{"text": "## Hankui wrote it on Nov 1, 2016\n## confusion matrix \n## source('confusion.r');\n\n# source(\"getKappa.r\");\n\nconfusion <- function(predicted, actual, filename=\"temp.csv\") \n{\n\tc_m <- as.matrix(table(Actual=actual, Predicted=predicted)) # create the confusion matrix\n\t# c_m <- as.data.frame(table(Predicted = predicted, Actual = actual )) # create the confusion matrix\n\tuniq_p <- sort(unique(predicted));\n\tuniq_t <- sort(unique(actual ));\n\tnr <- nrow(c_m) # number of classes\n\n\t##***************************************************************\n\t## important for those classes never been classified (Mar 13 2018 Hank) \n\tblock_list <- list();\n\tzeros <- rowsums <- apply(c_m, 1, sum)*0; \n\tempty_index <- vector();\n\tfor (i in 1:length(uniq_t)) {\n\t\tnc <- ncol(c_m);\n\t\tis_find <- 0;\n\t\tuniq_t_i <- uniq_t[i];\n\t\tfor (j in 1:length(uniq_p)) if (uniq_p[j]==uniq_t_i) {is_find <- j; break; }\n\t\t\n\t\tif(is_find==0) { \n\t\t\tblock_list[[i]] <- zeros;\n\t\t\tcat(\"\\tclass\", uniq_t_i, \"is never predicted\\n\");\n\t\t\tempty_index <- c(empty_index, i);\n\t\t} else {\n\t\t\tblock_list[[i]] <- c_m[,is_find];\n\t\t}\n\t}\n\tc_m_new <- block_list[[1]];\n\tfor (i in 2:length(uniq_t)) c_m_new <- cbind(c_m_new, block_list[[i]]);\n\t\n\tc_m <- c_m_new;\n\tcolnames(c_m) <- rownames(c_m);\n\n\t\n\t##***************************************************************\n\t## accuracy \n\tn <- sum(c_m) # number of instances\n\tdiags <- diag(c_m) # number of correctly classified instances per class \n\trowsums <- apply(c_m, 1, sum) # number of instances per class\n\tcolsums <- apply(c_m, 2, sum) # number of predictions per class\n\t\n\taccuracy = sum(diags) / n; \n\t# print(c_m)\n\t# print(accuracy)\n\t# precision = diags / colsums \n\t# recall = diags / rowsums \n\tuses <- diags / colsums * 100;\n\tproduces <- diags / rowsums * 100;\n\tc_m <- cbind(c_m, rowsums, produces);\n\t\n\tcolsumss <- c(colsums, n, 0 );\n\tuses.s <- c(uses , 0, accuracy);\n\tc_m <- rbind(c_m, colsumss, uses.s);\n\t\n\t# colnames(c_m) <- c(colnames(c_m)[1:nr],\"Total truth\",\"Producer's accuracy\");\n\t# rownames(c_m) <- c(rownames(c_m)[1:nr],\"Total predicted\",\"User's accuracy\");\n\trownames(c_m) <- c(as.numeric(colnames(c_m)[1:nr])-1,\"Total truth\",\"Producer's accuracy\");\n\tcolnames(c_m) <- c(as.numeric(rownames(c_m)[1:nr]) ,\"Total predicted\",\"User's accuracy\");\n\t\n\t# print(precision)\n\toptions(scipen=999);\n\t# print(format(c_m, digits=1));\n\twrite.csv(c_m, file=filename);\n\t\n\t## *******************************************************************************************\n\t## Kappa\n\tppp <- rowsums / n; # distribution of instances over the actual classes\n\tqqq <- colsums / n; # distribution of instances over the predicted classes\n\texpAccuracy = sum(ppp*qqq);\n\tKappa = (accuracy - expAccuracy) / (1 - expAccuracy);\n\tcat(\"\\tKappa = \");\n\tcat(format(Kappa, digits=4));\n\tcat(\"\\tAccuracy = \");\n\tprint(format(accuracy*100, digits=4)); \n\tlist(\"Kappa\"=Kappa, \"Accuracy\"=accuracy, \"Confusion\"=c_m);\n}\n## *******************************************************************************************\n## test data\n# set.seed(0)\n# actual = c('a','b','c')[runif(100, 1,4)] # actual labels\n# predicted = actual # predicted labels\n# predicted[runif(30,1,100)] = actual[runif(30,1,100)] # introduce incorrect predictions\n# predicted[runif(30,1,10)] = c('a','b','c')[runif(10, 1,4)] # introduce incorrect predictions\n# confusion(predicted, actual);\n\n## *******************************************************************************************\n## test data global random forest\n# prop <- 67;\n# prop <- 100;\n# load(paste(\"./file.plot/conus.urbanFALSE.prop\",prop,\".tree500\", sep=\"\"));\n# actual <- truth;\n# confusion(predicted, actual, \"global.confusion.csv\");\n", "meta": {"hexsha": "5cb13b34be70887f7fd1eb75e2e632c5f0ddbcfd", "size": 3635, "ext": "r", "lang": "R", "max_stars_repo_path": "confusion.r", "max_stars_repo_name": "hankui/cnn_time_series", "max_stars_repo_head_hexsha": "b35a3430f6a2326cb44f126e983112123f6babb2", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "confusion.r", "max_issues_repo_name": "hankui/cnn_time_series", "max_issues_repo_head_hexsha": "b35a3430f6a2326cb44f126e983112123f6babb2", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "confusion.r", "max_forks_repo_name": "hankui/cnn_time_series", "max_forks_repo_head_hexsha": "b35a3430f6a2326cb44f126e983112123f6babb2", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.7171717172, "max_line_length": 101, "alphanum_fraction": 0.5763411279, "num_tokens": 1046, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7826624688140726, "lm_q2_score": 0.7025300511670689, "lm_q1q2_score": 0.549843904262495}} {"text": "# # # # # # # # # # # # # # # # # # # # # # # # # #\n\n# tttt iiii lllllll \n# ttt:::t i::::i l:::::l \n# t:::::t iiii l:::::l \n# t:::::t l:::::l \n# uuuuuu uuuuuuttttttt:::::ttttttt iiiiiii l::::l ssssssssss \n# u::::u u::::ut:::::::::::::::::t i:::::i l::::l ss::::::::::s \n# u::::u u::::ut:::::::::::::::::t i::::i l::::l ss:::::::::::::s \n# u::::u u::::utttttt:::::::tttttt i::::i l::::l s::::::ssss:::::s\n# u::::u u::::u t:::::t i::::i l::::l s:::::s ssssss \n# u::::u u::::u t:::::t i::::i l::::l s::::::s \n# u::::u u::::u t:::::t i::::i l::::l s::::::s \n# u:::::uuuu:::::u t:::::t tttttt i::::i l::::l ssssss s:::::s \n# u:::::::::::::::uu t::::::tttt:::::ti::::::il::::::ls:::::ssss::::::s\n# u:::::::::::::::u tt::::::::::::::ti::::::il::::::ls::::::::::::::s \n# uu::::::::uu:::u tt:::::::::::tti::::::il::::::l s:::::::::::ss \n# uuuuuuuu uuuu ttttttttttt iiiiiiiillllllll sssssssssss \n\n# # # # # # # # # # # # # # # # # # # # # # # # # #\n\n# # # backprop\n# backpropagate error and update weights\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nbackprop <- function(out_wts, in_wts, out_activation, current_target, \n hid_activation, hid_activation_raw, ins_w_bias, learning_rate){\n\n # # # calc error on output units\n out_delta <- 2 * (out_activation - current_target)\n \n # # # calc error on hidden units\n hid_delta <- out_delta %*% t(out_wts)\n hid_delta <- hid_delta[,2:ncol(hid_delta)] * sigmoid_grad(hid_activation_raw)\n \n # # # calc weight changes\n out_delta <- learning_rate * (t(hid_activation) %*% out_delta)\n hid_delta <- learning_rate * (t(ins_w_bias) %*% hid_delta)\n\n # # # adjust wts\n out_wts <- out_wts - out_delta\n in_wts <- in_wts - hid_delta\n\n return(list(out_wts = out_wts, \n in_wts = in_wts))\n\n}\n\n# # # forward_pass\n# conduct forward pass\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nforward_pass <- function(in_wts, out_wts, inputs, out_rule) {\n # # # init needed vars\n num_feats <- ncol(out_wts)\n num_cats <- dim(out_wts)[3]\n num_stims <- nrow(inputs)\n if (is.null(num_stims)) {num_stims <- 1}\n\n \n # # # add bias to ins\n bias_units <- matrix(rep(1, num_stims), ncol = 1, nrow = num_stims)\n ins_w_bias <- cbind(bias_units,\n matrix(inputs, nrow = num_stims, ncol = num_feats, byrow = TRUE))\n\n # # # ins to hids propagation\n hid_activation_raw <- ins_w_bias %*% in_wts\n hid_activation <- sigmoid(hid_activation_raw)\n\n # # # add bias unit to hid activation\n hid_activation <- cbind(bias_units, hid_activation) \n\n # # # hids to outs propagation\n out_activation <- array(rep(0, (num_stims * num_feats * num_cats)), \n dim = c(num_stims, num_feats, num_cats))\n \n # # NEED VECTORIZED HERE?\n # # # get output activation\n for (category in 1:num_cats) {\n \tout_activation[,,category] <- hid_activation %*% out_wts[,,category]\n }\n \n # # # apply output activatio rule\n if(out_rule == 'sigmoid') {\n \tout_activation <- sigmoid(out_activation)\n }\n\n return(list(out_activation = out_activation, \n hid_activation = hid_activation,\n hid_activation_raw = hid_activation_raw, \n ins_w_bias = ins_w_bias))\n\n}\n\n# # # get_wts\n# generate net weights\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nget_wts <- function(num_feats, num_hids, num_cats, wts_range, wts_center) {\n # # # set bias\n bias <- 1\n \n # # # generate wts between ins and hids\n in_wts <- \n (matrix(runif((num_feats + bias) * num_hids), ncol = num_hids) - 0.5) * 2 \n in_wts <- wts_center + (wts_range * in_wts)\n\n # # # generate wts between hids and outs\n out_wts <- \n (array(runif((num_hids + bias) * num_feats * num_cats), \n dim = c((num_hids + bias), num_feats, num_cats)) - 0.5) * 2\n out_wts <- wts_center + (wts_range * out_wts) \n \n return(list(in_wts = in_wts, \n out_wts = out_wts))\n\n}\n\n# # # global_scale\n# scale inputs to 0/1\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nglobal_scale <- function(x) { x / 2 + 0.5 }\n\n# # # train_plot\n# function to produce line plot of training\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\ntrain_plot <- function(training) {\n # # # get dimensions\n n_cats <- dim(training)[2]\n xrange <- c(1, dim(training)[1])\n yrange <- range(training)\n\n # # # open plot device\n pdf('training_plot.pdf')\n\n # # # create frame\n plot(xrange, c(.01, yrange[2]), xlab = 'Block Number', ylab = 'Accuracy')\n\n # # # aesthetics \n colors <- rainbow(n_cats)\n line_type <- c(1:n_cats)\n plot_char <- seq(18, 18 + n_cats, 1)\n\n # # # plot lines\n for (i in 1:n_cats) {\n target_cat <- training[,i]\n lines(seq(1, xrange[2], 1), target_cat, type = 'b', lwd = 1.5, \n lty = line_type[i], col = colors[i], pch = plot_char[i])\n }\n\n # # # title and legend\n title('DIVA Training Accuracy across Blocks')\n legend('bottomright', y = NULL, 1:n_cats, cex = 0.8, col = colors, \n pch = plot_char, lty = line_type, title = 'SHJ Categories')\n\n # # # produce plot\n dev.off()\n\n}\n\n# response_rule\n# convert output activations to classification\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nresponse_rule <- function(out_activation, target_activation, beta_val){\n num_feats <- ncol(out_activation)\n num_cats <- dim(out_activation)[3]\n num_stims <- nrow(target_activation)\n if (is.null(num_stims)) {num_stims <- 1}\n\n # # # calc error \n ssqerror <- array(as.vector(\n apply(out_activation, 3, function(x) {x - target_activation})),\n c(num_stims, num_feats, num_cats))\n ssqerror <- ssqerror ^ 2\n ssqerror[ssqerror < 1e-7] <- 1e-7\n\n # # # if focusing is on:\n if (beta_val > 0) {\n # # # get list of channel comparisons\n pairwise_comps <- combn(1:num_cats, 2)\n \n # # # get differences for each feature between categories\n diff_matrix <- \n abs(apply(pairwise_comps, 2, function(x) {\n out_activation[,,x[1]] - out_activation[,,x[2]]}))\n\n # # # reconstruct activation array and get feature diversity means\n diff_array <- array(diff_matrix, dim = c(num_stims, num_feats, num_cats))\n feature_diffs <- apply(diff_array, 2, mean)\n\n # # # calculate diversities\n diversities <- exp(beta_val * feature_diffs)\n diversities[diversities > 1e+7] <- 1e+7\n\n # # # divide diversities by sum of diversities\n fweights = diversities / sum(diversities)\n\n # # # apply focus weights; then get sum for each category\n ssqerror <- t(apply(ssqerror, 3, function(x) sum(x * fweights))) \n\n # # # otherwise, set focus weights to NULL\n } else {\n fweights <- NULL\n }\n\n # # # calculate inverse sse\n ssqerror <- 1 / ssqerror\n\nreturn(list(ps = (ssqerror / sum(ssqerror)), \n fweights = fweights, \n ssqerror = ssqerror))\n\n}\n\n# run_diva\n# trains vanilla diva\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nrun_diva <- function(model) {\n \n # # # get new seed\n model.seed <- runif(1) * 100000 * runif(1)\n set.seed(model.seed)\n \n # # # set mean value of weights\n model$wts_center <- 0 \n # # # convert targets to 0/1\n model$targets <- global_scale(model$inputs) \n # # # init size parameter variables\n model$num_feats <- ncol(model$inputs)\n model$num_stims <- nrow(model$inputs)\n model$num_cats <- length(unique(model$labels))\n model$num_updates <- model$num_blocks * model$num_stims\n # # # init training accuracy matrix\n training <- \n matrix(rep(NA, model$num_updates * model$num_inits), \n nrow = model$num_updates, ncol = model$num_inits)\n \n # # # initialize and run DIVA models\n for (model_num in 1:model$num_inits) {\n\n # # # generate weights\n wts <- get_wts(model$num_feats, model$num_hids, model$num_cats, \n model$wts_range, model$wts_center)\n \n # # # generate random presentation order\n prez_order <- as.vector(apply(replicate(model$num_blocks, \n seq(1, model$num_stims)), 2, sample, model$num_stims))\n\n # # # iterate over each trial in the presentation order \n for (trial_num in 1:model$num_updates) {\n current_input <- model$inputs[prez_order[[trial_num]], ]\n current_target <- model$targets[prez_order[[trial_num]], ]\n current_class <- model$labels[prez_order[[trial_num]]] \n\n # # # complete forward pass\n fp <- forward_pass(wts$in_wts, wts$out_wts, current_input, model$out_rule)\n\n # # # calculate classification probability\n response <- response_rule(fp$out_activation, current_target, model$beta_val)\n\n # # # store classification accuracy\n training[trial_num, model_num] = response$ps[current_class]\n\n # # # back propagate error to adjust weights\n class_wts <- wts$out_wts[,,current_class]\n class_activation <- fp$out_activation[,,current_class]\n\n adjusted_wts <- backprop(class_wts, wts$in_wts, class_activation, current_target, \n fp$hid_activation, fp$hid_activation_raw, fp$ins_w_bias, model$learning_rate)\n\n # # # set new weights\n wts$out_wts[,,current_class] <- adjusted_wts$out_wts\n wts$in_wts <- adjusted_wts$in_wts\n \n }\n\n }\n\ntraining_means <- \n rowMeans(matrix(rowMeans(training), nrow = model$num_blocks, ncol = model$num_stims, byrow = TRUE))\n\nreturn(list(training = training_means,\n model = model))\n\n}\n\n# shj_cats\n# loads shj category structures\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\ndemo_cats <- function(type){\n \n in_pattern <- \n matrix(c(-1, -1, -1,\n\t -1, -1, 1,\n\t -1, 1, -1,\n \t -1, 1, 1,\n\t 1, -1, -1,\n\t 1, -1, 1,\n\t 1, 1, -1,\n\t 1, 1, 1), \n nrow = 8, ncol = 3, byrow = TRUE)\t\t\n\n cat_assignment <- \n matrix(c(1, 1, 1, 1, 2, 2, 2, 2, # type I\n 1, 1, 2, 2, 2, 2, 1, 1, # type II\n 1, 1, 2, 1, 1, 2, 2, 2, # type III\n 1, 1, 1, 2, 1, 2, 2, 2, # type IV\n 2, 1, 1, 1, 1, 2, 2, 2, # type V\n 1, 2, 2, 1, 2, 1, 1, 2, # type VI\n 1, 1, 2, 2, 3, 3, 4, 4), # type II multiclass \n ncol = 8, byrow = TRUE)\n\nreturn(list(inputs = in_pattern, \n\t\t\t labels = cat_assignment[type,]))\n\n}\n\n# sigmoid\n# returns sigmoid evaluated elementwize in X\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nsigmoid <- function(x) {\n g = 1 / (1 + exp(-x))\n\nreturn(g)\n\n}\n\n# sigmoid gradient\n# returns the gradient of the sigmoid function evaluated at x\n# # # # # # # # # # # # # # # # # # # # # # # # # # #\nsigmoid_grad <- function(x) {\n \nreturn(g = ((sigmoid(x)) * (1 - sigmoid(x))))\n\n}", "meta": {"hexsha": "f95820f87bcc67bdef27158a673edcca219148c8", "size": 11107, "ext": "r", "lang": "R", "max_stars_repo_path": "utils.r", "max_stars_repo_name": "ghonk/divaR", "max_stars_repo_head_hexsha": "55a0d2f45e71fdb6e2524f128330e40b8dc8f16c", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "utils.r", "max_issues_repo_name": "ghonk/divaR", "max_issues_repo_head_hexsha": "55a0d2f45e71fdb6e2524f128330e40b8dc8f16c", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "utils.r", "max_forks_repo_name": "ghonk/divaR", "max_forks_repo_head_hexsha": "55a0d2f45e71fdb6e2524f128330e40b8dc8f16c", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 33.4548192771, "max_line_length": 101, "alphanum_fraction": 0.5337174755, "num_tokens": 3786, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8723473614033683, "lm_q2_score": 0.6297746143530797, "lm_q1q2_score": 0.549382223109733}} {"text": "## Load some packages that will be useful\r\ninstall.packages(\"mgcv\") # load the MGCV package with Simon Wood's gam() functions\r\ninstall.packages(\"readr\") # Hadley Wickham's package for reading in data easily\r\ninstall.packages(\"ggplot2\") # for plotting with ggplot\r\ninstall.packages(\"ggfortify\")\r\nlibrary(mgcv) # load the MGCV package with Simon Wood's gam() functions\r\nlibrary(readr) # Hadley Wickham's package for reading in data easily\r\nlibrary(ggplot2) # for plotting with ggplot\r\nlibrary(ggfortify) # for autoplot\r\n\r\n##################################################################################################\r\n## We'll first examine the bias-variance tradeoff in a simplified form with a small exercise.\r\n## If you remember, this tradeoff is one of the concepts behind finding the optimal amount of smoothing.\r\n## We'll fit linear regressions to parts of the data, increasing the number of parts from 1 to 10.\r\n##################################################################################################\r\n\r\n# First, read in the bioluminescence data.\r\n# Sources is the number of bioluminescent plankton seen in each water sample\r\n# SampleDepth is the depth of the water sample, in m\r\n# Station is the ID of the location at which multiple water samples were taken\r\nISIT <- read_tsv(url('https://github.com/aeda2021/2021_master/raw/main/raw-data/ISIT.txt')) # tsv for tab-separated values\r\n\r\n# Subset it to one station. This is what we'll work with for now.\r\nISITsub <- subset(ISIT, Station == 8)\r\n\r\n# Plot the data\r\nggplot(ISITsub, aes(x=SampleDepth, y=Sources)) +\r\n\tgeom_point()\r\n\r\n# Now you can fit a linear regression to the ISITsub dataset, with Sources as the response and SampleDepth as the explanatory variable\r\n# Please call the output \"mod1\", as I've started to fill in for you\r\nmod1 <- lm (ISITsub$Sources ~ ISITsub$SampleDepth, data = ISITsub)\r\n\r\n# Q1: What was your code for fitting a linear regression of Sources vs. SampleDepth?\r\n#see above\r\n\r\n# We'll predict the mean response and confidence intervals from this model and save the output\r\nISITsub$mod1 <- predict(mod1)\r\nISITsub$mod1lwr <- predict(mod1, interval='confidence', level=0.95)[,'lwr']\r\nISITsub$mod1upr <- predict(mod1, interval='confidence', level=0.95)[,'upr']\r\n\r\n# Now plot the linear model fit to the full dataset on top of the data\r\nggplot(ISITsub, aes(x=SampleDepth, y=Sources)) +\r\n\tgeom_point() +\r\n\tgeom_line(aes(y=mod1)) + \r\n\tgeom_ribbon(aes(ymin=mod1lwr, ymax=mod1upr), alpha=0.3)\r\n\r\n# Q2: Use your skills from earlier in this course and evaluate this linear model (autoplot may be useful here). How does this fit look to you? Does a linear regression look reasonable? Any well-justified answer is appropriate. \r\nautoplot(mod1)\r\n#we want residuals around 0, so the residuals are not very normal; therefore a linear regression is not reasonable. \r\n\r\n# For comparison, let's fit linear models to two halves of the data. This involves a fair bit of coding in R, so please ask for help understanding the code where things are confusing. \r\n# First, we have to find the halves of the data\r\nquants2 <- quantile(ISITsub$SampleDepth, probs=seq(0, 1, by=0.5)) # define the ends and the half-way point in the data with the quantile function\r\nISITsub$split2 <- cut(ISITsub$SampleDepth, breaks=quants2, include.lowest=TRUE, labels=1:2) \r\n\r\n# Q3: In your own words, what does the cut() function do? How are we using it here?\r\n#The cut function allows us to split up data into two lables\r\n#here we are using it to split the data along the x axis into two subsets (labeled 1 and 2) based on where they fall (first half or second)\r\n\r\n# Next, let's fit a linear model to each subset (half) of the data. We do this in a loop to make it easier.\r\nmod2s <- vector('list', length=2) # A list to hold each of our models in a single, convenient object\r\nISITsub$mod2 <- ISITsub$mod2lwr <- ISITsub$mod2upr <- NA # Initialize the variables we'll use and fill with NA\r\nfor(i in 1:length(mod2s)){ # Loop through each model\r\n\tmod2s[[i]] <- lm(Sources ~ SampleDepth, data=ISITsub[ISITsub$split2==i,]) # Fit a linear regression to the appropriate subset of the data. Now see why we made the split2 vector?\r\n\tISITsub$mod2[ISITsub$split2==i] <- predict(mod2s[[i]]) # Now predict from that model. Notice we're only predicting for part of the data.\r\n\tISITsub$mod2lwr[ISITsub$split2==i] <- predict(mod2s[[i]], interval='confidence', level=0.95)[,'lwr'] # Same for the confidence intervals\r\n\tISITsub$mod2upr[ISITsub$split2==i] <- predict(mod2s[[i]], interval='confidence', level=0.95)[,'upr']\r\n}\r\n\r\n# Plot the data, the 1-model fit, and the 2-model fit\r\n# There will be an odd 3rd line segment connecting the end of the first model fit to the beginning of the second model fit. Just pretend it's not there.\r\nggplot(ISITsub, aes(x=SampleDepth, y=Sources)) +\r\n\tgeom_point() +\r\n\tgeom_line(aes(y=mod1)) +\r\n\tgeom_ribbon(aes(ymin=mod1lwr, ymax=mod1upr), alpha=0.3) +\r\n\tgeom_line(aes(y=mod2, color='red')) +\r\n\tgeom_ribbon(aes(ymin=mod2lwr, ymax=mod2upr), alpha=0.3, fill='red')\r\n\r\n\r\n# Q4: Has the fit between the model and the data improved? What's the downside of continuing the split the data into finer and finer chunks and fitting more models?\r\n#Yes, the fit has improved! However, the downside of continuing to split the data/fit more models is the trap of overfitting\r\n#you're not modelling the general trend, you'll eventually just show the exact data set\r\n\r\n# Now, it's your turn. Please fit 10 linear regression models. I'd recommend using the 2-model code above as a guide and tweaking it.\r\n\r\nquants10 <- quantile(ISITsub$SampleDepth, probs=seq(0, 1, by=0.1)) # define the ends and the 10 points\r\nISITsub$split10 <- cut(ISITsub$SampleDepth, breaks=quants10, include.lowest=TRUE, labels=1:10) \r\nmod10 <- vector('list', length=10) # A list to hold each of our models in a single, convenient object\r\nISITsub$mod10 <- ISITsub$mod10lwr <- ISITsub$mod10upr <- NA # Initialize the variables we'll use and fill with NA\r\nfor(i in 1:length(mod10)){ # Loop through each model\r\n mod10[[i]] <- lm(Sources ~ SampleDepth, data=ISITsub[ISITsub$split10==i,]) # Fit a linear regression to the appropriate subset of the data. Now see why we made the split2 vector?\r\n ISITsub$mod10[ISITsub$split10==i] <- predict(mod10[[i]]) # Now predict from that model. Notice we're only predicting for part of the data.\r\n ISITsub$mod10lwr[ISITsub$split10==i] <- predict(mod10[[i]], interval='confidence', level=0.95)[,'lwr'] # Same for the confidence intervals\r\n ISITsub$mod10upr[ISITsub$split10==i] <- predict(mod10[[i]], interval='confidence', level=0.95)[,'upr']\r\n}\r\n\r\n\r\n# Now plot the 10 models on top of the data and on top of the 1- and 2-model fits. The plot with the data, 1-model, and 2-model fit is probably a useful guide here.\r\n\r\nggplot(ISITsub, aes(x=SampleDepth, y=Sources)) +\r\n geom_point() +\r\n geom_line(aes(y=mod10)) +\r\n geom_ribbon(aes(ymin=mod10lwr, ymax=mod10upr), alpha=0.3) +\r\n geom_line(aes(y=mod10, color='red')) +\r\n geom_ribbon(aes(ymin=mod10lwr, ymax=mod10upr), alpha=0.3, fill='red')\r\n\r\n\r\n\r\n# Q5: Compare your 10-model fit to the 1-model fit. Which model's predictions are furthest from the observed data? Which model has the widest confidence bounds? How does this (or does this not) illustrate the bias-variance tradeoff?\r\n#Model 1 was the furthest from observed data, but Model 10 had the widest confidence bounds\r\n#bias-variance tradeoff = increasing bias to reduce variance\r\n#mod1 is biased because we assume a linear fit; mod10 has high variance, but it fits so well\r\n\r\n\r\n\r\n####################################################\r\n## Now you'll fit some GAMs and evaluate your models\r\n####################################################\r\n#install.packages(\"maps\") # if needed\r\nlibrary(maps) # has map data in it\r\n\r\n# load the data\r\nspdata <- readRDS(url('https://github.com/aeda2021/2021_master/raw/main/raw-data/gadusmorhua.rds')) # load the spdata data.frame with abundance and environmental data from cod surveys\r\nspdata$presfit01 <- as.numeric(spdata$presfit) # make a vector that's nice for plotting\r\nspdata <- spdata[order(spdata$presfit01),] # order the data from absent to present for ease of plotting\r\n\r\n# examine the data\r\nhead(spdata) # look at the dataset\r\nsummary(spdata)\r\n\r\n# make a map of the data\r\nworld <- map_data('world')\r\n\r\nggplot() + \r\n\tgeom_polygon(data=world, aes(x=long, y=lat, group=group), color='black', fill=NA) + # the map\r\n\txlim(-100, -45) + \r\n\tylim(23, 62) +\r\n\tgeom_point(data=spdata, aes(x=lon, y=lat, color=presfit, alpha=0.1), size=0.01) # the data points. note that because the dataframe is sorted from absent to present, the presences are plotted on top of the absences\r\n\r\n\r\n# Q6: Based on the map, do you think cod prefer warmer or cooler waters? Why do you think that?\r\n#Cod like cooler water. I think that because all of the blue points are in the nothern area = colder water\r\n\r\n\r\n# Q7: Plot cod presence/absence (the presfit vector) vs. the bottom temperature (SBT.actual). At what range of temperatures have cod been observed?\r\n\r\nggplot(spdata, aes(x=SBT.actual, y=presfit01)) +\r\n geom_point()\r\n#cod can be found in the range ~0 - 20 Celsius\r\n\r\n# Q8: Fit a GAM for presfit against SBT.actual. Check the model. Does it look like the assumptions have been met? Why or why not?\r\n\r\ngam1 <- gam(presfit01 ~ s(SBT.actual, k=8), data=spdata)\r\ngam.check(gam1)\r\n\r\n#does not meet assumptions, using plots we can see that it is violating normality\r\n\r\n\r\n# Q9: Fit a new GAM with a more appropriate error structure and save it as \"mod2\". The gam() function takes the same family= argument as does glm(). Which error structure and link function did you choose?\r\nmod2 <- gam(presfit01 ~ s(SBT.actual, k=8), data=spdata, family=binomial)\r\ngam.check(mod2)\r\n\r\n\r\n# Q10: Check your mod2 GAM. Does it look like the assumptions have been met? Make sure to read the text output as well as look at the graphs. If something doesn't look right, what would you do to fix it?\r\n\r\n#yes, the deviance residuals vs. theoretical quantiles plot follows the 1:1 ratio\r\n#I changed the k value because the output showed a low p-value, indicating that the edf and k value were too close\r\n\r\n\r\n# Q11: Interpret your mod2 GAM (summary function). How much deviance is explained? Is the SBT.actual term significant at alpha=0.05? How wiggly do you expect the smooth fit to be?\r\n#deviance = 25.4%\r\n#significance = yes, the SBT.actual is significant\r\n#line = pretty curvy, since edf is a 9\r\nsummary(mod2)\r\n\r\n\r\n# Now let's make predictions from our last model (mod2) and plot them.\r\nnd <- data.frame(SBT.actual = seq(-5,30, by=0.5)) # make a data.frame of new explanatory variables\r\nnd$mod2 <- predict(mod2, newdata=nd, type='response') # predict on the scale of the response\r\nnd$mod2se <- predict(mod2, newdata=nd, type='response', se.fit=TRUE)$se.fit # get the standard errors of the fit\r\n\r\nggplot(spdata, aes(x=SBT.actual, y=presfit01)) +\r\n\tgeom_point() +\r\n\tgeom_line(data=nd, aes(x=SBT.actual, y=mod2, color='red')) +\r\n\tgeom_ribbon(data=nd, aes(x=SBT.actual, ymin=mod2-mod2se, ymax=mod2+mod2se), alpha=0.3, fill='red', inherit.aes=FALSE)\r\n\r\n\r\n# Q12: Does the fit look realistic? Why or why not? Does it look overfit? Why or why not?\r\n#I feel like the fit is not realistc because it has many curves; it looks overfit\r\n#it captures too many of the ups and downs to be realistic \r\n\r\n\r\n\r\n\r\n# Q13: This dataset is actually a concatenation of seven different surveys which each have slightly different abilities to catch cod. Fit a new GAM that includes a categorical predictor for survey (the region vector). Based on AIC, which model would you choose? How confident would you be?\r\nmod3 <- gam(presfit01 ~ s(SBT.actual, k=6) + region, data=spdata, family=binomial)\r\nlibrary(MuMIn) # for model selection\r\nmodel.sel(mod3, mod2)\r\n\r\n#AIC shows that model 3, which incorporates region, is the better model. Pretty confident because it is quite a big difference between models. \r\n\r\n", "meta": {"hexsha": "701aea004bbd6bcd327f036e72a726a82c0e9d05", "size": 11960, "ext": "r", "lang": "R", "max_stars_repo_path": "9_GAMs.r", "max_stars_repo_name": "aeda2021/konnovitch_theresa", "max_stars_repo_head_hexsha": "b0a43e1c2169e0573d05dbb422934bee353ffab0", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "9_GAMs.r", "max_issues_repo_name": "aeda2021/konnovitch_theresa", "max_issues_repo_head_hexsha": "b0a43e1c2169e0573d05dbb422934bee353ffab0", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "9_GAMs.r", "max_forks_repo_name": "aeda2021/konnovitch_theresa", "max_forks_repo_head_hexsha": "b0a43e1c2169e0573d05dbb422934bee353ffab0", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 58.9162561576, "max_line_length": 290, "alphanum_fraction": 0.7143812709, "num_tokens": 3267, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7057850278370112, "lm_q2_score": 0.7772998663336158, "lm_q1q2_score": 0.5486066077979762}} {"text": "#' Principal Components Analysis\n#' \n#' Performs the principal components analysis.\n#' \n#' @details\n#' \\code{prcomp()} performs the principal components analysis on the data\n#' matrix by taking the SVD. Sometimes core R and kazaam will disagree\n#' slightly in what the rotated variables are because of how the SVD is\n#' caluclated.\n#' \n#' @section Communication:\n#' The communication is an \\code{allreduce()} call, quadratic on the number of\n#' columns. Most of the run time should be dominated by relatively expensive\n#' local operations.\n#' \n#' @param x \n#' A shaq.\n#' @param center \n#' Should columns are zero centered?\n#' @param scale. \n#' Should columns are rescaled to unit variance?\n#' @param retx \n#' Should the rotated variables be returned?\n#' @param tol \n#' The cutoff for components.\n#' @param ...\n#' Ignored.\n#' \n#' @return \n#' A list of elements \\code{sdev}, \\code{rotation}, \\code{center}, \\code{scale},\n#' and \\code{x}, as with R's own \\code{prcomp()}. The elements are,\n#' respectively, a regular vector, a regular matrix, a regular vector, a regular\n#' vector, and a shaq.\n#' \n#' @examples\n#' \\dontrun{\n#' library(kazaam)\n#' x = ranshaq(runif, 10, 3)\n#' pca = prcomp(x)\n#' \n#' comm.print(pca)\n#' \n#' finalize()\n#' }\n#' \n#' @name prcomp\n#' @rdname prcomp\n#' @export\nprcomp.shaq = function(x, retx=TRUE, center=TRUE, scale.=FALSE, tol=NULL, ...)\n{\n x <- scale(x, center=center, scale=scale.)\n x.center <- attr(DATA(x), \"scaled:center\")\n x.scale <- attr(DATA(x) , \"scaled:scale\")\n if (any(x.scale == 0))\n comm.stop(\"cannot rescale a constant/zero column to unit variance\")\n \n s <- svd(x, nu=0)\n s$d <- s$d/sqrt(max(1, nrow(x) - 1))\n \n if (!is.null(tol))\n {\n rank <- max(as.integer(sum(s$d > (s$d[1L] * tol))), 1L)\n if (rank < ncol(x))\n {\n s$v <- s$v[, 1L:rank, drop=FALSE]\n s$d <- s$d[1L:rank]\n }\n }\n \n if (is.null(x.center))\n center = FALSE\n else\n center = x.center\n \n if (is.null(x.scale))\n scale = FALSE\n else\n scale = x.scale\n \n r <- list(sdev=s$d, rotation=s$v, center=center, scale=scale)\n if (retx)\n r$x <- x %*% s$v\n \n class(r) <- \"prcomp\"\n \n return(r)\n}\n", "meta": {"hexsha": "e73cfc472ac050105e5894a66a97f66b64633014", "size": 2147, "ext": "r", "lang": "R", "max_stars_repo_path": "R/pca.r", "max_stars_repo_name": "snoweye/kazaam", "max_stars_repo_head_hexsha": "0b6f290516f1a6e505ced665b6f2730ba9bc45d4", "max_stars_repo_licenses": ["BSD-2-Clause"], "max_stars_count": 4, "max_stars_repo_stars_event_min_datetime": "2017-07-16T19:21:57.000Z", "max_stars_repo_stars_event_max_datetime": "2020-06-24T13:07:08.000Z", "max_issues_repo_path": "R/pca.r", "max_issues_repo_name": "snoweye/kazaam", "max_issues_repo_head_hexsha": "0b6f290516f1a6e505ced665b6f2730ba9bc45d4", "max_issues_repo_licenses": ["BSD-2-Clause"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2017-06-24T21:33:17.000Z", "max_issues_repo_issues_event_max_datetime": "2018-08-07T03:33:48.000Z", "max_forks_repo_path": "R/pca.r", "max_forks_repo_name": "snoweye/kazaam", "max_forks_repo_head_hexsha": "0b6f290516f1a6e505ced665b6f2730ba9bc45d4", "max_forks_repo_licenses": ["BSD-2-Clause"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2017-06-24T21:22:10.000Z", "max_forks_repo_forks_event_max_datetime": "2017-06-24T21:22:10.000Z", "avg_line_length": 24.3977272727, "max_line_length": 80, "alphanum_fraction": 0.6217978575, "num_tokens": 649, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7879311956428947, "lm_q2_score": 0.6959583124210896, "lm_q1q2_score": 0.5483672652235604}} {"text": "#'@title calcHMAppendageRes\n#'\n#'@description Calculate appendage resistance (\\code{Rapp}) (kN) from the\n#'Holtrop & Mennen method.\n#'\n#'@param shipSpeed Ship actual speed (vector of numericals, m/s) (see \n#' \\code{\\link{calcSpeedUnitConversion}})\n#'@param Cf Frictional resistance coefficient (vector of numericals, \n#' dimensionless) (see \\code{\\link{calcCf}})\n#'@param appendagesList List of appendages on ship (vector of strings) \\itemize{\n#'\\item\"rudder behind skeg\"\n#'\\item\"rudder behind stern\"\n#'\\item\"twin-screw balance rudders\"\n#'\\item\"shaft brackets\"\n#'\\item\"skeg\"\n#'\\item\"strut bossings\"\n#'\\item\"hull bossings\"\n#'\\item\"shafts\"\n#'\\item\"stabilizer fins\"\n#'\\item\"dome\"\n#'\\item\"bilge keels\"\n#'}\n#'@param wettedAppSAList List of wetted surface areas corresponding to list of appendages\n#' (vector of numericals, m^2)\n#'@param seawaterDensity Sea water density. Default = 1.025 (g/cm^3). Can \n#' supply either a vector of numericals corresponding to the ship speed, provide a single\n#' number, or rely on the default\n#'\n#'@return \\code{Rapp} (vector of numericals, kN)\n#'\n#'@references\n#'Holtrop, J. and Mennen, G. G. J. 1982. \"An approximate power prediction\n#'method.\" International Shipbuilding Progress 29.\n#'\n#'@seealso \\itemize{\n#'\\item \\code{\\link{calcSpeedUnitConversion}}\n#'\\item \\code{\\link{calcCf}} }\n#'\n#'@family Holtrop-Mennen Calculations\n#'@family Resistance Calculations\n#'\n#'@examples calcHMAppendageRes(seq(1,5,1),0.0015,\"rudder behind skeg\",50,seawaterDensity=1.025)\n#' calcHMAppendageRes(seq(1,5,1),0.0015,NA,0,seawaterDensity=1.025)\n#'\n#'@export\n\ncalcHMAppendageRes<- function(shipSpeed,Cf,\n appendagesList,wettedAppSAList,seawaterDensity=1.025){\n\n\nAppResFactor <- data.frame(appendage=c(\"rudder behind skeg\",\"rudder behind stern\",\n\"twin-screw balance rudders\",\"shaft brackets\",\"skeg\",\"strut bossings\",\"hull bossings\",\n\"shafts\",\"stabilizer fins\",\"dome\",\"bilge keels\"),\nresistance.factor=c(1.5,1.4,2.8,3,1.75,3,2,3,2.8,2.7,1.4))\n\nwettedAppSAList[is.na(wettedAppSAList)==TRUE]<-0\nappendagesList<- tolower(appendagesList)\n\nFormFactorEqu=0\n\nfor(i in 1:length(appendagesList)){\n if(length(AppResFactor$resistance.factor[AppResFactor$appendage==paste(appendagesList[i])]*\n wettedAppSAList[i])>0){\n j<- AppResFactor$resistance.factor[grepl(appendagesList[i],AppResFactor$appendage)==TRUE]*\n wettedAppSAList[i]}else{j<-0}\n FormFactorEqu=FormFactorEqu+j\n}\n\nFormFactorEqu<-FormFactorEqu/sum(wettedAppSAList)\nFormFactorEqu[is.na(FormFactorEqu)]<-0\nRapp<-0.5*seawaterDensity*(shipSpeed^2)*wettedAppSAList*FormFactorEqu*Cf\n\nreturn(Rapp)\n}\n\n", "meta": {"hexsha": "41fe9152fd0da59cc50d3cb658c6741e568c4890", "size": 2568, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcHMAppendageRes.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcHMAppendageRes.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcHMAppendageRes.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 33.7894736842, "max_line_length": 95, "alphanum_fraction": 0.742211838, "num_tokens": 846, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104789086703225, "lm_q2_score": 0.6757646140788307, "lm_q1q2_score": 0.5476929669366324}} {"text": "# Current method\n\n# TODO Consider implementing this from scratch so that you know the details\n# It should be straightforward, see here: https://bookdown.org/egarpor/NP-UC3M/kre-i-kre.html\n# Using kernsmooth currently because I expect their implementation is faster and more robust\n\n# The following functions contains Kernsmooth and ROT estimators. ROT is selected currently\n\n# Returns Lambda, or P(+|S=t)\nEstLambda = function(S, X, t, idx, h = NULL){\n \n tryCatch({\n Sorder <- order(S, decreasing=TRUE)\n m <- length(S)\n idx.range <- c(max(1, idx - 1000), min(length(S), idx + 1000))\n \n S.ordered <- S[Sorder][idx.range[1]:idx.range[2]]\n X.ordered <- X[Sorder][idx.range[1]:idx.range[2]]\n \n ## --1 Kernsmooth bandwidth estimator\n ## more accurate, but prone to errors\n # if(is.null(h)) h <- KernSmooth::dpill(x = S.ordered, y = X.ordered, gridsize = 100)\n \n ## --2 Rule of thumb estimator\n ## Should not throw an error\n if(is.null(h)) h <- (m^{-1/5}) * sd(S)\n \n lp0 <- KernSmooth::locpoly(x = S.ordered, y = X.ordered, bandwidth = h, degree = 0,\n range.x = range(S.ordered), gridsize = 1000)\n \n lp0.idx <- which.min(abs(lp0$x - t))\n lam <- lp0$y[lp0.idx]\n }, error = function(e) {\n stop(\"Lambda estimate failed. Try setting h manually.\")\n }\n )\n \n return(lam)\n}\n\n# --3 JHO \n#\n# # Returns Lambda, or P(+|S=t)\n# EstLambda = function(S, X, t, ...){\n# #h is the bandwidth, t is called c in the JZ paper\n# #m is the sample size, S and X are the vectors of scores and labels\n# m <- length(S)\n# h <- (m^{-1/5}) * sd(S)\n# ind.win <- (S < t + h) & (S > t - h)\n# exp.X.and.I <- sum(X*ind.win)/m\n# exp.I <- sum(ind.win)/m\n# return(exp.X.and.I/exp.I)\n# }\n\n# --4 np method\n\n# # Returns Lambda, or P(+|S=t)\n# EstLambda = function(S, X, t, idx, Sorder, h = NULL){\n# \n# idx.range <- c(max(1, idx - 100), min(length(S), idx + 100))\n# \n# S.ordered <- S[Sorder][idx.range[1]:idx.range[2]]\n# X.ordered <- X[Sorder][idx.range[1]:idx.range[2]]\n# \n# # TODO This is still not silent!\n# if(is.null(h)) {\n# bwa <- invisible(np::npregbw(formula = X.ordered ~ S.ordered, bwtype = \"adaptive_nn\",\n# regtype = \"lc\"))\n# } else {\n# bwa <- invisible(np::npregbw(formula = X.ordered ~ S.ordered, bws = h,\n# regtype = \"lc\"))\n# }\n# np.model <- invisible(np::npreg(bwa))\n# \n# return(predict(np.model, newdata = data.frame(S.ordered = t)))\n# }\n\n# --5 JZ original method\n# \n# Returns Lambda, or P(+|S=t)\n# EstLambda = function(S, X, t, ...){\n# #h is the bandwidth, t is called c in the JZ paper\n# #m is the sample size, S and X are the vectors of scores and labels\n# m <- length(S)\n# h <- (m^{-1/5})\n# ind.win <- (S < t + h) & (S > t - h)\n# exp.X.and.I <- sum(X*ind.win)/m\n# exp.I <- sum(ind.win)/m\n# return(exp.X.and.I/exp.I)\n# }\n\n\n# Bootstrap percentile CI\nBootCI = function(X, S, m, pi.0, boot.rep, metric, plus, r, myseed=111){\n storeout=matrix(NA, nrow=m, ncol=1+boot.rep)\n r.all <- (1:m)/m\n idx <- which(r.all %in% r)\n storeout <- storeout[idx, ]\n storeout[,1] <- r\n set.seed(myseed)\n for(i in 1:boot.rep){\n boot.samp <- sample(1:m, m, replace = T)\n X.star = X[boot.samp]\n S.star = S[boot.samp]\n Sorder <- order(S.star,decreasing=TRUE)\n hits <- cumsum(X.star[Sorder])[idx] \n pi.0 <- mean(X.star)\n pi <- (hits)/(m*r)\n k <- (hits)/(sum(X.star))\n if (plus) {\n # using random plus correction\n plus.yes <- rbinom(1, 4, .5)\n pi <- (hits+plus.yes)/(m*r+4)\n k <- r/pi.0*pi\n }\n if(metric == \"rec\") {\n storeout[,1+i] = k\n } else if(metric == \"prec\") {\n storeout[,1+i] = pi\n } else if(metric == \"lift\") {\n storeout[,1+i] = k/r\n }\n # print(i/boot.rep)\n }\n A = apply(storeout[,-1],MARGIN=1,FUN=function(x){quantile(x, probs = c(.025, .975))})\n # Returns 95% percentile bootstrap intervals at quantiles\n A=t(A)\n}\n\n#' Construct a confidence band for a recall or precision curve\n#' \n#' \\code{PerfCurveBands} takes a pair of score and activity vectors as input.\n#' A performance curve and confidence band is created for the selected testing fractions.\n#' \n#' @param S a vector of scores.\n#' @param X a vector of activities.\n#' @param r a vector of testing fractions.\n#' @param metric the performance curve to use. Options are recall (\"rec\") and precision (\"prec\").\n#' @param type specifies whether a point-wise confidence interval \n#' (\"pointwise\") or a confidence band (\"band\") should be constructed.\n#' @param method the method to use. Point-wise confidence interval options\n#' are \"binomial\", \"JZ\", \"bootstrap\". Confidence band options are \"sup-t\", \"theta-proj\".\n#' @param plus should plus correction be used or not?\n#' @param conf.level the confidence level for the bands.\n#' @param boot.rep the number of replicates to use for the bootstrap method.\n#' @param mc.rep the number of Monte Carlo replicates to use for the sup-t method.\n#' @param myseed the random seed.\n#' \n#' @export\nPerfCurveBands <- function(S, X, r, metric = \"rec\", type = \"band\", method = \"sup-t\",\n plus = T, conf.level = .95, boot.rep = 100,\n mc.rep = 100000, myseed = 111, h = NULL){\n \n # TODO add support for EF\n \n # Some error handeling\n \n if (!(method %in% c(\"binomial\", \"JZ\", \"bootstrap\", \"sup-t\", \"theta-proj\", \"bonf\"))) {\n stop(\"'method' should be a string specifiying a performance curve method in chemmodlab.\n Point-wise confidence interval options are 'binomial', 'JZ', 'bootstrap'. Confidence band options\n are 'sup-t', 'theta-proj'.\")\n }\n \n if (!(metric %in% c(\"rec\", \"prec\"))) {\n stop(\"'metric' should be a string specifiying a performance curve metric in chemmodlab.\n Point-wise confidence interval options are 'binomial', 'JZ', 'bootstrap'. Options\n are recall ('rec') and precision ('prec').\")\n }\n \n if (!(type %in% c(\"band\", \"pointwise\"))) {\n stop(\"'type' should be a string specifiying a performance curve type in chemmodlab.\n Point-wise confidence interval options are 'binomial', 'JZ', 'bootstrap'. Options\n are point-wise confidence interval ('pointwise') or a confidence band ('band').\")\n }\n \n set.seed(myseed)\n alpha <- 1-conf.level\n m <- length(S) #total sample size\n nact <- sum(X) #total number of actives\n ntest <- m*r #total number tested\n r.all <- (1:m)/m\n idx <- which(r.all %in% r)\n \n # Convert fractions in r to thresholds, for both sets of scores\n # Also accumulate the number of actives, for each score and jointly\n Fcdf <- ecdf(S); yyobs <- sort(unique(S))\n Finv <- stepfun(x = Fcdf(yyobs), y = c(yyobs, max(yyobs)), right=TRUE, f=1)\n hits <- vector(length = length(r))\n t <- vector(length = length(r))\n for(i in 1:length(r)) {\n t[i] <- Finv(1-r[i])\n hits[i] <- sum(X*(S > t[i]))\n }\n \n Lam.vec <- vector(length = length(idx))\n for(j in seq_along(idx)){\n Lam.vec[j] <- EstLambda(S, X, t = t[j], idx = idx[j], h)\n }\n \n pi <- (hits)/(m*r)\n k <- hits/nact\n \n # Plus 2 correction\n if (plus) {\n hits <- hits + 2\n nact <- nact + 4\n ntest <- ntest + 2\n m <- m + 4\n }\n \n pi.0 <- nact/m\n r <- ntest/m\n k.c <- hits/nact\n pi.c <- pi.0/r*k.c\n k.ide <- (cumsum(rev(sort(X)))[idx]/sum(X))\n \n CI.int <- matrix(ncol = 2, nrow = length(k))\n if(type == \"pointwise\") {\n \n # Z Quantile for CIs\n quant <- qnorm(1-alpha/2)\n \n if(metric == \"rec\") {\n if(method == \"JZ\"){\n for(j in seq_along(k)) {\n Lam <- Lam.vec[j]\n var.k <- ((k.c[j]*(1-k.c[j]))/(m*pi.0))*(1-2*Lam) + \n (Lam^2*(1-r[j])*r[j])/(m*pi.0^2)\n # Check to see if var.k is negative due to machine precision problem\n var.k <- ifelse(var.k < 0, 0, var.k)\n sd.k <- sqrt(var.k)\n lcl <- k.c[j] - quant*sd.k\n lcl <- ifelse(lcl < 0, 0, lcl)\n ucl <- k.c[j] + quant*sd.k\n ucl <- ifelse(ucl > k.ide[j], k.ide[j], ucl)\n CI.int[j, ] <- c(lcl, ucl)\n }\n } else if(method == \"bootstrap\") {\n # bootstrap quantiles\n CI.int <- BootCI(X, S, m, pi.0, boot.rep, metric = \"rec\", plus, r, myseed=myseed)\n CI.int[, 1] <- ifelse(CI.int[, 1] < 0, 0, CI.int[, 1])\n CI.int[, 2] <- ifelse(CI.int[, 2] > k.ide, k.ide, CI.int[, 2])\n } else if(method == \"binomial\") {\n for(j in seq_along(k)) {\n var.k <- ((m*pi.0)^-1)*k.c[j]*(1-k.c[j])\n var.k <- ifelse(var.k < 0, 0, var.k)\n sd.k <- sqrt(var.k)\n lcl <- k.c[j] - quant*sd.k\n lcl <- ifelse(lcl < 0, 0, lcl)\n ucl <- k.c[j] + quant*sd.k\n # TODO consider removing due to coverage issues\n ucl <- ifelse(ucl > k.ide[j], k.ide[j], ucl)\n CI.int[j, ] <- c(lcl, ucl)\n }\n }\n } else if(metric == \"prec\") {\n if(method == \"JZ\") {\n for(j in seq_along(k)) {\n Lam <- Lam.vec[j]\n var.pi <- (pi.c[j]*(1-pi.c[j]))/(m*r[j]) +\n (1-r[j])*(pi.c[j]-Lam)^2/(m*r[j])\n # Check to see if var.pi is negative due to machine precision issues\n var.pi <- ifelse(var.pi < 0, 0, var.pi)\n sd.pi <- sqrt(var.pi)\n lcl <- pi.c[j] - quant*sd.pi\n lcl <- ifelse(lcl < 0, 0, lcl)\n ucl <- pi.c[j] + quant*sd.pi\n ucl <- ifelse(ucl > 1, 1, ucl)\n CI.int[j, ] <- c(lcl, ucl)\n }\n } else if(method == \"bootstrap\") {\n # bootstrap quantiles\n CI.int <- BootCI(X, S, m, pi.0, boot.rep, metric = \"prec\", plus, r, myseed=myseed)\n CI.int[, 1] <- ifelse(CI.int[, 1] < 0, 0, CI.int[, 1])\n CI.int[, 2] <- ifelse(CI.int[, 2] > 1, 1, CI.int[, 2])\n } else if(method == \"binomial\") {\n for(j in seq_along(pi)) {\n var.pi <- ((m*r[j])^-1)*pi.c[j]*(1-pi.c[j])\n # Check to see if var.pi is negative due to machine precision issues\n var.pi <- ifelse(var.pi < 0, 0, var.pi)\n sd.pi <- sqrt(var.pi)\n lcl <- pi.c[j] - quant*sd.pi\n lcl <- ifelse(lcl < 0, 0, lcl)\n ucl <- pi.c[j] + quant*sd.pi\n ucl <- ifelse(ucl > 1, 1, ucl)\n CI.int[j, ] <- c(lcl, ucl)\n }\n }\n }\n } else if(type == \"band\") {\n if(metric == \"rec\") {\n if(method == \"sup-t\") {\n cor.C <- matrix(NA, ncol = length(k), nrow = length(k))\n for(f in seq_along(k)) {\n for(e in 1:f) {\n Lam1 <- Lam.vec[e]\n var.k1 <- (((k.c[e]*(1-k.c[e]))/(m*pi.0))*(1-2*Lam1) +\n (Lam1^2*(1-r[e])*r[e])/(m*pi.0^2))\n var.k1 <- ifelse(var.k1 < 0, 0, var.k1)\n Lam2 <- Lam.vec[f]\n var.k2 <- (((k.c[f]*(1-k.c[f]))/(m*pi.0))*(1-2*Lam2) +\n (Lam2^2*(1-r[f])*r[f])/(m*pi.0^2))\n var.k2 <- ifelse(var.k2 < 0, 0, var.k2)\n cov.k <- ((m^-1*pi.0^-2)*(pi.0*(k.c[e]-k.c[e]*k.c[f])*(1-Lam1-Lam2) +\n (r[e]-r[e]*r[f])*Lam1*Lam2))\n if(e == f) {\n # If the covariance serves as a variance then it cant be non-negative\n # otherwise, it can be negative\n cov.k <- ifelse(cov.k < 0, 0, cov.k)\n }\n cor.k <- cov.k/(sqrt(var.k1)*sqrt(var.k2))\n cor.k <- ifelse(var.k1 == 0 | var.k2 == 0, ifelse(e == f, 1, 0), cor.k)\n cor.C[e, f] <- cor.k\n cor.C[f, e] <- cor.k\n }\n }\n mc.samples <- MASS::mvrnorm(n = mc.rep, rep(0, length = length(k)), cor.C, tol = 1)\n max.q <- vector(length = mc.rep)\n for(j in 1:mc.rep) {\n max.q[j] <- max(abs(mc.samples[j, ]))\n }\n # Should be 1-alpha, see Montiel Olea and Plagborg-Møller\n quant <- quantile(max.q, probs = 1-alpha)\n } else if(method == \"theta-proj\") {\n quant <- sqrt(qchisq(1-alpha, length(k)))\n } else if(method == \"bonf\") {\n quant <- qnorm(1-alpha/(2*length(k)))\n } else {\n stop(\"Invalid method selected\")\n }\n for(j in seq_along(k)) {\n Lam <- Lam.vec[j]\n var.k <- ((k.c[j]*(1-k.c[j]))/(m*pi.0))*(1-2*Lam) + \n (Lam^2*(1-r[j])*r[j])/(m*pi.0^2)\n # Check to see if var.k is negative due to machine precision problem\n var.k <- ifelse(var.k < 0, 0, var.k)\n sd.k <- sqrt(var.k)\n lcl <- k.c[j] - quant*sd.k\n lcl <- ifelse(lcl < 0, 0, lcl)\n ucl <- k.c[j] + quant*sd.k\n ucl <- ifelse(ucl > k.ide[j], k.ide[j], ucl)\n CI.int[j, ] <- c(lcl, ucl)\n }\n } else if(metric == \"prec\") {\n if(method == \"sup-t\") {\n cor.C <- matrix(NA, ncol = length(k), nrow = length(k))\n for(f in seq_along(pi)) {\n for(e in 1:f) {\n Lam1 <- Lam.vec[e]\n var.pi1 <- (pi.c[e]*(1-pi.c[e]))/(m*r[e]) + \n (1-r[e])*(pi.c[e]-Lam1)^2/(m*r[e])\n var.pi1 <- ifelse(var.pi1 < 0, 0, var.pi1)\n Lam2 <- Lam.vec[f]\n var.pi2 <- (pi.c[f]*(1-pi.c[f]))/(m*r[f]) + \n (1-r[f])*(pi.c[f]-Lam2)^2/(m*r[f])\n var.pi2 <- ifelse(var.pi2 < 0, 0, var.pi2)\n cov.pi <- (((m*r[e]*r[f])^{-1})*(r[e]*pi.c[e]*(1 - pi.c[e]) + \n (pi.c[e]-Lam1)*(pi.c[e]-Lam2)*(r[e] - r[e]*r[f])))\n if(e == f) {\n # If the covariance serves as a variance then it cant be non-negative\n # otherwise, it can be negative\n cov.pi <- ifelse(cov.pi < 0, 0, cov.pi)\n }\n cor.pi <- cov.pi/(sqrt(var.pi1)*sqrt(var.pi2))\n cor.pi <- ifelse(var.pi1 == 0 | var.pi2 == 0, ifelse(e == f, 1, 0), cor.pi)\n cor.C[e, f] <- cor.pi\n cor.C[f, e] <- cor.pi\n }\n }\n mc.samples <- MASS::mvrnorm(n = mc.rep, rep(0, length = length(k)), cor.C, tol = 1)\n max.q <- vector(length = mc.rep)\n for(j in 1:mc.rep) {\n max.q[j] <- max(abs(mc.samples[j, ]))\n }\n quant <- quantile(max.q, probs = 1-alpha)\n } else if(method == \"theta-proj\") {\n quant <- sqrt(qchisq(1-alpha, length(pi)))\n } else if(method == \"bonf\") {\n quant <- qnorm(1-alpha/(2*length(k)))\n } else {\n stop(\"Invalid method selected\")\n }\n for(j in seq_along(pi)) {\n Lam <- Lam.vec[j]\n var.pi <- (pi.c[j]*(1-pi.c[j]))/(m*r[j]) + (1-r[j])*(pi.c[j]-Lam)^2/(m*r[j])\n # Check to see if var.k is negative due to machine precision problem\n var.pi <- ifelse(var.pi < 0, 0, var.pi)\n sd.pi <- sqrt(var.pi)\n lcl <- pi.c[j] - quant*sd.pi\n lcl <- ifelse(lcl < 0, 0, lcl)\n ucl <- pi.c[j] + quant*sd.pi\n ucl <- ifelse(ucl > 1, 1, ucl)\n CI.int[j, ] <- c(lcl, ucl)\n }\n }\n }\n \n list(CI = CI.int, rec = k, prec = pi, h = h)\n \n }\n\n", "meta": {"hexsha": "ad63ac8913462ae52d3648d752c5f93a3a1cb798", "size": 14799, "ext": "r", "lang": "R", "max_stars_repo_path": "simulation_files/confidence_bands/confidence_bands.r", "max_stars_repo_name": "jrash/Enrichment-Inference-Supplemental-Materials", "max_stars_repo_head_hexsha": "0f0191cfcc5bf4a80cc39a593e912a59e5970cbb", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "simulation_files/confidence_bands/confidence_bands.r", "max_issues_repo_name": "jrash/Enrichment-Inference-Supplemental-Materials", "max_issues_repo_head_hexsha": "0f0191cfcc5bf4a80cc39a593e912a59e5970cbb", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "simulation_files/confidence_bands/confidence_bands.r", "max_forks_repo_name": "jrash/Enrichment-Inference-Supplemental-Materials", "max_forks_repo_head_hexsha": "0f0191cfcc5bf4a80cc39a593e912a59e5970cbb", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.8134328358, "max_line_length": 106, "alphanum_fraction": 0.5150347996, "num_tokens": 4775, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7826624789529375, "lm_q2_score": 0.6992544210587586, "lm_q1q2_score": 0.5472801986046492}} {"text": "#' Multiple variance Brownian motion estimation\n#'\n#' Computes rescaled branch lengths using multiple variance Brownian motion as described in Smaers et al. (2016) \n#' @param data a vector of tip values for species; should be in the same order as tiplabels in the tree\n#' @param tree an object of class \"phylo\".\n#' @param sigma sig2 value from a Bayesian MCMC run using standard Brownian motion (e.g. ?anc.Bayes)\n#' @return dataframe with rescaled branch lengths (rBL) for all branches in the tree\n#' @references Smaers, Mongle & Kandler (2016) A multiple variance Brownian motion framework for estimating variable rates and inferring ancestral states. Biological Journal of the Linnean Society. 118 (1): 78-94.\n#' @examples tree<-pbtree(n=50) # simulate tree ('pbtree' requires the 'phytools' package)\n#' x<-fastBM(tree,sig2=2) # simulate values ('pbtree' requires the 'phytools' package)\n#' BMsigma2<-ace(x,tree,method=\"REML\")$sigma2[1] # get BM sigma2 ('ace' requires the 'ape' package)\n#' mvBMresults<-mvBM(x,tree,BMsigma2) # calculate rescaled branch lengths using mvBM\n#' tree_mvBM<-tree\n#' tree_mvBM$edge.length<-mvBMresults$rBL # create new tree with rescaled branch lengths\n#' plot(tree_mvBM) # plot mvBM tree\n#' ace(x,tree_mvBM,method=\"REML\") # get ancestral estimates using mvBM tree \n#' # To obtain the mvBM branch specific rate estimate: calculate sig2 for the trait in question using the mvBM tree; multiply the mvBM sig2 with the mvBM rescaled branch length of the lineage of interest; divide this value by the phylogenetic branch length of the lineage of interest. \n#' # MCMC posterior distributions can be obtained by using anc.Bayes for the above calculations. ('anc.Bayes' requires the 'phytools' package)\n#' @examples for other examples see https://smaerslab.com/software/\n\n#' @export\nmvBM <- function (data, tree, sigma2) \n{\n data_original <- data\n N = length(data)\n phy.matrix = data.frame(tree$edge, tree$edge.length, data[tree$edge[, \n 2]])\n names(phy.matrix) = c(\"Anc\", \"Desc\", \"Length\", \"Value\")\n nodes_extant = 1:N\n nodes_extinct = (N + 1):(N + (N - 1))\n extant_values <- data.frame(data, nodes_extant)\n Pk_values <- c()\n \n dist.tree <- dist.nodes(tree)\n \n for (j in nodes_extinct) {\n nominator = c()\n denominator = c()\n\n for (i in nodes_extant) {\n nominator = rbind(nominator, (extant_values[i, 1]/dist.tree[i, \n j]^2))\n denominator = rbind(denominator, (1/dist.tree[i, \n j]^2))\n }\n \n Pk = sum(nominator)/sum(denominator)\n Pk_values = rbind(Pk_values, Pk)\n }\n \n \n Pk <- c()\n Pk <- cbind(Pk, Pk_values)\n Pk <- cbind(Pk, nodes_extinct)\n colnames(Pk) <- c(\"value\", \"nodes\")\n rownames(Pk) <- 1:length(nodes_extinct)\n Pk <- as.data.frame(Pk)\n nodes_extinct_reverse = sort(nodes_extinct, decreasing = TRUE)\n rBL <- c()\n rBL_edge <- c()\n for (i in nodes_extinct_reverse) {\n sister_branches <- which(phy.matrix$Anc == i)\n X1 <- phy.matrix$Value[sister_branches[1]]\n X2 <- phy.matrix$Value[sister_branches[2]]\n Y1 <- sqrt(phy.matrix$Length[sister_branches[1]])\n Y2 <- sqrt(phy.matrix$Length[sister_branches[2]])\n Pk_anc <- Pk$value[which(Pk$nodes == i)]\n Ax <- (Pk_anc + X1 + X2)/3\n T1 <- ((X1 - Ax)^2)/sigma2 + Y1^2\n T2 <- ((X2 - Ax)^2)/sigma2 + Y2^2\n phy.matrix$Value[which(phy.matrix$Desc == i)] <- Ax\n rBL <- c(rBL, T1)\n rBL <- c(rBL, T2)\n rBL_edge <- c(rBL_edge, sprintf(\"%05.f\", sister_branches[1]))\n rBL_edge <- c(rBL_edge, sprintf(\"%05.f\", sister_branches[2]))\n names(rBL) <- rBL_edge\n }\n Output <- data.frame(tree$edge, tree$edge.length, rBL[sort(names(rBL))])\n names(Output) <- c(\"node_anc\", \"node_desc\", \"BL\", \"rBL\")\n return(Output)\n}\n", "meta": {"hexsha": "9a0542a8d9ebc81822bfb59a0fc1922e08f52e91", "size": 3858, "ext": "r", "lang": "R", "max_stars_repo_path": "R/mvBM.r", "max_stars_repo_name": "JeroenSmaers/evomap", "max_stars_repo_head_hexsha": "dfa7dfdc560d1fd04414dffedab7b6be765d8175", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2016-01-07T04:20:28.000Z", "max_stars_repo_stars_event_max_datetime": "2018-07-15T19:42:12.000Z", "max_issues_repo_path": "R/mvBM.r", "max_issues_repo_name": "JeroenSmaers/evomap", "max_issues_repo_head_hexsha": "dfa7dfdc560d1fd04414dffedab7b6be765d8175", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/mvBM.r", "max_forks_repo_name": "JeroenSmaers/evomap", "max_forks_repo_head_hexsha": "dfa7dfdc560d1fd04414dffedab7b6be765d8175", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 4, "max_forks_repo_forks_event_min_datetime": "2015-10-14T18:26:29.000Z", "max_forks_repo_forks_event_max_datetime": "2019-01-10T04:57:55.000Z", "avg_line_length": 47.0487804878, "max_line_length": 285, "alphanum_fraction": 0.6516329705, "num_tokens": 1120, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8688267796346599, "lm_q2_score": 0.6297745935070806, "lm_q1q2_score": 0.5471650319724839}} {"text": "model_phylsowingdatecorrection <- function (sowingDay = 1,\n latitude = 0.0,\n sDsa_sh = 1.0,\n rp = 0.0,\n sDws = 1,\n sDsa_nh = 1.0,\n p = 120.0){\n #'- Name: PhylSowingDateCorrection -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: PhylSowingDateCorrection Model\n #' * Author: Loic Manceau\n #' * Reference: Modeling development phase in the \n #' Wheat Simulation Model SiriusQuality.\n #' See documentation at http://www1.clermont.inra.fr/siriusquality/?page_id=427\n #' * Institution: INRA Montpellier\n #' * Abstract: Correction of the Phyllochron Varietal parameter according to sowing date \n #'- inputs:\n #' * name: sowingDay\n #' ** description : Day of Year at sowing\n #' ** parametercategory : species\n #' ** datatype : INT\n #' ** min : 1\n #' ** max : 365\n #' ** default : 1\n #' ** unit : d\n #' ** uri : some url\n #' ** inputtype : parameter\n #' * name: latitude\n #' ** description : Latitude\n #' ** parametercategory : soil\n #' ** datatype : DOUBLE\n #' ** min : -90\n #' ** max : 90\n #' ** default : 0.0\n #' ** unit : °\n #' ** uri : some url\n #' ** inputtype : parameter\n #' * name: sDsa_sh\n #' ** description : Sowing date at which Phyllochrone is maximum in southern hemispher\n #' ** parametercategory : species\n #' ** inputtype : parameter\n #' ** datatype : DOUBLE\n #' ** min : 1\n #' ** max : 365\n #' ** default : 1.0\n #' ** unit : d\n #' ** uri : some url\n #' * name: rp\n #' ** description : Rate of change of Phyllochrone with sowing date\n #' ** parametercategory : species\n #' ** inputtype : parameter\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 365\n #' ** default : 0\n #' ** unit : d-1\n #' ** uri : some url\n #' * name: sDws\n #' ** description : Sowing date at which Phyllochrone is minimum\n #' ** parametercategory : species\n #' ** datatype : INT\n #' ** default : 1\n #' ** min : 1\n #' ** max : 365\n #' ** unit : d\n #' ** uri : some url\n #' ** inputtype : parameter\n #' * name: sDsa_nh\n #' ** description : Sowing date at which Phyllochrone is maximum in northern hemispher\n #' ** parametercategory : species\n #' ** datatype : DOUBLE\n #' ** default : 1.0\n #' ** min : 1\n #' ** max : 365\n #' ** unit : d\n #' ** uri : some url\n #' ** inputtype : parameter\n #' * name: p\n #' ** description : Phyllochron (Varietal parameter)\n #' ** parametercategory : species\n #' ** datatype : DOUBLE\n #' ** default : 120\n #' ** min : 0\n #' ** max : 1000\n #' ** unit : °C d leaf-1\n #' ** uri : some url\n #' ** inputtype : parameter\n #'- outputs:\n #' * name: fixPhyll\n #' ** description : Phyllochron Varietal parameter \n #' ** variablecategory : auxiliary\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 1000\n #' ** unit : °C d leaf-1\n if (latitude < 0.0)\n {\n if (sowingDay > as.integer(sDsa_sh))\n {\n fixPhyll <- p * (1 - (rp * min((sowingDay - sDsa_sh), sDws)))\n }\n else\n {\n fixPhyll <- p\n }\n }\n else\n {\n if (sowingDay < as.integer(sDsa_nh))\n {\n fixPhyll <- p * (1 - (rp * min(sowingDay, sDws)))\n }\n else\n {\n fixPhyll <- p\n }\n }\n return (list('fixPhyll' = fixPhyll))\n}", "meta": {"hexsha": "21b57fdad1b8cb4355d460f691f78484033307b8", "size": 5340, "ext": "r", "lang": "R", "max_stars_repo_path": "src/r/SQ_Wheat_Phenology/Phylsowingdatecorrection.r", "max_stars_repo_name": "Crop2ML-Catalog/SQ_Wheat_Phenology", "max_stars_repo_head_hexsha": "8e9ec229e5f0754d0f4b9d79ac96a084d75dde67", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-06-21T18:58:04.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-29T21:32:28.000Z", "max_issues_repo_path": "src/r/SQ_Wheat_Phenology/Phylsowingdatecorrection.r", "max_issues_repo_name": "Crop2ML-Catalog/SQ_Wheat_Phenology", "max_issues_repo_head_hexsha": "8e9ec229e5f0754d0f4b9d79ac96a084d75dde67", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 27, "max_issues_repo_issues_event_min_datetime": "2018-12-04T15:35:44.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-11T08:25:03.000Z", "max_forks_repo_path": "src/r/SQ_Wheat_Phenology/Phylsowingdatecorrection.r", "max_forks_repo_name": "Crop2ML-Catalog/SQ_Wheat_Phenology", "max_forks_repo_head_hexsha": "8e9ec229e5f0754d0f4b9d79ac96a084d75dde67", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 7, "max_forks_repo_forks_event_min_datetime": "2019-04-20T02:25:22.000Z", "max_forks_repo_forks_event_max_datetime": "2021-11-04T07:52:35.000Z", "avg_line_length": 44.8739495798, "max_line_length": 115, "alphanum_fraction": 0.3284644195, "num_tokens": 1136, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8175744761936437, "lm_q2_score": 0.6688802669716107, "lm_q1q2_score": 0.5468594339055791}} {"text": "#' @export\n#'\n#' @title checkIsAutocorrelated - Check whether MCMC chains are autocorrelated.\n#'\n#' @description Run autocorrelation on every parameter in an mcmc.list and make\n#' a decision whether they are significantly auto correlated. Ideally, chains\n#' are not autocorrelated when converged.\n#'\n#' @param obj An mcmc.list object from the \\code{rjags} package. These are output\n#' by the \\code{coda.samples} function.\n#'\n#' @param criterion scalar specifying the level of acceptable autocorrelation at\n#' the assessed \\code{lag}. If all average lag2 autocorrelations are less than this number, the\n#' whole mcmc object is deems 'random' or not autocorrelated.\n#'\n#' @param lag The lag at which to assess autocorrelation. This should generally be a\n#' low number like 1, 2, 5, or 10.\n#'\n#' @param quiet Whether to print any output on screen.\n#'\n#' @details This routine calculates auto correlation separately at the specified lag\n#' for every chain in the mcmc.list, then averages autocorrelations across\n#' chains.\n#'\n#' @return A list with the following components\n#' \\itemize{\n#' \\item autoCorrelated: If the average autocorrelations at the specified\n#' lag are all less than\n#' \\code{criterion}, the chain is declared un-autocorrelated and is set FALSE.\n#' Otherwise, this component of the returns is TRUE.\n#' \\item autoCorrs: the computed autocorrelations at the specified lag.\n#' This value is the average across chains in \\code{obj}.\n#' }\n#'\n#' @author Trent McDonald\n#'\n#'\n#' @seealso \\code{\\link{checkIsConverged}}\n#'\n#' @examples\n#' \\dontrun{\n#' jags = jags.model(file=\"someJAGSFile.txt\",\n#' data=someJAGS.data, inits=someJAGS.inits,\n#' n.chains=3,n.adapt=100)\n#' update(jags, n.iter=1000)\n#' out = coda.samples(jags,\n#' variable.names=c(\"someParameter\"),\n#' n.iter=1000,\n#' thin=2)\n#'\n#' auto <- chechIsAutocorrelated(out)\n#' }\n\n\ncheckIsAutocorrelated <- function(obj, criterion=0.4, lag=1, quiet=FALSE){\n\n if(!quiet) cat(paste0(\"\\n\", paste(rep(\"-\",4),collapse=\"\"),\" Checking autocorrelation:\\n\"))\n vn <- varnames(obj)\n ncovars <- nvar(obj)\n nchains <- nchain(obj)\n acflag2 <- rep(NA,ncovars)\n names(acflag2) <- vn\n for( i in 1:ncovars ){\n tmp <- rep(NA,nchains)\n for(j in 1:nchains){\n tmp.2 <- acf(as.matrix(obj[[j]][,i]), lag.max = lag, plot=FALSE)\n tmp[j] <- tmp.2$acf[tmp.2$lag ==lag,,]\n }\n acflag2[i] <- mean(tmp)\n if(!quiet) cat(paste( vn[i], \": mean(acf[\",lag, \"]) =\", round(acflag2[i],5), \"\\n\") )\n }\n if(any(acflag2>criterion, na.rm=TRUE)){\n if(!quiet) cat(paste0(\"\\nSome autocorrelations exceed \",criterion, \".\\nInspect acfplot(mcmc.list) and increase thinning.\\n\"))\n ans <- TRUE\n } else {\n ans <- FALSE\n }\n\n list(autoCorrelated = ans, autoCorrs=acflag2)\n\n}\n", "meta": {"hexsha": "afd2d4275f92d08dcda570420218b1f7e51756a5", "size": 2759, "ext": "r", "lang": "R", "max_stars_repo_path": "R/checkIsAutocorrelated.r", "max_stars_repo_name": "tmcd82070/EoAR", "max_stars_repo_head_hexsha": "30bdd48e88046332fdb1c97d55fb9a6a1a983e06", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/checkIsAutocorrelated.r", "max_issues_repo_name": "tmcd82070/EoAR", "max_issues_repo_head_hexsha": "30bdd48e88046332fdb1c97d55fb9a6a1a983e06", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/checkIsAutocorrelated.r", "max_forks_repo_name": "tmcd82070/EoAR", "max_forks_repo_head_hexsha": "30bdd48e88046332fdb1c97d55fb9a6a1a983e06", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 33.6463414634, "max_line_length": 129, "alphanum_fraction": 0.679956506, "num_tokens": 823, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7745833737577158, "lm_q2_score": 0.7057850278370112, "lm_q1q2_score": 0.5466893480096755}} {"text": "## 6장 소스코드\n\n# CSV 파일 읽어들이기\nad.data <- read.csv(\"ad_result.csv\", header = T, stringsAsFactors = F)\nad.data\n\n# TV 광고의 광고비용과 신규 유저수의 산점도를 그리기\ninstall.packages(\"ggplot2\")\ninstall.packages(\"scales\")\nlibrary(ggplot2)\nlibrary(scales)\n\nggplot(ad.data, aes(x = tvcm, y = install)) + geom_point() + \n xlab(\"TV 광고비\") + ylab(\"신규 유저수\") + \n scale_x_continuous(label = comma) +\n scale_y_continuous(label = comma) + geom_smooth(method=\"lm\")\n\n# 잡지매체의 광고비용과 신규 유저수의 산점도를 그리기\nggplot(ad.data, aes(x = magazine, y = install)) + geom_point() + \n xlab(\"잡지 광고비\") + ylab(\"신규 유저수\") + \n scale_x_continuous(label = comma) + \n scale_y_continuous(label = comma) + geom_smooth(method=\"lm\")\n\n# 회귀분석 실행\nfit <- lm(install ~ ., data = ad.data[, c(\"install\", \"tvcm\", \"magazine\")])\nfit\n\n# 회귀분석 결과를 해석하기\nsummary(fit)\noptions(scipen=1000)\n\n", "meta": {"hexsha": "5bc054e0fb860e987a59eb2f3ea334c297ec1682", "size": 808, "ext": "r", "lang": "R", "max_stars_repo_path": "ch6.r", "max_stars_repo_name": "yedam-Lee/myfirstrepo", "max_stars_repo_head_hexsha": "39a96f0e6702b4b821cb56ab3ec49a83abe5fb19", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2019-03-13T02:26:29.000Z", "max_stars_repo_stars_event_max_datetime": "2019-03-13T02:26:29.000Z", "max_issues_repo_path": "ch6.r", "max_issues_repo_name": "yedam-Lee/myfirstrepo", "max_issues_repo_head_hexsha": "39a96f0e6702b4b821cb56ab3ec49a83abe5fb19", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ch6.r", "max_forks_repo_name": "yedam-Lee/myfirstrepo", "max_forks_repo_head_hexsha": "39a96f0e6702b4b821cb56ab3ec49a83abe5fb19", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 25.25, "max_line_length": 74, "alphanum_fraction": 0.6707920792, "num_tokens": 347, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8887587905460026, "lm_q2_score": 0.615087848460224, "lm_q1q2_score": 0.5466647322770516}} {"text": "#'@title calcKristCa\n#'\n#'@description Calculate incremental hull (roughness) resistance coefficient\n#'(\\code{Ca}) (dimensionless) using the Kristensen method.\n#'\n#'@param shipType Ship type (vector of strings, see \\code{\\link{calcShipType}}).\n#'Must align with \\code{tankerBulkCarrierGCargoShipTypes} and\n#' \\code{containerShipTypes} groupings\n#'@param actualDisplacement Actual loaded displacement (vector of numericals,\n#' m^3) (see \\code{\\link{calcActualDisp}})\n#'@param tankerBulkCarrierGCargoShipTypes Ship types specified in input\n#'\\code{shipTypes} to be modeled as tankers, bulk carriers and general cargo\n#'vessels (vector of strings)\n#'@param containerShipTypes Ship types specified in input \\code{shipTypes} to be\n#'modeled as container ships (vector of strings)\n#'\n#'@details\n#'Models the effect of realistic hull roughness on resistance, which is not\n#'captured in the frictional and residual resistance coefficients from tank\n#'towing operations.\n#'\n#' This method this requires ship types to be grouped. Use the\n#' \\code{tankerBulkCarrierGCargoShipTypes}, \\code{containerShipTypes} grouping\n#' parameters to provide these ship type groupings. Any ship types not included\n#' in these groupings will be considered as miscellaneous\n#'\n#'@return \\code{Ca} (vector of numericals, dimensionless)\n#'\n#'@references\n#'Kristensen, H. O. and Lutzen, M. 2013. \"Prediction of Resistance and Propulsion\n#'Power of Ships.\"\n#'\n#'\\href{https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}{Kristensen, H. O.\n#'\"Ship-Desmo-Tool.\" https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}\n#'\n#'@seealso \\code{\\link{calcActualDisp}}\n#'\n#'@family Kristensen Calculations\n#'@family Resistance Calculations\n#'\n#'@examples\n#'calcKristCa(c(\"bulk.carrier\",\"container.ship\"),c(73663.27,216726.45))\n#'\n#'@export\n\n\ncalcKristCa<-function(shipType,actualDisplacement,\n tankerBulkCarrierGCargoShipTypes=c(\"tanker\",\"general.cargo\",\"chemical.tanker\",\"liquified.gas.tanker\",\"oil.tanker\",\"other.tanker\",\"bulk.carrier\"),\n containerShipTypes=c(\"container.ship\")){\n\nCa<- ifelse(shipType%in%containerShipTypes,\n (0.5*log10(actualDisplacement)-0.1*(log10(actualDisplacement))^2)/1000,\n ifelse(shipType%in%tankerBulkCarrierGCargoShipTypes,\n (pmax(-0.1, 0.5*log10(actualDisplacement)-0.1*(log10(actualDisplacement))^2))/1000,\n NA\n)\n)\nreturn(Ca)\n}\n", "meta": {"hexsha": "3fac6051cb103ea881bdc147f363ce616f010d1b", "size": 2376, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcKristCa.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcKristCa.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcKristCa.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 39.6, "max_line_length": 167, "alphanum_fraction": 0.7428451178, "num_tokens": 654, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8577680977182186, "lm_q2_score": 0.6370307944803831, "lm_q1q2_score": 0.5464246927693637}} {"text": "# d: dataset, m: mean, sd: width, a: aplitude\nvOrigDeriv = Inf\n\naApproxDeriv = 0\nmApproxDeriv = 0\nsdApproxDeriv = 0\n\ndataDeriv <- convolve(data, c(-1,1), type= 'filter')\n\nfor (sd in sdTab) {\n for(m in mTab) {\n dnFast <- dnorm(1:length(dataDeriv), mean = m, sd = sd)\n \n n <- (log2Population - 1):1\n aTabPow <- 2 ** n\n aApproxCur = 2 ** log2Population\n vAtAApproxCur = Inf\n for (a in aTabPow) {\n v <- var(dataDeriv - (dnFast * aApproxCur))\n vAp <- var(dataDeriv - (dnFast * (aApproxCur + a)))\n vAm <- var(dataDeriv - (dnFast * (aApproxCur - a)))\n \n if (v > vAp && vAm > vAp) {\n aApproxCur = aApproxCur + a\n } else if (v > vAm && vAp > vAm) {\n aApproxCur = aApproxCur - a\n }\n if (v < vAtAApproxCur) {\n aApproxCurMin = aApproxCur\n vAtAApproxCur = v\n }\n }\n if (vAtAApproxCur < vOrigDeriv) {\n sdApproxDeriv = sd\n mApproxDeriv = m\n aApproxDeriv = aApproxCurMin\n vOrigDeriv = vAtAApproxCur\n }\n }\n}\nprint('sdApproxDeriv: ')\nprint(sdApproxDeriv)\nprint('mApproxDeriv: ')\nprint(mApproxDeriv)\nprint('aApproxDeriv: ')\nprint(aApproxDeriv)\nprint('vOrigDeriv: ')\nprint(vOrigDeriv)\n", "meta": {"hexsha": "d5daf001cc407c050d50b20b3dd2891b59fcd719", "size": 1189, "ext": "r", "lang": "R", "max_stars_repo_path": "rstudio/approxDeriv.r", "max_stars_repo_name": "Taack/coronavirus", "max_stars_repo_head_hexsha": "10823cfc094def6f1102f49e9eaa18f7c040e64b", "max_stars_repo_licenses": ["CC0-1.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "rstudio/approxDeriv.r", "max_issues_repo_name": "Taack/coronavirus", "max_issues_repo_head_hexsha": "10823cfc094def6f1102f49e9eaa18f7c040e64b", "max_issues_repo_licenses": ["CC0-1.0"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2020-03-23T05:43:14.000Z", "max_issues_repo_issues_event_max_datetime": "2020-03-23T05:43:14.000Z", "max_forks_repo_path": "rstudio/approxDeriv.r", "max_forks_repo_name": "Taack/coronavirus", "max_forks_repo_head_hexsha": "10823cfc094def6f1102f49e9eaa18f7c040e64b", "max_forks_repo_licenses": ["CC0-1.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 24.2653061224, "max_line_length": 59, "alphanum_fraction": 0.5929352397, "num_tokens": 446, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9046505402422645, "lm_q2_score": 0.6039318337259583, "lm_q1q2_score": 0.5463472596496897}} {"text": "\naux_step03_Weight_of_Norm_matrix <-function(inMatrix, inWeight){\n \n outMat <- inMatrix\n ncols<-dim(inMatrix)[2]\n nrows<-dim(inMatrix)[1]\n \n #iMcalc2 <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=TRUE)\n iMcalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=FALSE)\n \n nC <- dim(iMcalc)[2] # Convert weigths into a matrix in order to perform Mult \n aMmul <- matrix(rep(inWeight, times=nrows), ncol=nC, byrow=TRUE)\n \n auxCalc <- iMcalc * aMmul\n \n outMat[,2:ncols]<-auxCalc\n \n return(outMat)\n}\n\n\n#' \n#' @param Matrices with Benefits and Costs, \n#' @param Vectors with Benefits and Costs weights,\n#' @return Score of Paths \n#' @author Bruno Sousa\n#' @note v1.0 no arguments validation\n#' @title METH_runTOPSIS\n#' @name METH_runTOPSIS\nMETH_runTOPSIS <- function(iMBen, iMCost, iVecBen,iVecCost ){\n # Important Global Variables\n MIN_COST_TOPSIS_TENDENCY <- 0.0001 # To avoid divisions by zero\n BETA_TOPSIS_TENDENCY_WEIGHTS <- 0.5 \n MINSUM_TOPSIS <- 1e-99 # To avoid divisions by zero in normalization\n \n #source(\"libNormalization.R\")\n \n mBen_Criteria <- iMBen\n mCost_Criteria <- iMCost\n \n vBen_weight <- iVecBen\n vCost_weight <- iVecCost\n \n \n # Euclidean Distance\n fAux_Euclidean_Dist<- function(i1, i2){\n adif <- (i1 - i2)^2\n \n return(adif)\n }\n \n \n \n \n #\n # Convert weight costs into Benefit costs\n #\n aux_step01_Tendency_Cnv <- function(inCosts){\n nCol <- dim(inCosts)[2]\n aux <- matrix(inCosts[,2:nCol], ncol=nCol-1)\n \n aux[which(aux==0) ] <- MIN_COST_TOPSIS_TENDENCY \n aux <- 1/aux\n \n return (aux)\n }\n \n \n # \n # Append Benefits and Cost weigths \n #\n aux_step01_Tendend_Weigths <- function(inWeBen, inWeCost){\n \n ret_Ben <- BETA_TOPSIS_TENDENCY_WEIGHTS * inWeBen\n ret_Cost <- (1 - BETA_TOPSIS_TENDENCY_WEIGHTS) * inWeCost\n \n ret_ <- c(ret_Ben, ret_Cost)\n \n return(ret_)\n }\n \n \n #\n # Normalize matrix\n # Using Vector normalization instead of min-max\n #\n aux_step02_Normalization <- function(inMatrix){\n \n #internal function to help in normalization\n fSum <- function(i){\n sumrow <- MINSUM_TOPSIS\n aSu <- sum(i^2) + sumrow\n return(aSu)\n }\n \n outMat <- inMatrix\n ncols<-dim(inMatrix)[2]\n nrows<-dim(inMatrix)[1]\n \n iMcalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=TRUE)\n \n #Apply Sum by col\n auxSum <- apply(iMcalc, 2, FUN=fSum)\n \n nC <- dim(iMcalc)[2]\n aMSum <- matrix(rep(auxSum, times=nrows), ncol=nC, byrow=TRUE)\n auxCalc <- iMcalc / sqrt(aMSum)\n \n outMat[,2:ncols]<-auxCalc\n return(outMat)\n \n }\n \n # \n # Determine PIS according to DiA Method\n #\n aux_step04_PIS <-function(inM){\n ncolu <- dim(inM)[2]\n nrows <- dim(inM)[1]\n iMcalc <- matrix(inM[,2:ncolu], nrow=nrows, ncol=ncolu-1, byrow=TRUE)\n auxRet <- apply(iMcalc, 2, max)\n return (auxRet) \n }\n \n aux_step04_NIS<-function(inM){\n ncolu <- dim(inM)[2]\n nrows <- dim(inM)[1]\n iMcalc <- matrix(inM[,2:ncolu], nrow=nrows, ncol=ncolu-1, byrow=TRUE)\n auxRet <- apply(iMcalc, 2, min)\n return (auxRet)\t\n }\n \n \n #\n # Determine the distance to the ideal points (PIS and NIS)\n #\n aux_step05_distance_to_ideal <- function(inApos, inMatrix){\n #Be careful with the idx of Ideal....\n ncols<- dim(inMatrix)[2]\n nrows<- dim(inMatrix)[1]\n \n auxMBenef <- matrix(ncol=2,nrow=nrows)\n auxMBenef[,1] <- inMatrix[,1] \n \n mCalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=FALSE)\n \n for (nR in seq(from=1, to=nrows)){\n aSumBen <- 0\n for (nC in seq(from=1, to=ncol(mCalc))){\n aSumBen <- aSumBen + fAux_Euclidean_Dist(mCalc[nR, nC] , inApos[nC]) \n }\n auxMBenef[nR,2] <- aSumBen\n }\n \n \n ret_ <- auxMBenef\n return (ret_)\n }\n \n \n #\n # Determine R score\n #\n aux_step06_Ranking<-function(inSepPos, inSepNeg){\n \n nRow <- dim(inSepPos)[1]\n nCol <- dim(inSepPos)[2]\n auxCi <- matrix(ncol=2, nrow=nRow)\n auxCi[,1] <- inSepPos[,1]\n mCalc <- matrix(c(inSepPos[,2], inSepNeg[,2]), ncol=2, nrow=nRow, byrow=FALSE)\n \n \n \n fDist <- function(i){ # Apply a function on pairwise columns\n ipos <- mCalc[i,1]\n ineg <- mCalc[i,2]\n auxCj <- ineg / (ipos + ineg)\n return (auxCj)\n }\n \n # dist\n aDist <- sapply(1:nrow(mCalc), fDist)\n auxCi[,2] <- aDist \n return(auxCi)\n }\n \n \n ncolMB <- dim(mBen_Criteria)[2]\n TPSTOPsisBenefits <- as.matrix(mBen_Criteria )\n \n ncolMC <- dim(mCost_Criteria)[2]\n TPSTOPsisCosts <- as.matrix(mCost_Criteria )\n \n TPSWeiBenTOP <- vBen_weight\n TPSWeiCostTOP <- vCost_weight\n \n #\n # Step 01\n #\n TPSTOPsisCostsTended <- aux_step01_Tendency_Cnv(TPSTOPsisCosts)\n TPSWeiBenTOPall <- aux_step01_Tendend_Weigths(TPSWeiBenTOP, TPSWeiCostTOP)\n TPSTOPsisBenefits <- cbind(TPSTOPsisBenefits, TPSTOPsisCostsTended)\n TPSWeiBenTOP <- TPSWeiBenTOPall\n stopifnot(TPSTOPsisBenefits != NULL)\n \n #\n # Step 02\n #\n TPSTOPsisBenefitsTOP <- aux_step02_Normalization(TPSTOPsisBenefits)\n \n #\n # Step 03\n #\n TPSTOPsisBenefitsTOP <- aux_step03_Weight_of_Norm_matrix(TPSTOPsisBenefitsTOP, TPSWeiBenTOP)\n \n # Step 04 - Ideal Solutions\n TPSApos_Dia <- aux_step04_PIS(TPSTOPsisBenefitsTOP)\n TPSAneg_Dia <- aux_step04_NIS(TPSTOPsisBenefitsTOP)\n \n # Step 05 - Distance to Ideal\n TPSDposBen_Dia <- aux_step05_distance_to_ideal(TPSApos_Dia, TPSTOPsisBenefitsTOP )\n TPSDnegBen_Dia <- aux_step05_distance_to_ideal(TPSAneg_Dia, TPSTOPsisBenefitsTOP)\n \n #Step 06\n #RdistBen_Dia <- aux_step06_Rscore(DposBen_Dia, DnegBen_Dia)\n TPStopsisRanking <- aux_step06_Ranking(TPSDposBen_Dia, TPSDnegBen_Dia) \n #TPStopsisRanking\n \n #print(TPStopsisRanking)\n return(TPStopsisRanking[order(TPStopsisRanking[,2]),])\n}\n\n\n\n#' \n#' @param Matrices with Benefits and Costs, \n#' @param Vectors with Benefits and Costs weights,\n#' @return Score of Paths \n#' @author Bruno Sousa\n#' @note v1.0 no arguments validation\n#' @title METH_runDiA\n#' @name METH_runDiA\nMETH_runDiA <- function(iMBen, iMCost, iVecBen,iVecCost ){\n # Important Global Variables\n MIN_COST_TOPSIS_TENDENCY <- 0.0001 # To avoid divisions by zero\n BETA_TOPSIS_TENDENCY_WEIGHTS <- 0.5 \n MINSUM_TOPSIS <- 1e-99 # To avoid divisions by zero in normalization\n \n #source(\"libNormalization.R\")\n \n mBen_Criteria <- iMBen\n mCost_Criteria <- iMCost\n \n vBen_weight <- iVecBen\n vCost_weight <- iVecCost\n \n \n #\n # Convert weight costs into Benefit costs\n #\n aux_step01_Tendency_Cnv <- function(inCosts){\n nCol <- dim(inCosts)[2]\n aux <- matrix(inCosts[,2:nCol], ncol=nCol-1)\n \n aux[which(aux==0) ] <- MIN_COST_TOPSIS_TENDENCY \n aux <- 1/aux\n \n return (aux)\n }\n \n \n # \n # Append Benefits and Cost weigths \n #\n aux_step01_Tendend_Weigths <- function(inWeBen, inWeCost){\n \n ret_Ben <- BETA_TOPSIS_TENDENCY_WEIGHTS * inWeBen\n ret_Cost <- (1 - BETA_TOPSIS_TENDENCY_WEIGHTS) * inWeCost\n \n ret_ <- c(ret_Ben, ret_Cost)\n \n return(ret_)\n }\n \n \n #\n # Normalize matrix\n # Using Vector normalization instead of min-max\n #\n aux_step02_Normalization <- function(inMatrix){\n \n #internal function to help in normalization\n fSum <- function(i){\n sumrow <- MINSUM_TOPSIS\n aSu <- sum(i^2) + sumrow\n return(aSu)\n }\n \n outMat <- inMatrix\n ncols<-dim(inMatrix)[2]\n nrows<-dim(inMatrix)[1]\n \n iMcalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=TRUE)\n \n #Apply Sum by col\n auxSum <- apply(iMcalc, 2, FUN=fSum)\n \n nC <- dim(iMcalc)[2]\n aMSum <- matrix(rep(auxSum, times=nrows), ncol=nC, byrow=TRUE)\n auxCalc <- iMcalc / sqrt(aMSum)\n \n outMat[,2:ncols]<-auxCalc\n return(outMat)\n \n }\n \n \n # \n # Determine PIS according to DiA Method\n #\n aux_step04_PIS <-function(inM){\n ncolu <- dim(inM)[2]\n nrows <- dim(inM)[1]\n iMcalc <- matrix(inM[,2:ncolu], nrow=nrows, ncol=ncolu-1, byrow=TRUE)\n auxRet <- apply(iMcalc, 2, max)\n return (auxRet) \n }\n \n aux_step04_NIS<-function(inM){\n ncolu <- dim(inM)[2]\n nrows <- dim(inM)[1]\n iMcalc <- matrix(inM[,2:ncolu], nrow=nrows, ncol=ncolu-1, byrow=TRUE)\n auxRet <- apply(iMcalc, 2, min)\n return (auxRet)\t\n }\n \n \n #\n # Determine the distance to the ideal points (PIS and NIS)\n #\n aux_step05_distance_to_ideal <- function(inApos, inMatrix){\n #Be careful with the idx of Ideal....\n ncols<- dim(inMatrix)[2]\n nrows<- dim(inMatrix)[1]\n \n auxMBenef <- matrix(ncol=2,nrow=nrows)\n auxMBenef[,1] <- inMatrix[,1] \n \n mCalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=FALSE)\n #internal fx\n fAux_Positive_Dif <- function(i1, i2){\n adif <- (i1 - i2)\n if (adif < 0) {\n adif <- adif * (-1)\n }\n return(adif)\n }\n \n \n \n for (nR in seq(from=1, to=nrows)){\n aSumBen <- 0\n for (nC in seq(from=1, to=ncol(mCalc))){\n aSumBen <- aSumBen + fAux_Positive_Dif(mCalc[nR, nC] , inApos[nC]) \n }\n auxMBenef[nR,2] <- aSumBen\n }\n \n \n ret_ <- auxMBenef\n return (ret_)\n }\n \n \n #\n # Determine R score\n #\n aux_step06_Rscore<-function(inSepPos, inSepNeg){\n \n nRow <- dim(inSepPos)[1]\n nCol <- dim(inSepPos)[2]\n auxCi <- matrix(ncol=2, nrow=nRow)\n auxCi[,1] <- inSepPos[,1]\n mCalc <- matrix(c(inSepPos[,2], inSepNeg[,2]), ncol=2, nrow=nRow, byrow=FALSE)\n \n # get ideal point\n PIA_pos <- min(inSepPos[,2])\n PIA_neg <- max(inSepNeg[,2])\n \n fDist <- function(i){ # Apply a function on pairwise columns\n ipos <- mCalc[i,1]\n ineg <- mCalc[i,2]\n auxSqrt <- sqrt( (ipos - PIA_pos)^2 + (ineg - PIA_neg)^2 )\n return (auxSqrt)\n }\n \n # dist\n aDist <- sapply(1:nrow(mCalc), fDist)\n auxCi[,2] <- aDist \n return(auxCi)\n }\n \n \n ncolMB <- dim(mBen_Criteria)[2]\n DiATOPsisBenefits <- as.matrix(mBen_Criteria )\n \n ncolMC <- dim(mCost_Criteria)[2]\n DiATOPsisCosts <- as.matrix(mCost_Criteria )\n \n DiAWeiBenTOP <- vBen_weight\n DiAWeiCostTOP <- vCost_weight\n \n #\n # Step 01\n #\n DiATOPsisCostsTended <- aux_step01_Tendency_Cnv(DiATOPsisCosts)\n DiAWeiBenTOPall <- aux_step01_Tendend_Weigths(DiAWeiBenTOP, DiAWeiCostTOP)\n DiATOPsisBenefits <- cbind(DiATOPsisBenefits, DiATOPsisCostsTended)\n DiAWeiBenTOP <- DiAWeiBenTOPall\n stopifnot(DiATOPsisBenefits != NULL)\n \n #\n # Step 02\n #\n DiATOPsisBenefitsTOP <- aux_step02_Normalization(DiATOPsisBenefits)\n DiATOPsisBenefitsTOPNorm <- DiATOPsisBenefitsTOP\n \n #\n # Step 03\n #\n DiATOPsisBenefitsTOP <- aux_step03_Weight_of_Norm_matrix(DiATOPsisBenefitsTOP, DiAWeiBenTOP)\n \n # Step 04\n DiAApos_Dia <- aux_step04_PIS(DiATOPsisBenefitsTOP)\n DiAAneg_Dia <- aux_step04_NIS(DiATOPsisBenefitsTOP)\n \n # Step 05\n DiADposBen_Dia <- aux_step05_distance_to_ideal(DiAApos_Dia, DiATOPsisBenefitsTOP )\n DiADnegBen_Dia <- aux_step05_distance_to_ideal(DiAAneg_Dia, DiATOPsisBenefitsTOP)\n \n #Step 06\n DiARdistBen_Dia <- aux_step06_Rscore(DiADposBen_Dia, DiADnegBen_Dia)\n #DiARdistBen_Dia\n \n \n #print(DiARdistBen_Dia)\n return(DiARdistBen_Dia[order(DiARdistBen_Dia[,2]),])\n \n \n}\n\n\n\n#' \n#' @param Matrices with Benefits and Costs, \n#' @param Vectors with Benefits and Costs weights,\n#' @return Score of Paths \n#' @author Bruno Sousa\n#' @note v1.0 no arguments validation\n#' @title METH_runNMMD\n#' @name METH_runNMMD\nMETH_runNMMD <- function(iMBen, iMCost, iVecBen,iVecCost ){\n require(\"corpcor\")\n require(\"HDMD\")\n \n # Important Global Variables\n MIN_COST_TOPSIS_TENDENCY <- 0.0001 # To avoid divisions by zero\n BETA_TOPSIS_TENDENCY_WEIGHTS <- 0.5 \n MINSUM_TOPSIS <- 1e-99 # To avoid divisions by zero in normalization\n \n #source(\"libNormalization.R\")\n \n mBen_Criteria <- iMBen\n mCost_Criteria <- iMCost\n \n vBen_weight <- iVecBen\n vCost_weight <- iVecCost\n \n \n \n # Function to check if Mahalanobis can be processed\n # This is to avoid errors on singular matrices.\n # TODO: need to understand why this can happen\n #\n # TODO2: Need a positive definite matrix... Probably due to the fact of having few data\n # Some useful urls:\n # http://www.r-bloggers.com/dealing-with-non-positive-definite-matrices-in-r/\n # http://www2.gsu.edu/~mkteer/npdmatri.html\n fAux_Do_Mahalanobis <- function(iM){\n ret <- TRUE\n #auxCov <- cov(iM)\n auxCov2 <- cov.shrink(iM,verbose=FALSE)\n \n #if (det(auxCov)==0){ ret <- FALSE }\n if (!is.positive.definite(auxCov2)){\n ret <- FALSE\n }\n return (ret)\n }\n \n #Mahalanobis Distance\n fAux_Mahalanobis_Dist <- function(iM, me=NULL){\n \n if (length(me)==0){\n auxMean <- apply(iM, 2, mean)\n }else{\n auxMean <- me\n }\n #auxCov <- cov(iM)\n auxCov <- cov.shrink(iM, verbose=FALSE)\n \n # To avoid errors\n if (fAux_Do_Mahalanobis(iM)){\n #print(paste(\"doing Maha for\" ))\n #print(iM)\n #print(auxCov)\n #print(det(auxCov))\n auxDistMaha <- mahalanobis(iM, auxMean, auxCov)\n return (auxDistMaha)\n }else{\n return (-1)\n }\n }\n \n \n \n myPairwise.mahalanobis <- function (x, grouping = NULL, cov = NULL, inverted = FALSE, digits = 5, ...) \n {\n x <- if (is.vector(x)) \n matrix(x, ncol = length(x))\n else as.matrix(x)\n if (!is.matrix(x)) \n stop(\"x could not be forced into a matrix\")\n if (length(grouping) == 0) {\n grouping = t(x[1])\n x = x[2:dim(x)[2]]\n cat(\"assigning grouping\\n\")\n print(grouping)\n }\n n <- nrow(x)\n p <- ncol(x)\n if (n != length(grouping)) {\n cat(paste(\"n: \", n, \"and groups: \", length(grouping), \n \"\\n\"))\n stop(\"nrow(x) and length(grouping) are different\")\n }\n g <- as.factor(grouping)\n g\n lev <- lev1 <- levels(g)\n counts <- as.vector(table(g))\n if (any(counts == 0)) {\n empty <- lev[counts == 0]\n warning(sprintf(ngettext(length(empty), \"group %s is empty\", \n \"groups %s are empty\"), paste(empty, collapse = \" \")), \n domain = NA)\n lev1 <- lev[counts > 0]\n g <- factor(g, levels = lev1)\n counts <- as.vector(table(g))\n }\n ng = length(lev1)\n group.means <- tapply(x, list(rep(g, p), col(x)), mean)\n if (missing(cov)) {\n inverted = FALSE\n cov = cor(x)\n }\n else {\n if (dim(cov)[1] != p && dim(cov)[2] != p) \n stop(\"cov matrix not of dim = (p,p)\\n\")\n }\n Distance = matrix(nrow = ng, ncol = ng)\n dimnames(Distance) = list(names(group.means), names(group.means))\n Means = round(group.means, digits)\n Cov = round(cov, digits)\n Distance = round(Distance, digits)\n for (i in 1:ng) {\n Distance[i, ] = mahalanobis(group.means, group.means[i,], cov, inverted)\n }\n result <- list(means = group.means, cov = cov, distance = Distance)\n result\n }\n \n #Mahalanobis Distance\n fAux_Pairwise.Mahalanobis_Dist <- function(iM, me=NULL){\n \n if (length(me)==0){\n auxMean <- apply(iM, 2, mean)\n }else{\n auxMean <- me\n }\n #auxCov <- cov(iM)\n auxCov <- cov.shrink(iM, verbose=FALSE)\n \n # To avoid errors\n if (fAux_Do_Mahalanobis(iM)){\n #print(paste(\"doing Maha for\" ))\n #print(iM)\n #print(det(auxCov))\n #print(ncol(iM))\n #auxDistMaha <- pairwise.mahalanobis(iM, grouping=t(iM[,1]), digits=5, center=auxMean, cov=auxCov)\n auxDistMaha <- myPairwise.mahalanobis(iM, grouping=t(iM[,1]), digits=5, center=auxMean, cov=auxCov)\n return (auxDistMaha)\n }else{\n return (-1)\n }\n }\n \n \n #\n # Convert weight costs into Benefit costs\n #\n aux_step01_Tendency_Cnv <- function(inCosts){\n nCol <- dim(inCosts)[2]\n aux <- matrix(inCosts[,2:nCol], ncol=nCol-1)\n \n aux[which(aux==0) ] <- MIN_COST_TOPSIS_TENDENCY \n aux <- 1/aux\n \n return (aux)\n }\n \n # \n # Append Benefits and Cost weigths \n #\n aux_step01_Tendend_Weigths <- function(inWeBen, inWeCost){\n \n ret_Ben <- BETA_TOPSIS_TENDENCY_WEIGHTS * inWeBen\n ret_Cost <- (1 - BETA_TOPSIS_TENDENCY_WEIGHTS) * inWeCost\n \n ret_ <- c(ret_Ben, ret_Cost)\n \n return(ret_)\n }\n \n \n #\n # Normalize matrix\n # Using Vector normalization instead of min-max\n #\n aux_step02_Normalization <- function(inMatrix){\n \n #internal function to help in normalization\n fSum <- function(i){\n sumrow <- MINSUM_TOPSIS\n aSu <- sum(i^2) + sumrow\n return(aSu)\n }\n \n outMat <- inMatrix\n ncols<-dim(inMatrix)[2]\n nrows<-dim(inMatrix)[1]\n \n iMcalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=TRUE)\n \n #Apply Sum by col\n auxSum <- apply(iMcalc, 2, FUN=fSum)\n \n nC <- dim(iMcalc)[2]\n aMSum <- matrix(rep(auxSum, times=nrows), ncol=nC, byrow=TRUE)\n auxCalc <- iMcalc / sqrt(aMSum)\n \n outMat[,2:ncols]<-auxCalc\n return(outMat)\n \n }\n \n \n #\n # Weight the normalized matrix, by multiplying each value by its weight\n #\n aux_step03_Weight_of_Norm_matrix <-function(inMatrix, inWeight){\n \n outMat <- inMatrix\n ncols<-dim(inMatrix)[2]\n nrows<-dim(inMatrix)[1]\n \n iMcalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=TRUE)\n \n nC <- dim(iMcalc)[2] # Convert weigths into a matrix in order to perform Mult \n aMmul <- matrix(rep(inWeight, times=nrows), ncol=nC, byrow=TRUE)\n \n auxCalc <- iMcalc * aMmul\n \n outMat[,2:ncols]<-auxCalc\n \n return(outMat)\n }\n \n \n # \n # Determine PIS according to DiA Method\n #\n aux_step04_PIS <-function(inM){\n ncolu <- dim(inM)[2]\n nrows <- dim(inM)[1]\n iMcalc <- matrix(inM[,2:ncolu], nrow=nrows, ncol=ncolu-1, byrow=TRUE)\n auxRet <- apply(iMcalc, 2, max)\n return (auxRet)\t\n }\n \n aux_step04_NIS<-function(inM){\n ncolu <- dim(inM)[2]\n nrows <- dim(inM)[1]\n iMcalc <- matrix(inM[,2:ncolu], nrow=nrows, ncol=ncolu-1, byrow=TRUE)\n auxRet <- apply(iMcalc, 2, min)\n return (auxRet)\t\n }\n \n \n #\n # Determine the distance to the ideal points (PIS and NIS)\n #\n aux_step05_distance_to_idealNMMD <- function(inMatrix){\n #Be careful with the idx of Ideal....\n ncols<- dim(inMatrix)[2]\n nrows<- dim(inMatrix)[1]\n \n #auxMBenef <- matrix(ncol=2,nrow=nrows)\n #auxMBenef[,1] <- inMatrix[,1] \n \n mCalc <- matrix(inMatrix[,2:ncols], nrow=nrows, ncol=ncols-1, byrow=FALSE)\n \n auxMBenef <- fAux_Pairwise.Mahalanobis_Dist(mCalc, colMeans(mCalc))\n \n \n ret_ <- auxMBenef\n return (ret_)\n }\n \n #\n # Determine R score\n #\n aux_step06_RankingNMMD<-function(inPathsIDs, inDist ){\n \n nRow <- dim(inDist)[1]\n nCol <- dim(inDist)[2]\n auxCi <- matrix(ncol=nCol, nrow=nRow)\n #auxCi[,1] <- 1:nRow\n auxCi[,1] <- inPathsIDs \n mCalc <- inDist \n \n \n fDist <- function(i){ # Apply a function on pairwise columns\n auxCj <- sum(mCalc[i,]) / (ncol(mCalc))\n return (auxCj)\n }\n \n # dist\n aDist <- sapply(1:nrow(mCalc), fDist)\n auxCi[,2] <- aDist \n return(auxCi)\n }\n \n ncolMB <- dim(mBen_Criteria)[2]\n NMMDTOPsisBenefits <- as.matrix(mBen_Criteria )\n \n ncolMC <- dim(mCost_Criteria)[2]\n NMMDTOPsisCosts <- as.matrix(mCost_Criteria )\n \n NMMDWeiBenTOP <- vBen_weight\n NMMDWeiCostTOP <- vCost_weight\n \n #\n # Step 01\n #\n \n #Keep Path Ids\n NMMD_PATHIds <- mBen_Criteria[,1]\n \n NMMDTOPsisCostsTended <- aux_step01_Tendency_Cnv(NMMDTOPsisCosts)\n NMMDWeiBenTOPall <- aux_step01_Tendend_Weigths(NMMDWeiBenTOP, NMMDWeiCostTOP)\n NMMDTOPsisBenefits <- cbind(NMMDTOPsisBenefits, NMMDTOPsisCostsTended)\n NMMDWeiBenTOP <- NMMDWeiBenTOPall \n stopifnot(NMMDTOPsisBenefits != NULL)\n \n #\n # Step 02\n #\n #naCol <- dim(NMMDTOPsisBenefits)[2]\n #print(NMMDTOPsisBenefits[,2:naCol])\n #if (!fAux_Do_Mahalanobis(NMMDTOPsisBenefits[,2:naCol])){\n #\tprint(\"Doing Normalization\")\n NMMDTOPsisBenefitsTOP <- aux_step02_Normalization(NMMDTOPsisBenefits)\n #}\n \n #\n # Step 03\n #\n NMMDTOPsisBenefitsTOP <- aux_step03_Weight_of_Norm_matrix(NMMDTOPsisBenefitsTOP, NMMDWeiBenTOP)\n \n # Step 04 - Ideal Solutions\n \n \n # Step 05 - Distance to Ideal\n \n NMMDDistBen_Dia <- aux_step05_distance_to_idealNMMD( NMMDTOPsisBenefitsTOP)\n \n #Step 06\n \n NMMDtopsisRanking <- aux_step06_RankingNMMD(NMMD_PATHIds, NMMDDistBen_Dia$distance) \n #NMMDtopsisRanking\n \n #print(NMMDtopsisRanking)\n #print(\"NMMD\")\n #print(NMMDtopsisRanking[order(-NMMDtopsisRanking[,2])])\n #print(\"\\n\")\n return(NMMDtopsisRanking[order(-NMMDtopsisRanking[,2]),])\n \n \n}\n\n\n", "meta": {"hexsha": "3bcbcfcd77d0464d056acbf45f6649030d2d61dc", "size": 20436, "ext": "r", "lang": "R", "max_stars_repo_path": "METH_related.r", "max_stars_repo_name": "OneSourceConsult/FMC", "max_stars_repo_head_hexsha": "e3f86debeffe5b09d23f62d9cf997348949b5970", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "METH_related.r", "max_issues_repo_name": "OneSourceConsult/FMC", "max_issues_repo_head_hexsha": "e3f86debeffe5b09d23f62d9cf997348949b5970", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "METH_related.r", "max_forks_repo_name": "OneSourceConsult/FMC", "max_forks_repo_head_hexsha": "e3f86debeffe5b09d23f62d9cf997348949b5970", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 25.1674876847, "max_line_length": 105, "alphanum_fraction": 0.6338324525, "num_tokens": 6875, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7772998611746911, "lm_q2_score": 0.702530051167069, "lm_q1q2_score": 0.5460765112432113}} {"text": "\n #par(mfrow=c(2,1), mgp = c(1.5, 0.5, 0),tck=0.02)\n\n V<- 4.66*10^(-13)\n\n #V是基础弯月面体积\n\n vne1<- 0.5*q[1]/fv + 4.66*10^(-13)\n\n\n vne2<- (3*10^(-12))*(1-k)/fv\n\n #vne是体积增量\n\n fv<-c(1:200)\n q<-c(1.5*10^(-12),8.33*10^(-12),16.67*10^(-12),\n 1.5/60*10^(-12),27/60*10^(-12),54/60*10^(-12),3*10^(-12))\n k<-c(0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8)\n\n #fv,q,k分别为电压频率，流量和占空比\n\n plot(fv,vne1, mgp = c(1.5, 0.5, 0),tck=0.02,type=\"b\",pch=1,cex=0.6,lwd=1.5,lty=2,xlab = expression(italic(f[\"v\"])(Hz)),\n ylab = expression (italic(V[\"ne\"]+V[\"m\"](m^3))),xlim=c(0,200),ylim=c(4.66*10^(-13), 9*10^(-13)))\n\n coll<-rainbow(7)\n pcc<-c(1,2,3,4,5,6,7)\n\n for(i in 2:7){\n lines(fv,(0.5*q[i]/fv + 4.66*10^(-13)),col=coll[i],pch=pcc[i],cex=0.6, type=\"b\",lwd=1.5,\n lty=2)\n }\n\n abline(h=(V+0.05*V),col=\"red\",lwd=1.5,lty=4)\n abline(h=(V+0.1*V),col=\"red\",lwd=1.5,lty=4)\n abline(h=(V+0.2*V),col=\"red\",lwd=1.5,lty=4)\n abline(h=(V+0.6*V),col=\"red\",lwd=1.5,lty=4)\n\n legend(\"topright\",c(\"1.5nl/s\",\"8.33nl/s\",\"16.67nl/s\",\n \"1.5nl/min\",\"27nl/min\",\"54nl/min\",\n \"180nl/min\"),col=c(\"black\",\n coll[2],coll[3],coll[4],coll[5],coll[6],coll[7]),\n inset = .02,pch=c(1,2,3,4,5,6,7),bty = \"n\",lty=2,lwd=1.5,cex=0.8)\n", "meta": {"hexsha": "a59d2364385a7a049c494cd0e1ad6279510d7477", "size": 1385, "ext": "r", "lang": "R", "max_stars_repo_path": "thesis/chap4/fig4-1b.r", "max_stars_repo_name": "shuaimeng/r", "max_stars_repo_head_hexsha": "94fa0cce89c89a847b95000f07e64a0feac1eabe", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "thesis/chap4/fig4-1b.r", "max_issues_repo_name": "shuaimeng/r", "max_issues_repo_head_hexsha": "94fa0cce89c89a847b95000f07e64a0feac1eabe", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "thesis/chap4/fig4-1b.r", "max_forks_repo_name": "shuaimeng/r", "max_forks_repo_head_hexsha": "94fa0cce89c89a847b95000f07e64a0feac1eabe", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 32.2093023256, "max_line_length": 126, "alphanum_fraction": 0.4527075812, "num_tokens": 703, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8289387998695209, "lm_q2_score": 0.6584174938590245, "lm_q1q2_score": 0.5457878071725975}} {"text": "\"gen.data\"<-function(funind,n=NULL,p=NULL){\nif(is.null(n)) n=400\nif(is.null(p)) p=1000\ny0=NULL\nif(funind<=4){\ns=3*2^(funind-1)\nu=matrix(rnorm(n*(p-50)),n,p-50)\nv=matrix(rnorm(n*50),n,50)\nw=u[,1:s]\nwcoef=(-1)^(1:s+1)/5\nx0=w%*%wcoef\nx2=as.vector(x0)+sqrt(25-s)/5*v\nx=cbind(u,x2)\nbetatrue=(-1)^(1:s+1)\ny0=w%*%betatrue\ny=y0+rnorm(n)*sqrt(3)\n} else if(funind==5){\ns=3\nx=matrix(rnorm(n*p),n,p)\nx[,2]=-1/3*x[,1]^3+rnorm(n)\ny0= x[,1]+x[,2]+x[,3]\ny=y0+rnorm(n)*sqrt(3)\n}\n#if(funind<=4){\n#s=3*2^(funind-1)\n#u=matrix(rnorm(n*(p-50)),n,p-50)\n#v=matrix(rnorm(n*50),n,50)\n#w=u[,1:s]\n#wcoef=(-1)^(1:s+1)/5\n#x0=w%*%wcoef\n#x2=as.vector(x0)+sqrt(25-s)/5*v\n#x=cbind(u,x2)\n#betatrue=(-1)^(1:s+1)\n#y=w%*%betatrue+rnorm(n)\n#} else if(funind==5){\n#s=3\n#x=matrix(rnorm(n*p),n,p)\n#x[,2]=-1/3*x[,1]^3+rnorm(n)\n#y=x[,1]+x[,2]+x[,3]+rnorm(n)\n#}\n#\nelse if(funind==6){\ns=4\nx=matrix(runif(n*p),n,p)\ny0=5*f1(x[,1])+3*f2(x[,2])+4*f3(x[,3])+6*f4(x[,4])\ny=y0+rnorm(n)*sqrt(1.74)\n} else if(funind==7){\ns=4\nw=matrix(runif(n*p),n,p)\nu=runif(n)\nx=(w+u)/2\ny0=5*f1(x[,1])+3*f2(x[,2])+4*f3(x[,3])+6*f4(x[,4])\ny=y0+rnorm(n)*sqrt(1.74)\n} else if(funind==8){\ns=12\nx=matrix(runif(n*p),n,p)\ny0=(f1(x[,1])+f2(x[,2])+f3(x[,3])+f4(x[,4])+\n1.5*f1(x[,5])+1.5*f2(x[,6])+1.5*f3(x[,7])+1.5*f4(x[,8])+\n2*f1(x[,9])+2*f2(x[,10])+2*f3(x[,11])+2*f4(x[,12]))\ny=y0+rnorm(n)*sqrt(0.5184)\n} else if(funind==9){\ns=12\nw=matrix(runif(n*p),n,p)\nu=runif(n)\nx=(w+u)/2\ny0=(f1(x[,1])+f2(x[,2])+f3(x[,3])+f4(x[,4])+\n1.5*f1(x[,5])+1.5*f2(x[,6])+1.5*f3(x[,7])+1.5*f4(x[,8])+\n2*f1(x[,9])+2*f2(x[,10])+2*f3(x[,11])+2*f4(x[,12]))\ny=y0+rnorm(n)*sqrt(0.5184)\n} else if(funind==10){\ns=4\nx=matrix(rnorm(p*n, mean=0, sd=1), n, p)\nbetatrue <- c(2,2,2,-3*sqrt(2))\ntruerho=0.5\ncorrmat=diag(rep(1-truerho, p))+matrix(truerho, p, p)\ncorrmat[,4]=sqrt(truerho)\ncorrmat[4, ]=sqrt(truerho)\ncorrmat[4,4]=1\ncholmat=chol(corrmat)\nx=x%*%cholmat\ny0 <- x[,1:s]%*%betatrue\ny=y0+rnorm(n)\n} else if(funind==11){\ns=12\nx=matrix(runif(n*p),n,p)\ny=(f1(x[,1])+f2(x[,2])+f3(x[,3])+f4(x[,4])+\n1.5*f1(x[,5])+1.5*f2(x[,6])+1.5*f3(x[,7])+1.5*f4(x[,8])+\n2*f1(x[,9])+2*f2(x[,10])+2*f3(x[,11])+2*f4(x[,12]))+0.5*rnorm(n)*sqrt(0.5184)\n} else if(funind==12){\ns=12\nw=matrix(runif(n*p),n,p)\nu=runif(n)\nx=(w+u)/2\ny=(f1(x[,1])+f2(x[,2])+f3(x[,3])+f4(x[,4])+\n1.5*f1(x[,5])+1.5*f2(x[,6])+1.5*f3(x[,7])+1.5*f4(x[,8])+\n2*f1(x[,9])+2*f2(x[,10])+2*f3(x[,11])+2*f4(x[,12]))+0.5*rnorm(n)*sqrt(0.5184)\n} else if(funind>=18 & funind<=22)\n{\ns=4\nx=matrix(runif(n*p),n,p)\ny0=3*f1(x[,1])+3*f2(x[,2])+2*f3(x[,3])+2*f4(x[,4])\ny=y0+rnorm(n)*sqrt(3.3843)*sqrt(2^(20-funind))\n} else if(funind>=28 & funind<=32)\n{\ns=4\nw=matrix(runif(n*p),n,p)\nu=runif(n)\nx=(w+u)/2\ny0=3*f1(x[,1])+3*f2(x[,2])+2*f3(x[,3])+2*f4(x[,4])\ny=y0+rnorm(n)*sqrt(3.3843)*sqrt(2^(30-funind))\n}\nreturn(data=list(x=x,y=y,n=n,p=p,s=s,y0=y0))\n\n}\n\nf1<-function(x){x}\nf2<-function(x){(2*x-1)^2}\nf3<-function(x){sin(2*pi*x)/(2-sin(2*pi*x))}\nf4<-function(x){0.1*sin(2*pi*x)+0.2*cos(2*pi*x)+0.3*sin(2*pi*x)^2+0.4*cos(2*pi*x)^3+0.5*sin(2*pi*x)^3}\n\n\nsimuNIS<-function(funind=NULL, repgroup=NULL, n=NULL, p=NULL, trace.it=NULL, knots=NULL, maxloop=NULL, eps0=NULL, randSeed = 0)\n{\n if(is.null(funind)) stop('No given data generating scheme.')\n if(is.null(trace.it)) trace.it=FALSE\n if(is.null(maxloop)) maxloop=10\n if(is.null(eps0)) eps0=1e-6\n if(is.null(n)) n=400\n if(is.null(p)) p=1000\n if(is.null(knots)) knots=ceiling(n^0.2)\n\n trace.it=as.logical(trace.it)\n set.seed(randSeed)\n cat('Random seed=', randSeed, '...\\n')\n data=gen.data(funind,n,p)\n testdata=gen.data(funind,n/2,p)\n \n greedINIS.fit = greedINIS(data, testdata, knots=knots, eps0=eps0, DOISIS=TRUE,trace=trace.it)\n \n aINIS.fit = adaptINIS(data, testdata, knots=knots, eps0=eps0, DOISIS=TRUE, maxloop=maxloop,trace=trace.it)\n \n\n return(list(greedyINIS = greedINIS.fit, INIS = aINIS.fit))\n #filename=paste(newpath,'/','funind=',funind,'_knots=',knots,'_r0=',r0,'_r1=',r1,'.RData',sep='')\n #save(aINIS.fit, greedINIS.fit, penGAM.fit,lmsis.fit,results,file=filename)\n}\n\n\n###########################################################################\n#greedINIS: g-INIS algorithm, using knots=n^0.2\n###########################################################################\ngreedINIS <- function(data, testdata=NULL, lambda.pen.list=NULL, folds=NULL, quant=NULL, gnum=1, kfold=NULL, knots=NULL, eps0=1e-6, DOISIS=TRUE, maxloop=20, trace=FALSE, detailed=FALSE){\n t0=proc.time()[1]\n cat('starting greedINIS, g-INIS-penGAM algorithm, adatively choose number of variables\\n')\n x=data$x\n y=data$y\n n <- nrow(x)\n p <- ncol(x)\n\n\n #if(is.null(nsis)) nsis=min(floor(n/log(n)),p-1)\n if(is.null(knots)) knots=ceiling(n^0.2)\n if(is.null(folds)) {\n temp= sample(1:n, n, replace = FALSE)\n if(is.null(kfold)) kfold=5\n for(i in 1:kfold){\n folds[[i]]=setdiff(1:n, temp[seq(i, n, kfold)])\n }\n }\n if(is.null(quant)) quant=1\n df0 <- knots+1\n\n xbs=matrix(0,n,df0*p)\n\n\n for(i in 1:p)\n xbs[,(i-1)*(df0)+(1:df0)]=ns(x[,i],df=df0)\n\n tempresi <- rep(0,p)\n\n curloop=1\n for(i in 1:p)\n {\n\n tempfit <-lm.fit (x=cbind(1,xbs[,(i-1)*df0+1:df0]),y=y)\n tempresi[i] <- sum(tempfit$residuals^2)\n }\n\n\n used.list<- tempresi\n\n used.sort <- sort(used.list, method= \"sh\", index=TRUE, decreasing=FALSE)\n initRANKorder <- used.sort$ix\n\n\n mindex <- sample(1:n)\n mresi=NULL\n for(i in 1:p){\n tempfit <-lm.fit (x=cbind(1,xbs[,(i-1)*df0+1:df0]),y=y[mindex])\n mresi[i] <- sum(tempfit$residuals^2)\n }\n resi.thres = quantile(mresi,1-quant)\n nsis <- min(sum(used.list=maxloop) {\n test=0\n normal.exit=0\n }\n\n }\n final.fit<-fit.tmp$fit\n cv.error<-fit.tmp$cv.error\n if(!is.null(testdata)){\n testx=testdata$x\n testy=testdata$y\n pred.error<-mean((predict(final.fit,as.matrix(testx[,pick.ind]))-testy)^2)\n } else pred.error=NULL\n\n\n ISISind=sort(ISISind)\n SISind=sort(SISind)\n ptime = proc.time()[1]-t0\n cat('finishing adaptINIS...\\n')\n if(detailed){\n return(list(fit=fit.tmp, initRANKorder=initRANKorder, detail.pickind=detail.pickind, detail.ISISind=detail.ISISind,\n SISind=SISind, ISISind=ISISind, nsis=nsis, cv.error=cv.error, pred.error=pred.error, ptime = ptime, normal.exit=normal.exit))\n } else\n {\n return(list(fit=fit.tmp, ISISind=ISISind, cv.error=cv.error, pred.error=pred.error, ptime = ptime, normal.exit=normal.exit))\n }\n}\n\n###########################################################################\n#adaptINIS: INIS-penGAM algorithm, using knots=n^0.2\n###########################################################################\nadaptINIS <- function(data, testdata=NULL, lambda.pen.list=NULL, folds=NULL, quant=NULL, kfold=NULL, knots=NULL, eps0=1e-6, DOISIS=TRUE, maxloop=10, trace=FALSE, detailed=FALSE){\n t0=proc.time()[1]\n cat('starting adaptINIS, INIS-penGAM algorithm, adatively choose number of variables\\n')\n x=data$x\n y=data$y\n n <- nrow(x)\n p <- ncol(x)\n \n \n #if(is.null(nsis)) nsis=min(floor(n/log(n)),p-1)\n if(is.null(knots)) knots=ceiling(n^0.2)\n if(is.null(folds)) {\n temp= sample(1:n, n, replace = FALSE)\n if(is.null(kfold)) kfold=5\n for(i in 1:kfold){\n folds[[i]]=setdiff(1:n, temp[seq(i, n, kfold)])\n }\n }\n if(is.null(quant)) quant=1\n df0 <- knots+1\n \n xbs=matrix(0,n,df0*p)\n \n \n for(i in 1:p)\n xbs[,(i-1)*(df0)+(1:df0)]=ns(x[,i],df=df0)\n \n tempresi <- rep(0,p)\n \n curloop=1\n for(i in 1:p)\n {\n \n tempfit <-lm.fit (x=cbind(1,xbs[,(i-1)*df0+1:df0]),y=y)\n tempresi[i] <- sum(tempfit$residuals^2)\n }\n \n \n used.list<- tempresi\n \n used.sort <- sort(used.list, method= \"sh\", index=TRUE, decreasing=FALSE)\n initRANKorder <- used.sort$ix\n \n \n mindex <- sample(1:n)\n mresi=NULL\n for(i in 1:p){\n tempfit <-lm.fit (x=cbind(1,xbs[,(i-1)*df0+1:df0]),y=y[mindex])\n mresi[i] <- sum(tempfit$residuals^2)\n }\n resi.thres = quantile(mresi,1-quant)\n nsis <- max(min(sum(used.list=maxloop) {\n test=0\n normal.exit=0\n }\n \n }\n final.fit<-fit.tmp$fit\n cv.error<-fit.tmp$cv.error\n if(!is.null(testdata)){\n testx=testdata$x\n testy=testdata$y\n pred.error<-mean((predict(final.fit,testx[,pick.ind])-testy)^2)\n } else pred.error=NULL\n \n \n ISISind=sort(ISISind)\n SISind=sort(SISind)\n ptime = proc.time()[1]-t0\n cat('finishing adaptINIS...\\n')\n if(detailed){\n return(list(fit=fit.tmp, initRANKorder=initRANKorder, detail.pickind=detail.pickind, detail.ISISind=detail.ISISind,\n SISind=SISind, ISISind=ISISind, nsis=nsis, cv.error=cv.error, pred.error=pred.error, ptime = ptime, normal.exit=normal.exit))\n } else\n {\n return(list(fit=fit.tmp, ISISind=ISISind, cv.error=cv.error, pred.error=pred.error, ptime = ptime, normal.exit=normal.exit))\n }\n}\n\n\n#########################################################################################################\ncompModelSize<-function(funind=NULL, n=NULL, p=NULL, repgroup=NULL, r0=NULL, r1=NULL, knots=NULL, maxloop=NULL, eps0=NULL, path=NULL)\n{\n if(is.null(funind)) stop('No given data generating scheme.')\n\n if(is.null(eps0)) eps0=1e-6\n\n if(is.null(n)) n=400\n if(is.null(p)) p=1000\n if(is.null(knots)) knots=ceiling(n^0.2)\n if(is.null(r0)&is.null(r1)){\n if(is.null(repgroup))\n {\n r0=r1=1\n }\n else{\n r0=(repgroup-1)*10+1\n r1=(repgroup)*10\n }\n }\n\n if(is.null(path)) path='../'\n\n ########################################\n ####loading packages\n ########################################\n lib.loc=paste(path, 'packages', sep='')\n library(zoo, lib.loc=lib.loc)\n library(fda, lib.loc=lib.loc)\n library(grplasso, lib.loc=lib.loc)\n library(penGAM, lib.loc=lib.loc)\n library(SIS, lib.loc=lib.loc)\n\n library(mgcv)\n\n source('gen.datanew.r')\n\n newpath=paste('result',funind,sep='')\n if(!file.exists(newpath))\n dir.create(newpath)\n\n s0=gen.data(funind,n,p)$s\n truemodel<-1:s0\n\n sizelist=matrix(0,r1,3)\n outputfilename=paste('results/result', funind,'/',funind,'modelsize.txt',sep='')\n for(repind in r0:r1){\n set.seed(repind)\n cat('Random seed=', repind, '...\\n')\n data=gen.data(funind,n,p)\n sizelist[repind,]=c(getSISSize(data,truemodel=truemodel),getPenGamSize(data,truemodel=truemodel),getLmSISSize(data,truemodel=truemodel))\n cat(repind,sizelist[repind,],'\\n',sep=' ',file=outputfilename, append=T)\n }\n\n filename=paste('result', funind,'/',funind,'_r0=',r0,'_r1=',r1,'modelsize.RData',sep='')\n\n save(truemodel,sizelist,file=filename)\n}\n#######################################################################################################\n#######################################################################################################\ncv.penGAM<- function(data, testdata=NULL, knots=NULL, folds=NULL, kfold=NULL, lambda.pen.list=NULL, model=LinReg(), control = grpl.control(trace=0),\n eps0=1e-6 ,cv.trace=FALSE, detailed=FALSE){\n t0=proc.time()[1]\n x=data$x\n y=data$y\n x=as.matrix(x)\n n=nrow(x)\n\n if(is.null(folds)) {\n temp= sample(1:n, n, replace = FALSE)\n if(is.null(kfold)) kfold=5\n for(i in 1:kfold){\n folds[[i]]=setdiff(1:n, temp[seq(i, n, kfold)])\n }\n }\n kfold=length(folds)\n if(is.null(knots)) knots=ceiling(n^0.2)\n\n if(is.null(lambda.pen.list)){\n lambda.pen.list=seq(1,0.02,-0.02)\n }\n error <- rep(0, length(lambda.pen.list))\n if(cv.trace) cat('Beginning Cross Validation.....\\n')\n\n for(i in 1:kfold){\n if(cv.trace) cat('Cross Validation...', 'fold ', i,'...\\n')\n train <- folds[[i]]\n test <- setdiff(1:n, folds[[i]])\n fit <- penGAM(as.matrix(x[train,]), y[train], lambda.pen=lambda.pen.list, lambda.curv=0,\n knots=knots, model=LinReg(), control = grpl.control(trace=0))\n fit.pred <- predict(fit, as.matrix(x[test,]))\n fit.mat<-t(fit.pred[,1,])\n error <- error+apply((fit.mat-y[test])^2,2,mean)\n }\n\n if(cv.trace) cat('Ending Cross Validation.....\\n')\n best.lambda.pen.ind <- (1:length(lambda.pen.list))[which.min(error)]\n\n\n fit <- penGAM(x, y, lambda.pen=lambda.pen.list[best.lambda.pen.ind], lambda.curv=0,\n knots=knots, model=LinReg(), control = grpl.control(trace=0))\n\n cv.error <- mean((predict(fit,x)-y)^2)\n\n coef.mat <- matrix(fit$coef[1,1,][-1],nrow=knots+2,ncol=ncol(x))\n\n coef.norm <- apply(coef.mat^2, 2, sum)\n\n non.zero.ind <- which(coef.norm>eps0)\n\n if(!is.null(testdata)){\n testx=as.matrix(testdata$x)\n testy=testdata$y\n pred.error<-mean((predict(fit,testx)-testy)^2)\n } else pred.error=NULL\n\n ptime = proc.time()[1]-t0\n if(detailed){\n return (list=list(fit=fit, best.lambda.pen.ind=best.lambda.pen.ind, best.lambda=lambda.pen.list[best.lambda.pen.ind],\n penGAMind=non.zero.ind, cv.error=cv.error, pred.error=pred.error, penGAMfit=fit, ptime = ptime))\n }\n else\n {\n return (list=list(fit=fit, penGAMind=non.zero.ind, cv.error=cv.error, pred.error=pred.error, ptime = ptime))\n }\n\n}\n########################################################################################\ngetSISSize <- function(data, truemodel, knots=NULL){\n x = data$x\n y = data$y\n n = nrow(x)\n p = ncol(x)\n\n if(is.null(knots)) knots=ceiling(n^0.2)\n df0 <- knots+1\n\n xbs=matrix(0,n,df0*p)\n\n\n for(i in 1:p)\n xbs[,(i-1)*(df0)+(1:df0)]=ns(x[,i],df=df0)\n\n tempresi <- rep(0,p)\n\n curloop=1\n for(i in 1:p)\n {\n\n tempfit <-lm.fit (x=cbind(1,xbs[,(i-1)*df0+1:df0]),y=y)\n tempresi[i] <- sum(tempfit$residuals^2)\n }\n\n\n used.list<- tempresi\n\n used.sort <- sort(used.list, method= \"sh\", index=TRUE, decreasing=FALSE)\n amsis.ind <- used.sort$ix\n\n\n for(i in 1:p)\n {\n if(length(setdiff(truemodel,amsis.ind[1:i]))==0) {\n amsis.size=i\n break\n }\n }\n\n return(amsis.size)\n\n}\n#######################################################################################\n########################################################################################\ngetPenGamSize <- function(data, truemodel, lambda.pen.list=NULL, knots=NULL, eps0=1e-6){\n x = data$x\n y = data$y\n n = nrow(x)\n p = ncol(x)\n if(is.null(knots)) knots=ceiling(n^0.2)\n if(is.null(lambda.pen.list)){\n lambda.max=0.99\n lambda.pen.list=lambda.max*(2^seq(0,-10,-0.1))\n }\n lam.len=length(lambda.pen.list)\n\n fit <- penGAM(x, y, lambda.pen=lambda.pen.list, lambda.curv=0,\n knots=knots, model=LinReg(), control = grpl.control(trace=0))\n\n coef.mat <- array(fit$coef[,1,-1],c(lam.len,knots+2,p))\n\n coef.norm <- apply(coef.mat^2, c(1,3), sum)\n\n\n nonzeros = coef.norm[,truemodel]>eps0;\n lambdas = apply(nonzeros, 1, all);\n ##str(fit.a);\n\n pickone = max(which(lambdas==F))+1;\n\n if (pickone==lam.len+1){\n mlasso <- p;\n }\n if (pickoneeps0);\n }\n return(mlasso)\n\n}\n\n########################################################################################\ngetLmSISSize <- function(data, truemodel){\n x = data$x\n y = data$y\n n = nrow(x)\n p = ncol(x)\n\n\n lmsis.corr <- NULL\n for(i in 1:p)\n {\n lmsis.corr[i] <- -abs(cor(x[,i],y))\n }\n lmsis.sort <- sort(lmsis.corr, method= \"sh\", index=TRUE, decreasing=FALSE)\n lmsis.ind <- lmsis.sort$ix\n\n for(i in 1:p)\n {\n if(length(setdiff(truemodel,lmsis.ind[1:i]))==0) {\n lmsis.size=i\n break\n }\n }\n return(lmsis.size)\n\n}\n########################################################################################\ngetAmSISSize <- function(data, truemodel, rank.method='corr'){\n x = data$x\n y = data$y\n n = nrow(x)\n p = ncol(x)\n q <- ceiling(n^0.2)\n amsis.corr <- NULL\n amsis.resi <- NULL\n amsis.norm <- NULL\n for(i in 1:p)\n {\n #if(i%%(round(p/10))==0) cat('First Step:', round(i*100/p), 'percent complete\\n')\n amsis.fit <-gam (y~s(x[,i],k=q,fx=TRUE, bs=\"cr\"))\n amsis.corr[i] <- mean(fitted(amsis.fit)*y)\n amsis.resi[i] <- amsis.fit$deviance\n amsis.norm[i] <- sum(fitted(amsis.fit)^2)\n }\n amsis.list<-switch(rank.method, corr=-amsis.corr, resi=amsis.resi, norm=-amsis.norm)\n amsis.sort <- sort(amsis.list, method= \"sh\", index=TRUE, decreasing=FALSE)\n amsis.ind <- amsis.sort$ix\n lmsis.corr <- NULL\n for(i in 1:p)\n {\n lmsis.corr[i] <- -abs(cor(x[,i],y))\n }\n lmsis.sort <- sort(lmsis.corr, method= \"sh\", index=TRUE, decreasing=FALSE)\n lmsis.ind <- lmsis.sort$ix\n\n for(i in 1:p)\n {\n if(length(setdiff(truemodel,amsis.ind[1:i]))==0) {\n amsis.size=i\n break\n }\n }\n for(i in 1:p)\n {\n if(length(setdiff(truemodel,lmsis.ind[1:i]))==0) {\n lmsis.size=i\n break\n }\n }\n return(size=list(amsis.size=amsis.size,lmsis.size=lmsis.size))\n\n}\n####################\n\n####################################################\n#######################################################################################\n##################################################\n\"gen\"<-function(n,p,funind,t=0){\n if(funind==4){\n w=matrix(runif(n*p),n,p)\n u=runif(n)\n v=runif(n)\n } else {\n w=truncnorm.gen(n,p)\n u=as.vector(truncnorm.gen(n,1))\n v=as.vector(truncnorm.gen(n,1))\n }\n x=matrix(0,n,p)\n x[,1:4]=(w[,1:4]+t*u)/(1+t)\n x[,5:p]=(w[,5:p]+t*v)/(1+t)\n epsi=rnorm(n)\n s=3\n y=switch(funind, -3+3*x[,1]+4*x[,2]-8*x[,3]+5*x[,4]+epsi, -7+8*x[,1]-3*x[,2]+10*x[,3]^3-6*x[,4]*(x[,4]-1)+epsi , -5+8*x[,1]^3+10*x[,2]*(1-x[,2])-10*x[,3]^5-8*x[,4]^2+epsi, -4+4*x[,1]+cos(2*pi*x[,2])-8*x[,3]^3+sqrt(x[,4]*(1-x[,4]))*sin(2*pi*(1+2^((9-4*s)/5))/(x[,4]+2^((9-4*s)/5)))+epsi)\n data=list(x=x,y=y)\n}\n\n############################################################################################################\n\n\"truncnorm.gen\"<-function(n,p){\n norm.trunc<-rnorm(n*p)\n ind= abs(norm.trunc)>2\n test=sum(ind)\n count=0\n while(test>0){\n tmp=rnorm(test)\n test=sum(abs(tmp)>2)\n norm.trunc[ind]= tmp\n ind=abs(norm.trunc)>2\n }\n abs(matrix(norm.trunc,n,p))/2\n}\n\n", "meta": {"hexsha": "03c9b63eb34875ec9a297716f32ca8401f279e41", "size": 22955, "ext": "r", "lang": "R", "max_stars_repo_path": "NISfuns.r", "max_stars_repo_name": "statcodes/NIS", "max_stars_repo_head_hexsha": "73521a69eb0b390b9bc4d7ab8effea0d8709c5d1", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2021-09-09T13:41:42.000Z", "max_stars_repo_stars_event_max_datetime": "2021-11-25T12:50:36.000Z", "max_issues_repo_path": "NISfuns.r", "max_issues_repo_name": "statcodes/NIS", "max_issues_repo_head_hexsha": "73521a69eb0b390b9bc4d7ab8effea0d8709c5d1", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "NISfuns.r", "max_forks_repo_name": "statcodes/NIS", "max_forks_repo_head_hexsha": "73521a69eb0b390b9bc4d7ab8effea0d8709c5d1", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 28.0280830281, "max_line_length": 288, "alphanum_fraction": 0.5725985624, "num_tokens": 8303, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7879311956428946, "lm_q2_score": 0.6926419894793246, "lm_q1q2_score": 0.5457542309229175}} {"text": "# setwd(dirname(rstudioapi::getActiveDocumentContext()$path))\n\n#########################################################################\n# Dependencies\n\n# library(purrr)\n# library(glmnet)\n\n#########################################################################\n# Tuning parameter\n\n.Lambdahat <- function(X, t = .05, m = 500){\n n <- nrow(X)\n g <- rnorm(n * m, 0, 1)\n map(1:m, ~ (X * rnorm(n, 0, 1)) %>%\n apply(2, sum) %>%\n max) %>%\n as.numeric %>%\n quantile(1 - t)\n}\n\n.lambda <- function(X, sigma0hat, t = 0.05, c = 1.1) { 2 * c * sigma0hat * .Lambdahat(X, t = t) / nrow(X) }\n\n.sigma0_u.hat <- function(y, x, beta_est) { sd(y - x %*% beta_est) }\n\n#########################################################################\n# Estimation\n\n# First-stage\n.lambda_j.Alpha0hat <- function(X, Z, sigma0_v, tune_type = \"CV\"){\n px <- ncol(X); pz <- ncol(Z)\n if ( tune_type == \"CV\") {\n # fits <- map(1:px, ~ cv.glmnet(Z, X[,.], intercept = FALSE, nfolds = 5)) %>%\n # ncores <- 100\n # cl <- makeCluster(ncores, type=\"FORK\")\n # fits <- parLapply(cl, 1:px, function(j) cv.glmnet(Z, X[,j], intercept=F, nfolds=5))\n # stopCluster(cl)\n fits <- lapply(1:px, function(j) cv.glmnet(Z, X[,j], intercept=F, nfolds=5))\n fits <- fits %>%\n map(~ list(lambda_j = .$lambda.min,\n alpha0_jhat = coef(., s = \"lambda.min\")))\n } else {\n if ( tune_type == \"semioracle\" ) {\n sigma0_vhat <- sigma0_v\n } else if ( tune_type == \"feasible\" ) {\n # ...\n }\n lambda <- .lambda(Z, sigma0_vhat)\n fits <- map(1:px, ~ list(X_j = X[,.], fit = glmnet(Z, X[,.], intercept = FALSE))) %>%\n map(~ list(lambda_j = lambda,\n alpha0_jhat = predict(.$fit, type = \"coefficients\",\n s = lambda, exact = TRUE, x = Z, y = .$X_j)))\n }\n Alpha0hat <- map(fits, ~ as.numeric(.$alpha0_jhat)[2:(pz+1)]) %>%\n reduce(cbind) %>% { colnames(.) <- NULL; . }\n list(lambda_j = map_dbl(fits, ~ .$lambda_j), Alpha0hat = Alpha0hat)\n}\n\n# Second-stage\n.lambda.beta_Lasso <- function(y, Dhat, sigma0_u, tune_type = \"CV\", no_pen_ids = c()) {\n px <- ncol(Dhat)\n if ( length(no_pen_ids)==0 ) {\n penalties <- rep(1, px)\n } else {\n penalties <- rep(1, px); penalties[no_pen_ids] <- 0\n }\n if ( tune_type == \"CV\") {\n res <- cv.glmnet(Dhat, y, intercept = FALSE, standardize=F, nfolds = 5, penalty.factor=penalties) %>%\n { list(lambda = .$lambda.min,\n beta_Lasso = as.numeric(coef(., s = \"lambda.min\"))[2:(px+1)]) }\n } else {\n if ( tune_type == \"infeasible\" ) {\n sigma0_u.hat <- sigma0_u\n } else if ( tune_type == \"feasible\" ) {\n # ...\n }\n lambda <- .lambda(Dhat, sigma0_u.hat)\n res <- glmnet(Dhat, y, intercept = FALSE, penalty.factor=penalties) %>%\n { list(lambda = lambda,\n beta_Lasso = as.numeric(predict(., type = \"coefficients\", s = lambda,\n exact = TRUE, x = Dhat, y = y))[2:(px+1)]) }\n }\n res\n}\n\n# Empirical Gram matrix\n.Sigmahat <- function(x) { (t(x) %*% x) / nrow(x) }\n\nfind_mus <- function(Sigma.hat) {\n ##########################\n # Using lpSolve\n #\n # px <- ncol(Sigma.hat)\n # Id <- diag(1, px)\n # zeros <- matrix(ncol=px,nrow=px); zeros[,] <- 0\n #\n # c1 <- cbind(Sigma.hat, -1*Sigma.hat, -1*Id, rep(0, px))\n # c2 <- cbind(-1*Sigma.hat, Sigma.hat, -1*Id, rep(0, px))\n # c3 <- cbind(zeros, zeros, diag(1, px), rep(-1, px))\n # A <- rbind(c1, c2, c3)\n #\n # obj <- c(rep(0, 3*px), 1)\n # dir <- rep(\"<=\", 3*px)\n # mus <- lapply(\n # 1:px,\n # function(j) {\n # b1.j <- rep(0,px); b1.j[j] <- -1\n # b2.j <- rep(0,px); b2.j[j] <- 1\n # b3.j <- rep(0,px)\n # b.j <- c(b1.j, b2.j, b3.j)\n # res.j <- lp(direction=\"min\", objective.in=obj, const.mat=A, const.dir=dir, const.rhs=b.j)\n # res.j$objval\n # }\n # ) %>% as.numeric\n # mus\n\n ##########################\n # Using Mosek\n\n px <- ncol(Sigma.hat)\n Id <- diag(1, px)\n zeros <- matrix(ncol=px,nrow=px); zeros[,] <- 0\n mus <- numeric(px)\n\n # constraint marix A\n A1 <- cbind(Sigma.hat, -Id, rep(0,px))\n A2 <- cbind(-Sigma.hat, -Id, rep(0,px))\n A3 <- cbind(zeros, Id, rep(-1, px))\n A <- rbind(A1, A2, A3) %>%\n Matrix(sparse=T)\n\n # lower bounds on variables\n blx <- c(rep(-Inf, px), rep(0, px), 0)\n bux <- c(rep(Inf, 2*px), Inf)\n\n objective <- c(rep(0,2*px), 1)\n\n for ( j in 1:px ){\n blc <- rep(-Inf, 3*px)\n buc1 <- rep(0, px); buc1[j] <- 1\n buc2 <- rep(0, px); buc2[j] <- -1\n buc3 <- rep(0, px)\n buc <- c(buc1, buc2, buc3)\n\n mosek.prob <- list(\n sense=\"min\",\n c=objective,\n A=A,\n bc=rbind(blc, buc),\n bx=rbind(blx, bux)\n )\n mosek.res <- try(mosek(mosek.prob, list(verbose=0, soldetail=1)), silent=TRUE)\n mus[j] <- mosek.res$sol$bas$pobjval\n }\n mus\n}\n\n.Theta.hat_JM <- function(Sigma.hat, n) {\n Theta.hat <- InverseLinfty(Sigma.hat, n)\n Theta.hat\n}\n\n\n.Theta.hat_CLIME <- function(Sigma.hat, mus) {\n #################################\n # lpSolve method\n\n # px <- ncol(Sigma.hat)\n # c1 <- lapply(1:px,\n # function(j) { a <- rep(0,3*px); a[j] <- -1; a[j+px] <- 1; a[j+2*px] <- -1; a } ) %>%\n # reduce(rbind)\n # c2 <- lapply(1:px,\n # function(j) { a <- rep(0,3*px); a[j] <- 1;a[j+px] <- -1; a[j+2*px] <- -1; a } ) %>%\n # reduce(rbind)\n # c3 <- lapply(1:px,\n # function(j) { c(-1*Sigma.hat[j,], Sigma.hat[j,], rep(0, px)) }) %>%\n # reduce(rbind)\n # c4 <- lapply(1:px,\n # function(j) { c(Sigma.hat[j,], -1*Sigma.hat[j,], rep(0, px)) }) %>%\n # reduce(rbind)\n # A <- rbind(c1, c2, c3, c4)\n # dir <- rep(\"<=\", 4*px)\n # obj <- c(rep(0,px), rep(0,px), rep(1,px))\n #\n # Theta.hat <- lapply(\n # 1:px,\n # function(j) {\n # mu <- mus[j]\n # b1 <- rep(mu, px); b1[j] = mu-1\n # b2 <- rep(mu, px); b2[j] = mu+1\n # b <- c(rep(0, 2*px), b1, b2)\n # res_j <- lp(direction=\"min\", objective.in=obj, const.mat=A, const.dir=dir, const.rhs=b)\n # theta_j <- res_j$solution[1:px] - res_j$solution[(px+1):(2*px)]\n # theta_j\n # }\n # ) %>%\n # reduce(rbind)\n # Theta.hat\n\n #################################\n # Mosek method\n\n px <- ncol(Sigma.hat)\n Id <- diag(1, px)\n zeros <- matrix(ncol=px,nrow=px); zeros[,] <- 0\n Theta.hat <- matrix(ncol=px,nrow=px)\n\n # constraint matrix\n A1 <- cbind(Id, -1*Id)\n A2 <- cbind(-1*Id, -1*Id)\n A3 <- cbind(Sigma.hat, zeros)\n A4 <- cbind(-1*Sigma.hat, zeros)\n A <- rbind(A1, A2, A3, A4) %>%\n Matrix(sparse=T)\n\n # lower and upper bounds on variables\n blx <- c(rep(-Inf, px), rep(0, px))\n bux <- rep(Inf, 2*px)\n\n objective <- c(rep(0,px), rep(1,px))\n\n for ( j in 1:px ) {\n mu <- mus[j]\n\n # lower and upper bounds on constraints\n blc <- rep(-Inf, 4*px)\n buc3 <- rep(mu, px); buc3[j] <- mu+1\n buc4 <- rep(mu, px); buc4[j] <- mu-1\n buc <- c(rep(0, 2*px), buc3, buc4)\n\n mosek.prob <- list(\n sense=\"min\",\n c=objective,\n A=A,\n bc=rbind(blc, buc),\n bx=rbind(blx, bux)\n )\n mosek.res <- try(mosek(mosek.prob, list(verbose=0)), silent=TRUE)\n Theta.hat[j,] <- mosek.res$sol$bas$xx[1:px]\n }\n Theta.hat\n}\n\n# De-biased second-stage Lasso estimation\n\n.beta_debiased <- function(y, X, Dhat, beta_Lasso, Thetahat) {\n n <- length(y)\n lambda_kappahat <- t(Dhat) %*% (y - Dhat %*% beta_Lasso) / n\n beta_debiased <- beta_Lasso + Thetahat %*% lambda_kappahat + (Thetahat %*% t(Dhat) %*% (Dhat - X) %*% beta_Lasso)/n\n beta_debiased\n}\n\n#########################################################################\n# Standard errors\n\n# SE1\n# This is given by sigma_u.hat * theta_j^T * Sigma_d.hat * theta_j\n.SE1 <-function(Sigma_d.hat, Theta.hat, sd_u.hat) {\n sd_u.hat * (Theta.hat %*% Sigma_d.hat %*% t(Theta.hat)) %>% diag %>% sqrt\n}\n\n# SE2\n# This is given by E_n[{theta_j, d_i}^2u_i^2] %>% sqrt\n.SE2 <- function(D.hat, Theta.hat, u.hat) {\n # n <- length(u.hat); p <- ncol(D.hat)\n (Theta.hat %*% t(D.hat)) %>%\n apply(1, function(X) { mean(X^2 * u.hat^2) }) %>%\n sqrt\n}\n\n# SE3\n# This is given by sigma_u.hat *\n.SE3 <- function(Theta.hat, sd_u.hat) {\n sqrt(diag(Theta.hat)) * sd_u.hat\n}\n", "meta": {"hexsha": "c25d11da7869ffcea7b5741c2d0648b463fbf7ce", "size": 8129, "ext": "r", "lang": "R", "max_stars_repo_path": "src/estimation.r", "max_stars_repo_name": "LedererLab/InstrumentalVariables", "max_stars_repo_head_hexsha": "35d58b0555d334a669283e7f680248bf041c69fd", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2018-09-02T11:16:13.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-11T09:14:00.000Z", "max_issues_repo_path": "src/estimation.r", "max_issues_repo_name": "LedererLab/InstrumentalVariables", "max_issues_repo_head_hexsha": "35d58b0555d334a669283e7f680248bf041c69fd", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/estimation.r", "max_forks_repo_name": "LedererLab/InstrumentalVariables", "max_forks_repo_head_hexsha": "35d58b0555d334a669283e7f680248bf041c69fd", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2019-07-23T16:09:54.000Z", "max_forks_repo_forks_event_max_datetime": "2022-01-11T09:14:03.000Z", "avg_line_length": 29.4528985507, "max_line_length": 117, "alphanum_fraction": 0.5007996063, "num_tokens": 2993, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8244619091240701, "lm_q2_score": 0.6619228691808011, "lm_q1q2_score": 0.5457301924176854}} {"text": "# écriture de stochastic_model sans array\n# avec output true\n\n# modèle simplifié\n\n# écart de production -------------------\nog <- lagog + dog\ndog <- og_lag * (lagog - lag2og) +\n - og_mce * (lagog - 0) + \n og_k * ib + og_noise\n\nupdate(lagog) <- og\nupdate(lag2og) <- lagog\nupdate(og_noise) <- ogn_ar * og_noise + rnorm(0, ogn_sigma)\n\n# dette ------------\ndette <- lagdette + ddette\nupdate(lagdette) <- dette\nupdate(lag2dette) <- lagdette\nddette <- lagrho*lagdette - surplus\n\n# rho écart critique\nrho <- lagrho + drho\nupdate(lagrho) <- rho\ndrho <- -rho_mce*(lagrho - rhostar) + rho_d*(lagdette - dstarr)\n\n# surplus\nsurplus <- lagsurplus + sur_og * (og - lagog) - ib\nupdate(lagsurplus) <- surplus\nupdate(lag2surplus) <- lagsurplus\n\n# règle budgétaire --------------\nsstar <- rhostar*dstar\n\nib <- tpo_sstar * (lagsurplus - sstar ) + tpo_1sstar * (lagsurplus - lag2surplus) + tpo_sstar2 * (lagsurplus - sstar)^2 +\n tpo_dstar * (lagdette - dstar) + tpo_1dstar * (lagdette - lag2dette) + tpo_dstar2 * (lagdette - dstar)^2 +\n tpo_og*(lagog) + tpo_og2*(lagog)^2 + tpo_1og*(lagog - lag2og) +\n tpo_rho*(rho - rhostar) +\n phi[step]\n\n# fonction de perte ----------------------\n# utilisée pour la partie optimisation\nupdate(loss_d_only) <- loss_d_only + loss_d * (step >= loss_t)*(lagdette - dstar)^2\ninitial(loss_d_only) <- 0\nupdate(loss_no_d) <- loss_no_d + og^2/(1+loss_df)^(step-1) + loss_dog * (og-lagog)^2 + loss_ib * (ib)^2\ninitial(loss_no_d) <- 0\n\n# sorties ------\n \nupdate(dettep_o) <- dette \nupdate(og_o) <- og\nupdate(surplus_o) <- surplus\nupdate(ib_o) <- ib\nupdate(rho_o) <- rho\nupdate(loss_o) <- loss_no_d + loss_d_only\nupdate(loss_nd_o) <- loss_no_d\n\ninitial(surplus_o) <- 0\ninitial(og_o) <- 0\ninitial(dettep_o) <- 0\ninitial(ib_o) <- 0\ninitial(rho_o) <- 0\ninitial(loss_o) <- 0\ninitial(loss_nd_o) <- 0\n\n# init ----------------\ni_lagog <- user(0)\ni_lag2og <- user(0)\ni_lagrho <- user(0)\ni_lagdette <- user(1)\ni_lagsurplus <- user(-0.01)\ni_lag2surplus <- user(-0.01)\ni_lag2dette <- user(1)\n\n## Initial conditions\n# initial(og) <- init_og\ninitial(lagog) <- i_lagog\ninitial(lag2og) <- i_lag2og\ninitial(og_noise) <- i_ogn\ninitial(lagrho) <- i_lagrho\ninitial(lagdette) <- i_lagdette\ninitial(lag2dette) <- i_lag2dette\ninitial(lagsurplus) <- i_lagsurplus\ninitial(lag2surplus) <- i_lag2surplus\n\n# params -------------\nperiods <- user(50)\n\n## parameters\nog_lag <- user(0)\nog_mce <- user(0.15)\nog_k <- user(0.7)\nrho_d <- user(0.01)\nrho_mce <- user(0.15)\ndstar <- user(0.6)\ndstarr <- user(0.6)\ntpo_sstar <- user(0.4)\ntpo_dstar <- user(-0.1)\ntpo_sstar2 <- user(0)\ntpo_1sstar <- user(0)\ntpo_dstar2 <- user(0)\ntpo_1dstar <- user(0)\ntpo_1og <- user(0)\ntpo_og <- user(0)\ntpo_og2 <- user(0)\ntpo_rho <- user(0)\nloss_df <- user(0.02)\nloss_ib <- user(0)\nogn_ar <- user(0)\nogn_sigma <- user(0.005)\nrhostar <- user(-0.015)\ni_ogn <- user(0.005)\nsur_og <- user(0.5)\n\n# loss\nloss_dog <- user(0)\nloss_d <- user(0.5)\nloss_t <- user(20)\n\n# controls&exogenes -----------\nphi[] <- user(0)\n\ndim(phi) <- periods", "meta": {"hexsha": "dcb0f3c3b10027b60b9af61c2676267c7dd4b18e", "size": 3001, "ext": "r", "lang": "R", "max_stars_repo_path": "odin/dwrsimple.r", "max_stars_repo_name": "OFCE/debtwatchR", "max_stars_repo_head_hexsha": "53c71e386c923dae0d98c90c79d65e3605b018d2", "max_stars_repo_licenses": ["CECILL-B"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2021-10-19T21:52:56.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-03T19:33:06.000Z", "max_issues_repo_path": "odin/dwrsimple.r", "max_issues_repo_name": "OFCE/dwr", "max_issues_repo_head_hexsha": "53c71e386c923dae0d98c90c79d65e3605b018d2", "max_issues_repo_licenses": ["CECILL-B"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "odin/dwrsimple.r", "max_forks_repo_name": "OFCE/dwr", "max_forks_repo_head_hexsha": "53c71e386c923dae0d98c90c79d65e3605b018d2", "max_forks_repo_licenses": ["CECILL-B"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 24.3983739837, "max_line_length": 121, "alphanum_fraction": 0.6471176275, "num_tokens": 1140, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8198933183101078, "lm_q2_score": 0.6654105521116443, "lm_q1q2_score": 0.545565665609377}} {"text": "# take input from the user\nnum = as.integer(readline(prompt = \"Enter a number: \"))\nif(num < 0) {\nprint(\"Enter a positive number\")\n} else {\nsum = 0\n# use while loop to iterate until zero\nwhile(num > 0) {\nsum = sum + num\nnum = num - 1\n}\nprint(paste(\"The sum is\", sum))\n}", "meta": {"hexsha": "04ba8dd2348d07122b49743dece1d692f4d1dad1", "size": 268, "ext": "r", "lang": "R", "max_stars_repo_path": "find_sum_of_natural_numbers.r", "max_stars_repo_name": "MarcosFloresta/R_Examples", "max_stars_repo_head_hexsha": "1ca3a7db0c7c82b352653ef70a6c7311f295b212", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "find_sum_of_natural_numbers.r", "max_issues_repo_name": "MarcosFloresta/R_Examples", "max_issues_repo_head_hexsha": "1ca3a7db0c7c82b352653ef70a6c7311f295b212", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "find_sum_of_natural_numbers.r", "max_forks_repo_name": "MarcosFloresta/R_Examples", "max_forks_repo_head_hexsha": "1ca3a7db0c7c82b352653ef70a6c7311f295b212", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 20.6153846154, "max_line_length": 55, "alphanum_fraction": 0.6455223881, "num_tokens": 84, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.679178699175393, "lm_q2_score": 0.8031738034238807, "lm_q1q2_score": 0.5454985390211841}} {"text": "# 4. faza: Analiza podatkov\n\n#LINEARNA REGRESIJA\n\nlibrary(ggplot2)\nlibrary(GGally)\n\ng <- ggplot(tabela_tri_zdruzene, aes(x=skupaj.x, y=skupaj.y)) + geom_point()\ngraf13 <- g + geom_smooth(method=\"lm\", formula=y ~ x) + labs(x = \"Število prebivalcev (v stotisočicah)\", y = \"Število dijakov\", title = \"Število dijakov v Sloveniji na leto \\nglede na število prebivalcev v istem letu\")\n\n\np <- ggplot(tabela_tri_zdruzene, aes(x=skupaj.x, y=skupaj)) + geom_point()\ngraf14 <- p + geom_smooth(method=\"lm\", formula=y ~ x)+ labs(x = \"Število prebivalcev (v stotisočicah)\", y = \"Število diplomantov\", title = \"Število diplomantov v Sloveniji na leto \\nglede na število prebivalcev v istem letu\")\n", "meta": {"hexsha": "b978efcf76e6864db5c79f103e4a3c153a88f3ad", "size": 685, "ext": "r", "lang": "R", "max_stars_repo_path": "analiza/analiza.r", "max_stars_repo_name": "OkornA18/APPR-2020-21", "max_stars_repo_head_hexsha": "e9b1ea939963fd35796eef0913f4a8861d3bbcf4", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "analiza/analiza.r", "max_issues_repo_name": "OkornA18/APPR-2020-21", "max_issues_repo_head_hexsha": "e9b1ea939963fd35796eef0913f4a8861d3bbcf4", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2020-12-17T11:04:49.000Z", "max_issues_repo_issues_event_max_datetime": "2021-02-10T12:25:07.000Z", "max_forks_repo_path": "analiza/analiza.r", "max_forks_repo_name": "OkornA18/APPR-2020-21", "max_forks_repo_head_hexsha": "e9b1ea939963fd35796eef0913f4a8861d3bbcf4", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-01-13T20:48:44.000Z", "max_forks_repo_forks_event_max_datetime": "2021-01-13T20:48:44.000Z", "avg_line_length": 48.9285714286, "max_line_length": 226, "alphanum_fraction": 0.7226277372, "num_tokens": 271, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8031737963569014, "lm_q2_score": 0.679178699175393, "lm_q1q2_score": 0.5454985342214423}} {"text": "#' Calculate TSI values from input data\n#'\n#' This function calculates TSI values according to Utah's IR methods from input data containing values for of chlorophyll, total phosphorus, and secchi disk depth.\n#' Note that inputs for these parameters must be specified in units of ug/L, mg/L, and meters, respectively.\n#' @param x Input dataset\n#' @param in_format One of \"wide\" or \"flat\" to specify data input format. Note that only wide format inputs are currently supported.\n#' @param chl_ugL Name of chlorophyll-a variable in ug/L\n#' @param TP_mgL Name of total phosphorus variable in mg/L\n#' @param SD_m Name of secchi disk depth variable in m\n#' @examples \n#' data(ul_trophic)\n#' head(ul_trophic)\n#' tsi=calcTSI(ul_trophic,chl_ugL=\"ChlA\",TP_mgL=\"Phosphate.phosphorus.Total\",SD_m=\"Depth.Secchi.disk.depth\")\n#' head(tsi)\n#' plot(TSIchl~ChlA,tsi)\n\n#' @export\ncalcTSI=function(x,in_format=\"wide\",chl_ugL=\"Chlorophyll a\",TP_mgL=\"Phosphate-phosphorus\",SD_m=\"Depth, Secchi disk depth\", value_var='ResultMeasureValue', param_var='CharacteristicName'){\n#x=trophic_data\n#x=within(x, {\n#\tIR_Value[IR_Value == 0] = 0.001\n#})\n#any(x[,'IR_Value'] == 0, na.rm=T)\n#chl_ugL=\"Chlorophyll a\"\n#TP_mgL=\"Phosphate-phosphorus\"\n#SD_m=\"Depth, Secchi disk depth\"\n#value_var='IR_Value'\n#param_var='CharacteristicName'\n\n#default input units: chla (ug/L), TP (mg/L), SD (m)\n\tif(in_format==\"wide\"){\n\t\tTSIchl=9.81*log(x[,chl_ugL])+30.6\n\t\tTSItp=14.2*log(x[,TP_mgL]*1000)+4.15\n\t\tTSIsd=60-14.41*log(x[,SD_m])\n\t\ttsi=cbind(x, TSIchl, TSItp, TSIsd)\n\t}\n\tif(in_format==\"flat\"){\n\t\tx$TSI=NA\n\t\tx$TSI[x[,param_var] == chl_ugL] = 9.81*log(x[x[,param_var]==chl_ugL,value_var])+30.6\n\t\tx$TSI[x[,param_var] == TP_mgL] =14.2*log(x[x[,param_var]==TP_mgL,value_var]*1000)+4.15\n\t\tx$TSI[x[,param_var] == SD_m] = 60-14.41*log(x[x[,param_var]==SD_m,value_var])\n\t\ttsi=x\n\t}\n\treturn(tsi)\n}\n\n", "meta": {"hexsha": "f66090924a28ced341ea8fdd2f719b3d12ce4514", "size": 1837, "ext": "r", "lang": "R", "max_stars_repo_path": "R/calcTSI.r", "max_stars_repo_name": "utah-dwq/udwqTools", "max_stars_repo_head_hexsha": "e5016f35d310fe01d1defb1652bfa103f84d574c", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 4, "max_stars_repo_stars_event_min_datetime": "2020-10-07T16:50:59.000Z", "max_stars_repo_stars_event_max_datetime": "2021-07-07T17:44:33.000Z", "max_issues_repo_path": "R/calcTSI.r", "max_issues_repo_name": "utah-dwq/udwqTools", "max_issues_repo_head_hexsha": "e5016f35d310fe01d1defb1652bfa103f84d574c", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 5, "max_issues_repo_issues_event_min_datetime": "2019-01-19T00:39:03.000Z", "max_issues_repo_issues_event_max_datetime": "2021-02-23T22:39:49.000Z", "max_forks_repo_path": "R/calcTSI.r", "max_forks_repo_name": "utah-dwq/udwqTools", "max_forks_repo_head_hexsha": "e5016f35d310fe01d1defb1652bfa103f84d574c", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-07-26T09:20:56.000Z", "max_forks_repo_forks_event_max_datetime": "2021-07-26T09:20:56.000Z", "avg_line_length": 39.085106383, "max_line_length": 187, "alphanum_fraction": 0.7136635819, "num_tokens": 635, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8418256393148982, "lm_q2_score": 0.6477982315512489, "lm_q1q2_score": 0.5453331604226905}} {"text": "#' f_smoothh1\n#'\n#' Construct the $K_\\tau K_1$ vector $h_1$\n#'\n#' @param n1 matrix n1 output from f_n1\n#' @param g1 $G_1$ diagonal matrix\n#' @param g2 $G_2$ diagonal matrix\n#' @param sigmatau $\\Sigma_tau$ diagonal matrix\n#'\n#'\n#' @return $K_\\tau K_1$ vector $H_1$\n#'\n#' @examples\n#' \\dontrun{\n#' devtools::load_all()\n#'k1 <- 4; ktau <- 4; m1 <- 4; mtau <- 4; n <- 400\n#' v_dim <- c(ktau + mtau, k1 + m1)\n#' v_x <- runif(n); v_y <- log(v_x); sigma <- 0.4 * (0.5 + v_x)\n#' n1 <- f_n1(v_x, k1, m1); beta <- rnorm(prod(v_dim[1], v_dim[2]))\n#' m_sigmatau <- f_sigmatau(v_dim[1])\n#' m_g1 <- f_g1(ktau, mtau)\n#' m_g2 <- f_g2(ktau, mtau)\n#' #h1\n#' m_h1 <- f_smoothh1(n1, m_g1, m_g2, m_sigmatau)\n#' m_h1$h1\n#' m_h1$h0\n#' }\nf_smoothh1 <- function(n1, g1, g2, sigmatau) {\n #form N^{m_tau+1}(1) and N^{m_tau+2}(\\tau)|_0^1\n nm1 <- rep(0, nrow(sigmatau)); nm1[nrow(sigmatau)] <- 1\n h_1 <- nm1 - tcrossprod(g2, sigmatau) %*% nm1\n h_1 <- tcrossprod(g1, sigmatau) %*% h_1\n h_1 <- kronecker(h_1, t(n1))\n h_1 <- rowSums(h_1)\n\n h_0 <- tcrossprod(g1, sigmatau) %*% nm1\n h_0 <- kronecker(h_0, t(n1))\n h_0 <- 0.5 * rowSums(h_0) - h_1\n return(list(h1 = h_1,\n h0 = h_0))\n}\n\n\n\n#' f_smoothres\n#'\n#' Construct the $n n_tau$ residual vector\n#'\n#' @param ndesign design matrix $n n_\\tau \\times K_\\tau K_1$ $N(\\tau, x)$\n#' @param v_y vector of observations y_i\n#' @param sigma $\\Sigma$ diagonal matrix\n#' @param til_beta tilde_beta coefficients\n#'\n#' @param h bandwidth value\n#' @return $n n_tau$ residual vector\n#'\n#'\n#' @examples\n#' \\dontrun{\n#' s_ntau <- 500; v_taus <- seq(0, 1, length.out = s_ntau)\n#' m_ndesign <- kronecker(f_ntau(v_taus, ktau, mtau), n1)\n#' m_sigma <- f_sigma(ktau + mtau, k1 + m1)\n#' til_beta <- tilde_beta(beta, k1 + m1)\n#' h <- max((((log(n) + 1) / n) ^ 0.4), 0.05)\n#' v_res <- f_smoothres(m_ndesign, v_y, m_sigma, til_beta)\n#' }\nf_smoothres <- function(ndesign, v_y, sigma, til_beta, h) {\n #Calculate the residual\n c(ndesign %*% sigma %*% til_beta - v_y) / h\n}\n\n\n#' f_smoothhtau\n#'\n#' Construct the $K_\\tau K_1$ vector $h_\\tau$\n#'\n#' @param ndesign design matrix $n n_\\tau \\times K_\\tau K_1$ $N(\\tau, x)$\n#' @param pres error function of the $K_\\tau n_tau$ residual vector devided by h\n#' @param s_ntau number of taus values\n#'\n#'\n#' @return $K_\\tau K_1$ vector $h_\\tau$\n#'\n#'\n#' @examples\n#' \\dontrun{\n#' s_ntau <- 500; v_taus <- seq(0, 1, length.out = s_ntau)\n#' m_ndesign <- kronecker(f_ntau(v_taus, ktau, mtau), n1)\n#' m_sigma <- f_sigma(ktau + mtau, k1 + m1)\n#' til_beta <- tilde_beta(beta, k1 + m1)\n#' h <- max((((log(n) + 1) / n) ^ 0.4), 0.05)\n#'\n#' ptm <- proc.time()# Start the clock!\n#' v_res <- f_smoothres(m_ndesign, v_y, m_sigma, til_beta)\n#' print(proc.time() - ptm)# Stop the clock\n#' #Using the Gaussian kernel\n#' v_pres <- pnorm(v_res)\n#' m_htau <- f_smoothhtau(m_ndesign, v_pres, h, s_ntau)\n#' }\nf_smoothhtau <- function(ndesign, pres, s_ntau) {\n c(crossprod(pres, ndesign) / s_ntau)\n}\n\n\n\n#' f_smoothderiv\n#'\n#' Construct the $K_\\tau K_1$ gradient vector for the loss\n#'\n#' @param til_beta the current estimation fro $\\tilde{\\beta}$\n#' @param v_dim vector contain the dimension of covariates\n#' $(K_\\tau, K_1, K_2, K_3, \\cdots)$\n#'\n#' @param m_sigma $K_\\tau K_1K_2 \\times K_1K_2 \\Sigma$ diagonal block matix.\n#' @param h1 $K_\\tau K_1K_2$ vector\n#' @param h2 $K_\\tau K_1K_2$ vector\n#'\n#'\n#' @return the $K_\\tau K_1$ gradient vector for the loss\n#'\n#' @examples\n#' \\dontrun{\n#' f_smoothderiv(til_beta, v_dim, m_sigma, m_htau, m_h1)\n#' }\nf_smoothderiv <- function(til_beta, v_dim, m_sigma, htau, h1) {\n m_c <- f_matc(v_dim, til_beta)\n lossderiv <- m_c %*% crossprod(m_sigma, (htau - h1))\n return(c(lossderiv))\n}\n\n#' f_smoothgrad\n#'\n#' Construct the $K_\\tau K_1K_2$ gradient vector for $\\beta$\n#'\n#' @param beta the current estimation fro $\\beta$\n#' @param loss_deriv the derivative of the loss\n#' @param s $K_\\tau K_1K_2 \\times K_1K_2$ $S$ matrix\n#'\n#'\n#' @return the $K_\\tau K_1K_2$ gradient vector for $\\beta$\n#'\n#' @examples\n#' \\dontrun{\n#' f_grad(beta, v_lossderiv, m_s)\n#' }\nf_smoothgrad <- function(beta, loss_deriv, s) {\n loss_deriv + c(s %*% beta)\n}\n\n#' f_smoothloss\n#'\n#' Return the quantile loss criteria\n#'\n#' @param res $K_\\tau n_tau$ residual vector\n#' @param pres error function of the $K_\\tau n_tau$ residual vector devided by h\n#' @param h0sig $K_\\tau K_1$ vector $H_0$ times m_sigma\n#' @param til_beta estimation for $\\tilde{\\beta}$\n#'\n#' @param s_ntau number of taus values\n#' @param h bandwidth value\n#' @param s_n number of observation\n#'\n#' @return the smooth quantile loss\n#'\nf_smoothloss <- function(res, pres, h0sig, til_beta,\n s_ntau, h, s_n) {\n loss <- crossprod(h0sig, til_beta)\n resloss <- h / 2 * sqrt(2 / pi) * sum(exp(- (- res / 2)^2))\n resloss <- resloss - crossprod(res, (1 - 2 * pres))\n loss <- loss + resloss / s_ntau\n return(c(loss) / s_n)\n}\n\n\n#' f_smoothcrit\n#'\n#' Return the smooth QR criteria\n#'\n#' @param res $n n_tau$ residual vector\n#' @param pres error function of the $K_\\tau n_tau$ residual vector devided by h\n#' @param h0sig $K_\\tau K_1$ vector $H_0$ times m_sigma\n#' @param til_beta estimation for $\\tilde{\\beta}$\n#' @param beta the current estimation for $\\beta$\n#'\n#' @param s_ntau number of taus values\n#' @param h bandwidth value\n#' @param s_n number of observation\n#' @param s $K_1K_2 \\times K_1K_2$ $S$ matrix\n#'\n#' @return the smooth quantile criteria\n#'\nf_smoothcrit <- function(res, pres, h0sig, til_beta, beta,\n s_ntau, h, s_n, s) {\n loss <- f_smoothloss(res, pres, h0sig, til_beta, s_ntau, h, s_n)\n criteria <- loss + crossprod(beta, s %*% beta)\n return(c(criteria))\n}\n\n#' f_smoothbbgd\n#'\n#' The descent algorithm using bb stepsize\n#'\n#' @param v_y vector of y\n#' @param v_x vector of x\n#' @param v_k0 vector of internal knots for covariates\n#' @param v_m vector of orders for covariates\n#'\n#' @param init initialization for $\\beta$\n#' @param h bandwidth value\n#' @param s_ntau number of taus for numerical integration\n#' @param v_smooth vector of smoothing parameters\n#' @param maxit maximum number of iterations\n#' @param eps stopping critria threshhold\n#' @param eta stopping criteria for gradient size\n#' @param trace indicator whether to plot trace\n#'\n#'\n#' @return the estimated $\\beta$\n#'\nf_smoothbbgd <- function(v_y, v_x, v_k0, v_m, init = NULL,\n v_smooth = rep(1, 3),\n h=max((((log(length(v_y))+1)/length(v_y))^0.4), 0.05),\n s_ntau = 200,\n maxit = 1e4, eps = 1e-3, eta = 1e-3,\n trace = NULL, echostep = F) {\n\n#-------------------------------------------------------\n##### set up #####\n#-------------------------------------------------------\n function_call <- match.call()\n x_ord <- order(v_x)\n v_x <- v_x[x_ord]\n v_y <- v_y[x_ord]\n s_center <- mean(v_x)\n s_sd <- sd(v_x)\n v_x <- c(scale(v_x))\n iter <- 0L\n converge <- F\n s_n <- length(v_y)\n #Define dimensions\n k1 <- v_k0[2]; ktau <- v_k0[1]; m1 <- v_m[2]; mtau <- v_m[1]\n v_dim <- v_k0 + v_m\n #Require\n m_n1 <- f_n1(v_x, k1, m1)\n m_g1 <- f_g1(ktau, mtau)\n m_g2 <- f_g2(ktau, mtau)\n m_sigmatau <- f_sigmatau(v_dim[1])\n m_sigma <- f_sigma(v_dim[1], v_dim[2])\n m_s <- f_mats(v_dim[1], v_dim[2], v_smooth)\n # Set up design matrix\n v_taus <- seq(0, 1, length.out = s_ntau)\n m_ndesign <- kronecker(f_ntau(v_taus, ktau, mtau), m_n1)\n\n #Precalculate h1, h0, h0 * Sigma\n m_h1 <- f_smoothh1(m_n1, m_g1, m_g2, m_sigmatau)\n m_h0 <- m_h1$h1\n m_h1 <- m_h1$h1\n v_h0sig <- c(crossprod(m_h0, m_sigma))\n\n #Initialization\n if (is.null(init)) {\n init <- f_init(v_y, v_x, m_sigma, m_n1, v_k0, v_m, m_s)\n }\n til_beta0 <- init\n v_beta0 <- og_beta(til_beta0, v_dim[1], v_dim[2])\n v_grad0 <- 0\n\n #Calculate v_res (most time consuming)\n v_res <- f_smoothres(m_ndesign, v_y, m_sigma, til_beta0, h)\n #Using the Gaussian kernel for h_\\tau\n v_pres <- pnorm(v_res)\n m_htau <- f_smoothhtau(m_ndesign, v_pres, s_ntau)\n\n ##Loss\n s_loss <- f_smoothcrit(v_res, v_pres, v_h0sig, til_beta0, v_beta0,\n s_ntau, h, s_n, m_s)\n\n ##trace\n if (!is.null(trace)) {\n v_taus <- trace\n con_quan <- list(x = v_x, y = v_y, center = s_center, sd = s_sd,\n v_k0 = v_k0, v_m = v_m, m_sigma = m_sigma, n1 = m_n1,\n til_beta = til_beta0)\n sub <- paste(\"Iter:\", iter, \", Error=\", round(s_loss, 3))\n main <- paste(\"Smoothing:\", v_smooth[1:2], \"dim:\", v_dim)\n cq_plot(trace, con_quan, sub = sub, main = main)\n }\n\n #Standard gradient updates\n v_deriv <- f_smoothderiv(til_beta0, v_dim, m_sigma, m_htau, m_h1)\n v_grad1 <- f_smoothgrad(v_beta0, v_deriv / s_n, m_s)\n\n #Calculate new coefficients\n v_beta1 <- v_beta0 - v_grad1\n til_beta1 <- tilde_beta(v_beta1, v_dim[2])\n new_loss <- f_smoothcrit(v_res, v_pres, v_h0sig, til_beta1, v_beta1,\n s_ntau, h, s_n, m_s)\n\n #Repeat\n while (maxit > iter) {\n iter <- iter + 1\n\n #Calculate difference\n v_delta <- v_beta1 - v_beta0\n v_gdiff <- v_grad1 - v_grad0\n\n #bb stepsize\n s_delgrad <- crossprod(v_delta, v_gdiff)\n if (s_delgrad < 0) {\n if (echostep) print(s_delgrad)\n s_step <- 1\n } else {\n step1 <- crossprod(v_delta) / s_delgrad\n step2 <- s_delgrad / crossprod(v_gdiff)\n s_step <- min(step1, step2, 10)\n if (echostep) print(s_step)\n }\n\n #update v_beta0, til_beta0, v_grad0, s_loss\n v_beta0 <- v_beta1\n til_beta0 <- til_beta1\n v_grad0 <- v_grad1\n s_loss <- new_loss\n\n #Calculate v_res (most time consuming)\n v_res <- f_smoothres(m_ndesign, v_y, m_sigma, til_beta0, h)\n #Using the Gaussian kernel for h_\\tau\n v_pres <- pnorm(v_res)\n m_htau <- f_smoothhtau(m_ndesign, v_pres, s_ntau)\n\n #Calculate gradient\n v_deriv <- f_smoothderiv(til_beta0, v_dim, m_sigma, m_htau, m_h1)\n v_grad1 <- f_smoothgrad(v_beta0, v_deriv / s_n, m_s)\n\n #Update v_beta1, til_beta1\n v_beta1 <- v_beta0 - s_step * v_grad1\n til_beta1 <- tilde_beta(v_beta1, v_dim[2])\n new_loss <- f_smoothcrit(v_res, v_pres, v_h0sig, til_beta1, v_beta1,\n s_ntau, h, s_n, m_s)\n\n ##trace\n if (!is.null(trace)) {\n con_quan$til_beta <- til_beta1\n sub <- paste(\"Iter:\", iter, \", Error=\", round(new_loss, 3))\n cq_plot(trace, con_quan, sub = sub, main = main)\n }\n\n #Check stop criterion\n if (is.null(eta)) {\n converge <- (abs(new_loss - s_loss) <= eps * abs(s_loss))\n } else {\n converge <- (norm_2(v_grad1) / norm_2(v_beta0) <= eta)\n }\n\n #If gradient size criteria\n if (is.na(converge)) {\n converge <- F\n break\n }\n if (converge) break\n\n }\n\n#-------------------------------------------------------\n##### Output #####\n#-------------------------------------------------------\n out <- list(\n beta = v_beta1,\n til_beta = til_beta1,\n gradient = v_grad1,\n converge = converge,\n grad_size = sum(v_grad1 ^ 2),\n iter = iter,\n loss = s_loss,\n x = v_x,\n y = v_y,\n center = s_center,\n sd = s_sd,\n v_k0 = v_k0,\n v_m = v_m,\n n1 = m_n1,\n m_sigma = m_sigma,\n mat_s = m_s,\n call = function_call)\n\n return(out)\n}\n\n\n\n\n\n#' f_smoothdescent\n#'\n#' The descent algorithm using back tracking line search stepsize\n#'\n#' @param v_y vector of y\n#' @param v_x vector of x\n#' @param v_k0 vector of internal knots for covariates\n#' @param v_m vector of orders for covariates\n#'\n#' @param init initialization for $\\beta$\n#' @param alpha threshold $\\alpha$, within $(0, 0.5)$.\n#' @param bet shrinkage rate $\\bet$, within $(0, 1)$.\n#' @param v_smooth vector of smoothing parameters\n#' @param maxit maximum number of iterations\n#' @param eps stopping critria threshhold\n#' @param eta stopping criteria for gradient size\n#' @param trace indicator whether to plot trace\n#'\n#'\n#' @return the estimated $\\beta$\n#'\nf_smoothdescent <- function(v_y, v_x, v_k0, v_m, init = NULL,\n alpha = 0.3, bet = 0.8, v_smooth = rep(1, 3),\n h=max((((log(length(v_y))+1)/length(v_y))^0.4), 0.05),\n s_ntau = 200,\n maxit = 1e4, eps = 1e-3, eta = NULL, trace = NULL) {\n\n#-------------------------------------------------------\n##### set up #####\n#-------------------------------------------------------\n function_call <- match.call()\n x_ord <- order(v_x)\n v_x <- v_x[x_ord]\n v_y <- v_y[x_ord]\n s_center <- mean(v_x)\n s_sd <- sd(v_x)\n v_x <- c(scale(v_x))\n iter <- 0L\n converge <- F\n s_n <- length(v_y)\n #Define dimensions\n k1 <- v_k0[2]; ktau <- v_k0[1]; m1 <- v_m[2]; mtau <- v_m[1]\n v_dim <- v_k0 + v_m\n #Require\n m_n1 <- f_n1(v_x, k1, m1)\n m_g1 <- f_g1(ktau, mtau)\n m_g2 <- f_g2(ktau, mtau)\n m_sigmatau <- f_sigmatau(v_dim[1])\n m_sigma <- f_sigma(v_dim[1], v_dim[2])\n m_s <- f_mats(v_dim[1], v_dim[2], v_smooth)\n # Set up design matrix\n v_taus <- seq(0, 1, length.out = s_ntau)\n m_ndesign <- kronecker(f_ntau(v_taus, ktau, mtau), m_n1)\n\n #Precalculate h1, h0, h0 * Sigma\n m_h1 <- f_smoothh1(m_n1, m_g1, m_g2, m_sigmatau)\n m_h0 <- m_h1$h1\n m_h1 <- m_h1$h1\n v_h0sig <- c(crossprod(m_h0, m_sigma))\n\n #Initialization\n if (is.null(init)) {\n init <- f_init(v_y, v_x, m_sigma, m_n1, v_k0, v_m, m_s)\n }\n til_beta <- init\n v_beta <- og_beta(til_beta, v_dim[1], v_dim[2])\n v_grad <- 0\n\n #Calculate v_res (most time consuming)\n v_res <- f_smoothres(m_ndesign, v_y, m_sigma, til_beta, h)\n #Using the Gaussian kernel for h_\\tau\n v_pres <- pnorm(v_res)\n m_htau <- f_smoothhtau(m_ndesign, v_pres, s_ntau)\n\n ##Loss\n s_loss <- f_smoothcrit(v_res, v_pres, v_h0sig, til_beta, v_beta,\n s_ntau, h, s_n, m_s)\n\n ##trace\n if (!is.null(trace)) {\n v_taus <- trace\n con_quan <- list(x = v_x, y = v_y, center = s_center, sd = s_sd,\n v_k0 = v_k0, v_m = v_m, m_sigma = m_sigma, n1 = m_n1,\n til_beta = til_beta)\n sub <- paste(\"Iter:\", iter, \", Error=\", round(s_loss, 3))\n main <- paste(\"alpha=\", alpha, \"beta=\", bet,\n \"Smoothing:\", v_smooth[1:2], \"dim:\", v_dim)\n cq_plot(trace, con_quan, sub = sub, main = main)\n }\n\n #Repeat\n while (maxit > iter) {\n iter <- iter + 1\n #Calculate gradient\n v_deriv <- f_smoothderiv(til_beta, v_dim, m_sigma, m_htau, m_h1)\n v_grad <- f_smoothgrad(v_beta, v_deriv / s_n, m_s)\n grad_size <- sum(v_grad ^ 2)\n #backtracking line search\n t <- 1\n repeat{\n new_beta <- v_beta - t * v_grad\n new_tilbeta <- tilde_beta(new_beta, v_dim[2])\n new_loss <- f_smoothcrit(v_res, v_pres, v_h0sig,\n new_tilbeta, new_beta,\n s_ntau, h, s_n, m_s)\n t <- bet * t\n if (new_loss <= (s_loss - alpha * t * grad_size) || t < 1e-4) {\n break\n }\n }\n\n #Update v_beta\n v_beta <- new_beta\n til_beta <- new_tilbeta\n\n #Calculate v_res (most time consuming)\n v_res <- f_smoothres(m_ndesign, v_y, m_sigma, til_beta, h)\n #Using the Gaussian kernel for h_\\tau\n v_pres <- pnorm(v_res)\n m_htau <- f_smoothhtau(m_ndesign, v_pres, s_ntau)\n\n ##trace\n if (!is.null(trace)) {\n con_quan$til_beta <- til_beta\n sub <- paste(\"Iter:\", iter, \", Error=\", round(new_loss, 3))\n cq_plot(trace, con_quan, sub = sub, main = main)\n }\n\n #Check stop criterion\n if (is.null(eta)) {\n converge <- (abs(new_loss - s_loss) <= eps * abs(s_loss))\n } else {\n converge <- (sqrt(grad_size) / norm_2(v_beta) <= eta)\n }\n\n s_loss <- new_loss\n #If gradient size criteria\n if (is.na(converge)) {\n converge <- F\n break\n }\n if (converge) break\n\n }\n\n#-------------------------------------------------------\n##### Output #####\n#-------------------------------------------------------\n out <- list(\n beta = v_beta,\n til_beta = til_beta,\n gradient = v_grad,\n converge = converge,\n grad_size = sum(v_grad ^ 2),\n iter = iter,\n loss = s_loss,\n x = v_x,\n y = v_y,\n center = s_center,\n sd = s_sd,\n v_k0 = v_k0,\n v_m = v_m,\n n1 = m_n1,\n m_sigma = m_sigma,\n mat_s = m_s,\n alpha = alpha,\n bet = bet,\n call = function_call)\n\n return(out)\n}", "meta": {"hexsha": "e9d5d18217e3acca5bacfff90f4e5bc6aad158b4", "size": 17153, "ext": "r", "lang": "R", "max_stars_repo_path": "R/smoothQR.r", "max_stars_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_stars_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/smoothQR.r", "max_issues_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_issues_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/smoothQR.r", "max_forks_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_forks_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 30.4671403197, "max_line_length": 81, "alphanum_fraction": 0.5587360812, "num_tokens": 5573, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8418256551882382, "lm_q2_score": 0.647798211152541, "lm_q1q2_score": 0.5453331535332565}} {"text": "state_prediction <- function(obs,delta,gamma,lls,param_lls,h)\n {\n m <- ncol(gamma)\n n <- length(obs)\n\n p <- function(state, x){\n lls[[state]](x, param_lls[[state]])\n }\n p <- Vectorize(p)\n\n p_mat <- outer(1:m, obs, p)\n log_p_mat <- log(p_mat)\n\n log_delta <- log(delta)\n log_gamma <- log(gamma)\n log_alpha <- forward_logprobabilities(obs, gamma, p, delta)\n\n log_theta <- rep(NA, m)\n k <- max(log_alpha[, n])\n denominator <- k + log(sum(exp(log_alpha[, n] - k)))\n log_theta <- log_alpha[, n] - denominator\n\n log_gamma_power <- matrix(0, nrow = m, ncol = m)\n for(i in 1:h){\n log_gamma_power <- log_gamma + log_gamma_power\n }\n\n log_prob <- rep(NA, m)\n for(i in 1:m){\n temp <- log_gamma_power[,i] + log_theta\n k <- max(temp)\n log_prob[i] <- k + log(sum(exp(temp - k)))\n }\n\n exp(log_prob)\n}\n", "meta": {"hexsha": "8538d91de1947522b7d9a3382feccdebdd477a8b", "size": 821, "ext": "r", "lang": "R", "max_stars_repo_path": "R/state_prediction.r", "max_stars_repo_name": "asbjornholk/HiddenStateModels", "max_stars_repo_head_hexsha": "202ad33437e3f836a731a71f8c91e0ca6171b938", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/state_prediction.r", "max_issues_repo_name": "asbjornholk/HiddenStateModels", "max_issues_repo_head_hexsha": "202ad33437e3f836a731a71f8c91e0ca6171b938", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 5, "max_issues_repo_issues_event_min_datetime": "2021-11-20T14:48:14.000Z", "max_issues_repo_issues_event_max_datetime": "2021-12-09T20:02:44.000Z", "max_forks_repo_path": "R/state_prediction.r", "max_forks_repo_name": "asbjornholk/HiddenStateModels", "max_forks_repo_head_hexsha": "202ad33437e3f836a731a71f8c91e0ca6171b938", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2022-01-03T18:57:02.000Z", "max_forks_repo_forks_event_max_datetime": "2022-01-03T18:57:02.000Z", "avg_line_length": 22.1891891892, "max_line_length": 61, "alphanum_fraction": 0.6053593179, "num_tokens": 265, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8558511469672594, "lm_q2_score": 0.6370307806984444, "lm_q1q2_score": 0.5452035243142124}} {"text": "model_deterministic_simulateR <- function(init_obs, period_start, times, pars, fix_pars){\r\n ode_solveR <- function(stage_pars, fix_pars, old_values) {\r\n ## stage pars\r\n b = stage_pars[1]\r\n r = stage_pars[2]\r\n ## fixed pars\r\n alpha_p = fix_pars[1] \r\n alpha_u = fix_pars[2] \r\n beta_1 = fix_pars[3] \r\n beta_2 = fix_pars[4]\r\n delta_1 = fix_pars[5]\r\n delta_2 = fix_pars[6]\r\n lambda = fix_pars[7]\r\n mu = fix_pars[8]\r\n mu_c = fix_pars[9]\r\n De = fix_pars[10]\r\n Dr = fix_pars[11]\r\n f = fix_pars[12]\r\n N = fix_pars[13]\r\n ## old values\r\n S = old_values[1]\r\n E = old_values[2]\r\n U = old_values[3]\r\n P = old_values[4]\r\n F = old_values[5]\r\n RU = old_values[6]\r\n RR = old_values[7]\r\n DU = old_values[8]\r\n DR = old_values[9] \r\n ## new values\r\n \r\n S_new = S - b * S * (alpha_p * P + alpha_u * U + F) / N + lambda * N - mu * S\r\n E_new = E + b * S * (alpha_p * P + alpha_u * U + F) / N - E / De - mu * E\r\n U_new = U + (1 - r) * E / De - U / (beta_1 * Dr) - delta_1 * mu_c * U - mu * U\r\n P_new = P + r * (1 - f) * E / De - P / Dr - mu_c * P - mu * P\r\n F_new = F + r * f * E / De - F * beta_2 / Dr - mu_c * F / delta_2 - mu * F\r\n RU_new <- RU + U / (beta_1 * Dr) + F * beta_2 / Dr - mu * RU\r\n RR_new <- RR + P / Dr - mu * RR\r\n DU_new <- DU + delta_1 * mu_c * U + mu_c * F / delta_2\r\n DR_new <- DR + mu_c * P \r\n est_P_new_n <- E\r\n est_P_new_prob <- r * (1 - f) / De \r\n est_RD_new_n <- P\r\n est_RD_new_prob_R <- 1 / Dr \r\n est_RD_new_prob_D <- mu_c \r\n \r\n return(c(S_new, E_new, U_new, P_new, F_new, RU_new, RR_new, DU_new, DR_new, \r\n est_P_new_n, est_P_new_prob,\r\n est_RD_new_n, est_RD_new_prob_R, est_RD_new_prob_D))\r\n }\r\n n_period = length(period_start)\r\n ymat = matrix(0, length(times), length(init_obs) + 6)\r\n ymat[, 1] = times\r\n \r\n colnames(ymat) <- c(\"time\", names(init_obs), \"est_p_n\", \"est_p_prob\", \"est_RD_n\", \"est_RD_prob_R\", \"est_RD_prob_D\")\r\n \r\n which.period <- function(i, phase = period_start){ # function to determine which period i falls in\r\n sum(i >= phase)\r\n }\r\n \r\n for(i in 1:length(times)){\r\n stage_pars <- c(b = pars[which.period(i)], r = pars[n_period + which.period(i)])\r\n if(i == 1) {\r\n old_values <- init_obs\r\n } else {\r\n old_values <- ymat[i - 1, 2:10]\r\n }\r\n ymat[i, 2:15] <- ode_solveR(stage_pars = stage_pars, fix_pars = fix_pars, old_values = old_values)\r\n }\r\n \r\n return(ymat)\r\n}", "meta": {"hexsha": "68827b3040c277d80ee3f9b8a2e9f065951c8f1a", "size": 2486, "ext": "r", "lang": "R", "max_stars_repo_path": "model/seirfansy_f=0/model_deterministic_simulateR.r", "max_stars_repo_name": "umich-cphds/covid_india_wave2", "max_stars_repo_head_hexsha": "32124bf9d51908d007d409d9bcc160eaebd2c74f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "model/seirfansy_f=0/model_deterministic_simulateR.r", "max_issues_repo_name": "umich-cphds/covid_india_wave2", "max_issues_repo_head_hexsha": "32124bf9d51908d007d409d9bcc160eaebd2c74f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "model/seirfansy_f=0/model_deterministic_simulateR.r", "max_forks_repo_name": "umich-cphds/covid_india_wave2", "max_forks_repo_head_hexsha": "32124bf9d51908d007d409d9bcc160eaebd2c74f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 34.5277777778, "max_line_length": 118, "alphanum_fraction": 0.5518905873, "num_tokens": 857, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8652240825770432, "lm_q2_score": 0.6297746074044134, "lm_q1q2_score": 0.5448961569218012}} {"text": "# Copyright (C) 2019 Dr. Norbert Bátfai, nbatfai@gmail.com\n#\n# This program is free software: you can redistribute it and/or modify\n# it under the terms of the GNU General Public License as published by\n# the Free Software Foundation, either version 3 of the License, or\n# (at your option) any later version.\n#\n# This program is distributed in the hope that it will be useful,\n# but WITHOUT ANY WARRANTY; without even the implied warranty of\n# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n# GNU General Public License for more details.\n#\n# You should have received a copy of the GNU General Public License\n# along with this program. If not, see \n\nlibrary(matlab)\n\nstp <- function(x){\n\n primes = primes(x)\n diff = primes[2:length(primes)]-primes[1:length(primes)-1]\n idx = which(diff==2)\n t1primes = primes[idx]\n t2primes = primes[idx]+2\n rt1plust2 = 1/t1primes+1/t2primes\n return(sum(rt1plust2))\n}\n\nx=seq(13, 1000000, by=10000)\ny=sapply(x, FUN = stp)\nplot(x,y,type=\"b\")\n", "meta": {"hexsha": "461c2c4ff2aaa356a9d9476c830dc0c5d1aa2161", "size": 1062, "ext": "r", "lang": "R", "max_stars_repo_path": "Prog2/bhax/attention_raising/Primek_R/stp.r", "max_stars_repo_name": "mesterakos963/Prog2", "max_stars_repo_head_hexsha": "edc3e49567d29211510aa9cd3f4bf639c49c1ea6", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Prog2/bhax/attention_raising/Primek_R/stp.r", "max_issues_repo_name": "mesterakos963/Prog2", "max_issues_repo_head_hexsha": "edc3e49567d29211510aa9cd3f4bf639c49c1ea6", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Prog2/bhax/attention_raising/Primek_R/stp.r", "max_forks_repo_name": "mesterakos963/Prog2", "max_forks_repo_head_hexsha": "edc3e49567d29211510aa9cd3f4bf639c49c1ea6", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 33.1875, "max_line_length": 72, "alphanum_fraction": 0.7052730697, "num_tokens": 304, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.849971175657575, "lm_q2_score": 0.640635854839898, "lm_q1q2_score": 0.5445220107066637}} {"text": "\n#' @title Fitting bivariate MGL copula models for two continuous data\n#' @description \\code{MGL.mle} is used to fit bivariate copula regression models via maximum likelihood (ML) method for two continuous variables.\n#' @param U two-dimensional matrix with values in \\eqn{[0,1]}.\n#' @param copula copula 'MGL', 'MGL180', \"MGL-EV\", \"MGL-EV180\", \"MGB2\", \"Normal\" , \"t\".\n#' @param hessian Logical. Should a numerically differentiated Hessian matrix be returned?\n#' @param initpar Initial values for the parameters to be optimized over.\n#' @param ... additional arguments, see \\code{\\link[stats]{nlm}} for more details.\n#' @importFrom stats nlm\n#' @md\n#' @details\n#' The estimation method is performed via \\code{\\link[stats]{nlm}} function.\n#'\n#' copula:\n#' * \"MGB2\" is multivariate GB2.\n#' * \"Normal\" and \"t\" denote the Gaussian copula and Student-t copula respectively.\n#' * \"MGL\" and \"MGL-EV\" denote the MGL and MGL-EV copula respectively.\n#' * \"MGL180\" and \"MGL-EV180\" denote the survival MGL and survival MGL-EV copula respectively.\n#' * \"Gumbel\" is Gumbel copula.\n#'\n#' @return A list containing the following components:\n#' * loglike: the value of the estimated maximum of the loglikelihood function.\n#' * copula: the name of the fitted copula. \"MGL180\" and \"MGL-EV180\" denote the survival MGL and MGL-EV copula respectively.\n#' * estimates: the point at which the maximum value of the loglikelihood is obtained.\n#' * se: the standard errors of the estimators.\n#' * AIC, BIC: the goodness fit of the regression models.\n#' * hessian: the hessian at the estimated maximum of the loglikelihood (if requested).\n#'\n#' @references Zhang, F. Z. . \"A generalized beta copula with applications in modeling multivariate long-tailed data.\" Insurance: Mathematics and Economics (2011).\n#'\n#' @examples\n#' library(rMGLReg)\n#' Usim <- rcMGL.bivar(n = 500, pars = 0.5)\n#' m.MGL <- MGL.mle(Usim,\n#' copula = \"MGL\",\n#' initpar = c(2))\n#' # estimation results\n#' m.MGL\n#' @export\n#'\nMGL.mle <- function(U, copula = c(\n \"MGL\", \"MGL180\", \"MGL-EV\",\n \"MGL-EV180\",\n \"Gumbel\",\n \"Normal\", \"MGB2\", \"t\"\n ),\n hessian = TRUE,\n initpar, ...) {\n dnormcop <- function(U, param) {\n as.numeric(fCopulae::dellipticalCopula(U, rho = param[1], type = \"norm\"))\n } # normal copula\n\n dtcop <- function(U, param) {\n as.numeric(fCopulae::dellipticalCopula(U,\n rho = param[1], type = \"t\",\n param = param[2]\n ))\n } # t copula\n dgumcop <- function(U, param) {\n as.numeric(fCopulae::devCopula(U, type = \"gumbel\", param = param[1]))\n } # Bivariate Extreme\n\n\n dMGL <- function(U, param) {\n as.numeric(dcMGL.bivar(u1 = U[, 1], u2 = U[, 2], pars = param[1]))\n }\n\n dMGL180 <- function(U, param) {\n as.numeric(dcMGL180.bivar(u1 = U[, 1], u2 = U[, 2], pars = param[1]))\n }\n\n dMGB2 <- function(U, param) {\n as.numeric(dcMGB2.bivar(u1 = U[, 1], u2 = U[, 2], pars1 = param[1], pars2 = param[2], pars3 = param[3]))\n }\n\n dMGLEV <- function(U, param) {\n as.numeric(dcMGLEV.bivar(u1 = U[, 1], u2 = U[, 2], param = param[1])) # Bivariate Extreme\n }\n\n dMGLEV180 <- function(U, param) {\n as.numeric(dcMGLEV180.bivar(u1 = U[, 1], u2 = U[, 2], param = param[1])) # Bivariate Extreme\n }\n\n argnorm <- list(length = 1, lower = 0, upper = 1, name = \"Gaussian\")\n argt <- list(length = 2, lower = c(0, 0), upper = c(1, 100), name = \"Student\")\n arggum <- list(length = 1, lower = 1, upper = 50, name = \"Gumbel\")\n argMGB2 <- list(\n length = 3, lower = c(0, 0, 0), upper = c(50, 50, 50),\n name = \"MGB2\"\n )\n argMG180 <- list(length = 1, lower = 0, upper = 10, name = \"MGL180\")\n argMGLEV180 <- list(length = 1, lower = 0, upper = 10, name = \"MGL-EV180\")\n argMG <- list(length = 1, lower = 0, upper = 10, name = \"MGL\")\n argMGLEV <- list(length = 1, lower = 0, upper = 10, name = \"MGL-EV\")\n if (copula == \"MGL\") {\n dcop <- dMGL\n arg.cop <- argMG\n } else if (copula == \"MGL180\") {\n dcop <- dMGL180\n arg.cop <- argMG180\n } else if (copula == \"MGL-EV\") {\n dcop <- dMGLEV\n arg.cop <- argMGLEV\n } else if (copula == \"MGL-EV180\") {\n dcop <- dMGLEV180\n arg.cop <- argMGLEV180\n } else if (copula == \"Gumbel\") {\n dcop <- dgumcop\n arg.cop <- arggum\n } else if (copula == \"Normal\") {\n dcop <- dnormcop\n arg.cop <- argnorm\n } else if (copula == \"t\") {\n dcop <- dtcop\n arg.cop <- argt\n } else if (copula == \"MGB2\") {\n dcop <- dMGB2\n arg.cop <- argMGB2\n }\n\n # loglike.copula <- function(U, initpar, ...){\n copLogL <- function(x) {\n if (all(arg.cop$lower < x && arg.cop$upper > x)) {\n logL <- log(dcop(U, param = x))\n # res <- -sum((logL))\n remove.naninf <- function(z) z[!is.nan(z) & is.finite(z)]\n res <- -sum(remove.naninf(logL))\n } else {\n res <- 10000000000000\n }\n return(res)\n }\n # }\n\n resopt <- nlm(\n f = copLogL,\n p = initpar,\n hessian = hessian, ...\n )\n\n\n # results\n if (hessian == TRUE){\n out <- list(\n loglike = -resopt$minimum,\n copula = list(name = arg.cop$name),\n estimates = resopt$estimate,\n se = sqrt(diag(solve(resopt$hessian))),\n hessian = -resopt$hessian,\n AIC = 2 * length(resopt$estimate) + 2 * resopt$minimum,\n BIC = log(nrow(U)) * length(resopt$estimate) + 2 * resopt$minimum\n )\n } else {\n out <- list(\n loglike = -resopt$minimum,\n copula = list(name = arg.cop$name),\n estimates = resopt$estimate,\n AIC = 2 * length(resopt$estimate) + 2 * resopt$minimum,\n BIC = log(nrow(U)) * length(resopt$estimate) + 2 * resopt$minimum\n )\n }\n out\n}\n", "meta": {"hexsha": "d36f006261a1271e8c0d82ae2445127d11e97107", "size": 5682, "ext": "r", "lang": "R", "max_stars_repo_path": "R/MGL-mle.r", "max_stars_repo_name": "lizhengxiao/rMGLReg", "max_stars_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/MGL-mle.r", "max_issues_repo_name": "lizhengxiao/rMGLReg", "max_issues_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2021-08-10T13:04:07.000Z", "max_issues_repo_issues_event_max_datetime": "2021-11-23T11:13:27.000Z", "max_forks_repo_path": "R/MGL-mle.r", "max_forks_repo_name": "lizhengxiao/rMGLReg", "max_forks_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 34.6463414634, "max_line_length": 163, "alphanum_fraction": 0.5934530095, "num_tokens": 1905, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.84997116805678, "lm_q2_score": 0.6406358479787609, "lm_q1q2_score": 0.5445220000055532}} {"text": "#' Susceptibility and Infection\n#'\n#' Calculates infectiousness and susceptibility for each node in the graph\n#' @templateVar dynamic TRUE\n#' @templateVar toa TRUE\n#' @templateVar valued TRUE\n#' @template graph_template\n#' @param t0 Integer scalar. See \\code{\\link{toa_mat}}.\n#' @param normalize Logical. Whether or not to normalize the outcome\n#' @param K Integer scalar. Number of time periods to consider\n#' @param r Numeric scalar. Discount rate used when \\code{expdiscount=TRUE}\n#' @param expdiscount Logical scalar. When TRUE, exponential discount rate is used (see details).\n#' @param outgoing Logical scalar. When \\code{TRUE}, computed using outgoing ties.\n#' @family statistics\n#' @aliases susceptibility\n#' @keywords univar\n#' @seealso The user can visualize the distribution of both statistics\n#' by using the function \\code{\\link{plot_infectsuscep}}\n#' @details\n#'\n#' Normalization, \\code{normalize=TRUE}, is applied by dividing the\n#' resulting number from the infectiousness/susceptibility stat\n#' by the number of individuals who adopted the innovation at\n#' time \\eqn{t}.\n#'\n#' Given that node \\eqn{i} adopted the innovation in time \\eqn{t}, its\n#' Susceptibility is calculated as follows\n#'\n#' \\deqn{S_i = \\frac{%\n#' \\sum_{k=1}^K\\sum_{j=1}^n x_{ij(t-k+1)}z_{j(t-k)}\\times \\frac{1}{w_k}}{%\n#' \\sum_{k=1}^K\\sum_{j=1}^n x_{ij(t-k+1)}z_{j(1\\leq t \\leq t-k)} \\times \\frac{1}{w_k} }\\qquad \\mbox{for }i,j=1,\\dots,n\\quad i\\neq j}{%\n#' S(i) = [\\sum_k \\sum_j (x(ij,t-k+1) * z(j,t-k))/w(k)]/[\\sum_k \\sum_j (x(ij,t-k+1) * z(j, 1<=t<=t-k))/w(k)] for j != i}\n#'\n#' where \\eqn{x_{ij(t-k+1)}}{x(ij,t-k+1)} is 1 whenever there's a link from \\eqn{i}\n#' to \\eqn{j} at time \\eqn{t-k+1}, \\eqn{z_{j(t-k)}}{z(j,t-k)}\n#' is 1 whenever individual \\eqn{j} adopted the innovation at time \\eqn{t-k},\n#' \\eqn{z_{j(1\\leq t \\leq t-k)}}{z(j, 1<=t<=t-k)} is 1 whenever\n#' \\eqn{j} had adopted the innovation up to \\eqn{t-k}, and \\eqn{w_k}{w(k)} is\n#' the discount rate used (see below).\n#'\n#' Similarly, infectiousness is calculated as follows\n#'\n#' \\deqn{I_i = \\frac{%\n#' \\sum_{k=1}^K \\sum_{j=1}^n x_{ji(t+k-1)}z_{j(t+k)}\\times \\frac{1}{w_k}}{%\n#' \\sum_{k=1}^K \\sum_{j=1}^n x_{ji(t+k-1)}z_{j(t+k\\leq t \\leq T)}\\times \\frac{1}{w_k} }\\qquad \\mbox{for }i,j=1,\\dots,n\\quad i\\neq j}{%\n#' I(i) = [\\sum_k \\sum_j (x(ji,t) * z(j,t+1))/w(k)]/[\\sum_k \\sum_j (x(ji,t) * z(j, t+1<=t<=T))/w(k)] for j != i}\n#'\n#' It is worth noticing that, as we can see in the formulas, while susceptibility\n#' is from alter to ego, infection is from ego to alter.\n#'\n#' When \\code{outgoing=FALSE} the algorithms are based on incoming edges, this is\n#' the adjacency matrices are transposed swapping the indexes \\eqn{(i,j)} by\n#' \\eqn{(j,i)}. This can be useful for some users.\n#'\n#' Finally, by default both are normalized by the number of individuals who\n#' adopted the innovation in time \\eqn{t-k}. Thus, the resulting formulas,\n#' when \\code{normalize=TRUE}, can be rewritten as\n#'\n#' \\deqn{%\n#' S_i' = \\frac{S_i}{\\sum_{k=1}^K\\sum_{j=1}^nz_{j(t-k)}\\times \\frac{1}{w_k}} %\n#' \\qquad I_i' = \\frac{I_i}{\\sum_{k=1}^K\\sum_{j=1}^nz_{j(t-k)} \\times\\frac{1}{w_k}}}{%\n#' S(i)' = S(i)/[\\sum_k \\sum_j z(j,t-k)/w(k)]\n#'\n#' I(i)' = I(i)/[\\sum_k \\sum_j z(j,t-k)/w(k)]}\n#'\n#' For more details on these measurements, please refer to the vignette titled\n#' \\emph{Time Discounted Infection and Susceptibility}.\n#'\n#' @section Discount rate:\n#'\n#' Discount rate, \\eqn{w_k}{w(k)} in the formulas above, can be either exponential\n#' or linear. When \\code{expdiscount=TRUE}, \\eqn{w_k = (1 + r)^{k-1}}{w(k) = (1+r)^(k-1)}, otherwise\n#' it will be \\eqn{w_k = k}{w(k)=k}.\n#'\n#' Note that when \\eqn{K=1}, the above formulas are equal to the ones presented\n#' in Valente et al. (2015).\n#'\n#' @references\n#' Thomas W. Valente, Stephanie R. Dyal, Kar-Hai Chu, Heather Wipfli, Kayo\n#' Fujimoto Diffusion of innovations theory applied to global tobacco control\n#' treaty ratification, Social Science & Medicine, Volume 145, November 2015,\n#' Pages 89-97, ISSN 0277-9536\n#' \\doi{10.1016/j.socscimed.2015.10.001}\n#'\n#' Myers, D. J. (2000). The Diffusion of Collective Violence: Infectiousness,\n#' Susceptibility, and Mass Media Networks. American Journal of Sociology, 106(1),\n#' 173–208. \\doi{10.1086/303110}\n#'\n#' @examples\n#'\n#' # Creating a random dynamic graph\n#' set.seed(943)\n#' graph <- rgraph_er(n=100, t=10)\n#' toa <- sample.int(10, 100, TRUE)\n#'\n#' # Computing infection and susceptibility (K=1)\n#' infection(graph, toa)\n#' susceptibility(graph, toa)\n#'\n#' # Now with K=4\n#' infection(graph, toa, K=4)\n#' susceptibility(graph, toa, K=4)\n#'\n#' @export\n#' @return A numeric column vector (matrix) of size \\eqn{n} with either infection/susceptibility rates.\n#' @author George G. Vega Yon\ninfection <- function(graph, toa, t0=NULL,\n normalize=TRUE, K=1L, r=0.5, expdiscount=FALSE,\n valued = getOption(\"diffnet.valued\", FALSE),\n outgoing = getOption(\"diffnet.outgoing\", TRUE)) {\n\n # Checking the times argument\n if (missing(toa))\n if (!inherits(graph, \"diffnet\")) {\n stop(\"-toa- should be provided when -graph- is not of class 'diffnet'\")\n } else {\n toa <- graph$toa\n t0 <- min(graph$meta$pers)\n }\n\n # Checking baseline time\n if (!length(t0)) t0 <- min(toa, na.rm=TRUE)\n\n switch (class(graph),\n array = infection.array(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing),\n list = infection.list(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing),\n diffnet = infection.list(graph$graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing),\n stopifnot_graph(graph)\n )\n}\n\n# @rdname infection\n# @export\ninfection.array <- function(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing) {\n toa <- toa - t0 + 1L\n t <- dim(graph)[3]\n n <- nrow(graph)\n ngraph <- vector(\"list\", t)\n\n for(i in 1:t)\n ngraph[[i]] <- methods::as(graph[,,i], \"dgCMatrix\")\n\n out <- infection_cpp(ngraph, toa, normalize, K, r, expdiscount, n, valued, outgoing)\n\n # Naming\n rn <- rownames(graph)\n if (!length(rn)) rn <- 1:n\n structure(out, dimnames=list(rn, \"infection\"), dim=c(n,1))\n}\n\n# @rdname infection\n# @export\ninfection.list <- function(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing) {\n t <- length(graph)\n n <- nrow(graph[[1]])\n toa <- toa - t0 + 1L\n\n out <- infection_cpp(graph, toa, normalize, K, r, expdiscount, n, valued, outgoing)\n\n # Naming\n rn <- rownames(graph[[1]])\n if (!length(rn)) rn <- 1:n\n\n structure(out, dimnames=list(rn, \"infection\"), dim=c(n,1))\n}\n\n#' @rdname infection\n#' @export\nsusceptibility <- function(graph, toa, t0=NULL, normalize=TRUE, K=1L, r=0.5,\n expdiscount=FALSE,\n valued=getOption(\"diffnet.valued\",FALSE),\n outgoing=getOption(\"diffnet.outgoing\",TRUE)) {\n # Checking the toa argument\n if (missing(toa))\n if (!inherits(graph, \"diffnet\")) {\n stop(\"-toa- should be provided when -graph- is not of class 'diffnet'\")\n } else {\n toa <- graph$toa\n t0 <- min(graph$meta$pers)\n }\n\n # Checking baseline time\n if (!length(t0)) t0 <- min(toa, na.rm=TRUE)\n\n cls <- class(graph)\n\n if (\"array\" %in% cls) {\n susceptibility.array(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing)\n } else if (\"list\" %in% cls) {\n susceptibility.list(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing)\n } else if (\"diffnet\" %in% cls) {\n susceptibility.list(graph$graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing)\n } else\n stopifnot_graph(graph)\n\n}\n\n# @rdname infection\n# @export\nsusceptibility.list <- function(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing) {\n t <- length(graph)\n n <- nrow(graph[[1]])\n toa <- toa - t0 + 1L\n\n out <- susceptibility_cpp(graph, toa, normalize, K, r, expdiscount, n, valued, outgoing)\n\n # Naming\n rn <- rownames(graph[[1]])\n if (!length(rn)) rn <- 1:n\n\n structure(out, dimnames=list(rn, \"susceptibility\"), dim=c(n,1))\n\n}\n\n# @rdname infection\n# @export\nsusceptibility.array <- function(graph, toa, t0, normalize, K, r, expdiscount, valued, outgoing) {\n toa <- toa - t0 + 1L\n t <- dim(graph)[3]\n n <- nrow(graph)\n ngraph <- vector(\"list\", t)\n\n for(i in 1:t)\n ngraph[[i]] <- methods::as(graph[,,i], \"dgCMatrix\")\n\n out <- susceptibility_cpp(ngraph, toa, normalize, K, r, expdiscount, n, valued, outgoing)\n\n # Naming\n rn <- rownames(graph)\n if (!length(rn)) rn <- 1:n\n structure(out, dimnames=list(rn, \"susceptibility\"), dim=c(n,1))\n}\n", "meta": {"hexsha": "35258b1af61484273220342a594833cac345279a", "size": 8560, "ext": "r", "lang": "R", "max_stars_repo_path": "R/infect_suscept.r", "max_stars_repo_name": "USCCANA/netdiffuseR", "max_stars_repo_head_hexsha": "4cda66a4381c2df3ee2d738f6c97ac0b2916a5c4", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 69, "max_stars_repo_stars_event_min_datetime": "2015-12-15T02:49:46.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-08T02:48:37.000Z", "max_issues_repo_path": "R/infect_suscept.r", "max_issues_repo_name": "USCCANA/netdiffuseR", "max_issues_repo_head_hexsha": "4cda66a4381c2df3ee2d738f6c97ac0b2916a5c4", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 30, "max_issues_repo_issues_event_min_datetime": "2015-12-17T03:43:07.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-21T18:50:22.000Z", "max_forks_repo_path": "R/infect_suscept.r", "max_forks_repo_name": "USCCANA/netdiffuseR", "max_forks_repo_head_hexsha": "4cda66a4381c2df3ee2d738f6c97ac0b2916a5c4", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 19, "max_forks_repo_forks_event_min_datetime": "2015-12-28T21:47:05.000Z", "max_forks_repo_forks_event_max_datetime": "2021-11-22T19:48:08.000Z", "avg_line_length": 36.7381974249, "max_line_length": 134, "alphanum_fraction": 0.6420560748, "num_tokens": 2826, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8267117940706733, "lm_q2_score": 0.658417500561683, "lm_q1q2_score": 0.5443215131368775}} {"text": "# 4. faza: Analiza podatkov\n\n#linearni model s intervalom zaupanja\n\nmodel1 <- lm(neto_izvoz ~ leto , data = pdf1)\nprihodnost <- data.frame(leto = seq(2020,2024))\n\nnapoved1 <- prihodnost %>% mutate(neto_izvoz= predict(model1, .))\n\n\ntabela.napoved1 <- bind_rows( (pdf1)[c(1,4)],napoved1)\n\n\n\n\ngraf.napoved1 <- ggplot(tabela.napoved1, aes(x = leto, y = neto_izvoz)) + \n geom_point() + scale_x_continuous(name = \"Leto\", breaks = seq(1950,2024,2)) + ylab(\"Neto izvoz\") +\n ggtitle(\"Napoved gibanja neto izvoza-linearna\") +\n geom_smooth(method = 'lm', formula = y ~ x,col=\"red\") +\n theme(axis.text.x=element_text(angle=90, vjust=0.5, hjust=1),\n panel.grid.minor = element_blank())+\n scale_y_continuous(labels=comma)+\n geom_point(data=napoved1, aes(x=leto, y=neto_izvoz), fill='yellow',shape=21, size=2)\n \n \n\n#linearni model uvoz izvoz\n\nmodel2 <- lm(izvoz ~ uvoz , data = pdf1)\nnovi.uvozi <- data.frame(uvoz = c(500000,1000000,1200000))\n\nnapoved2 <- novi.uvozi %>% mutate(izvoz= predict(model2, .))\n\ntabela.napoved2 <- bind_rows( (pdf1)[c(2,3)],napoved2)\n\ngraf.napoved2 <- ggplot(tabela.napoved2, aes(x = uvoz, y = izvoz)) + \n geom_point() + scale_x_continuous(name = \"Uvoz\",labels = scales::comma, breaks = seq(0,1300000,100000)) + ylab(\"Izvoz\") +\n ggtitle(\"Napoved gibanja izvoza glede na uvoz\") +\n geom_smooth(method = 'lm', formula = y ~ x,col=\"red\") +\n theme(axis.text.x=element_text(angle=90, vjust=0.5, hjust=1),\n panel.grid.minor = element_blank())+\n scale_y_continuous(labels=comma)+\n geom_point(data=napoved2, aes(x=uvoz, y=izvoz), fill='yellow',shape=24, size=3.5)\n", "meta": {"hexsha": "74c69806ee98fee7b153d232b573bfbd23aebcf0", "size": 1596, "ext": "r", "lang": "R", "max_stars_repo_path": "analiza/analiza.r", "max_stars_repo_name": "mateabt/APPR-2020-21", "max_stars_repo_head_hexsha": "873dcb3a5cebdf6730f597273cfc6a1693b2f3b2", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "analiza/analiza.r", "max_issues_repo_name": "mateabt/APPR-2020-21", "max_issues_repo_head_hexsha": "873dcb3a5cebdf6730f597273cfc6a1693b2f3b2", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2020-12-23T13:37:49.000Z", "max_issues_repo_issues_event_max_datetime": "2021-01-17T22:11:27.000Z", "max_forks_repo_path": "analiza/analiza.r", "max_forks_repo_name": "mateabt/APPR-2020-21", "max_forks_repo_head_hexsha": "873dcb3a5cebdf6730f597273cfc6a1693b2f3b2", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.2727272727, "max_line_length": 123, "alphanum_fraction": 0.6829573935, "num_tokens": 586, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8267117983401362, "lm_q2_score": 0.658417487156366, "lm_q1q2_score": 0.5443215048656329}} {"text": "dt <- 0.01\r\nNstep <- 10000\r\nk1 <- 0.1\r\nlambda1 <- 1.0\r\nlambda2 <- 0.5\r\nk2 <- 0.2\r\nNcell <- 600\r\nF0 <- 0.1\r\np0 <- 0.7\r\n## F0 <- 0.8\r\n## p0 <- 0.8\r\nec <- 1.6\r\nsig1 <- 0\r\nsig2 <- 0\r\nalpha <- 0.5\r\nsig <- 0.0\r\n\r\nthetabarList <- NULL\r\nfor (nsample in seq(1))\r\n{\r\ne <- seq(1.2, 1.2, length.out=Ncell)\r\ntheta <- runif(Ncell, -pi/2, pi/2)\r\n\r\noutput <- NULL\r\n\r\nthetabar <- 0\r\nfor ( i in seq(Nstep) )\r\n{\r\n if ( i %% 20 == 0 )\r\n {\r\n n <- cbind(cos(theta), sin(theta))\r\n Q <- cbind(n[,1]*n[,1]-0.5, n[,1]*n[,2], n[,2]*n[,1], n[,2]*n[,2]-0.5)\r\n M <- matrix(colMeans(Q), nrow=2, ncol=2, byrow=TRUE)\r\n eigenvec <- eigen(M)$vectors[,1]\r\n thetabar <- atan(abs(eigenvec[2])/pmax(0.000001, abs(eigenvec[1])))\r\n }\r\n F <- max(0, F0 - alpha * sig/(1+abs(sig)))\r\n S <- mean(cos(2.0*theta))\r\n e <- e + dt*(F*cos(theta) - lambda1*(e-1)) + k1*sqrt(dt)*rnorm(Ncell, mean=0, sd=1)\r\n e <- pmax(e, 1)\r\n theta <- theta - dt*lambda2*(e-1)*(theta - thetabar) + k2*sqrt(dt)*rnorm(Ncell, mean=0, sd=1)\r\n ##theta <- theta - dt*S*lambda2*(theta-thetabar) + k2*sqrt(dt)*rnorm(Ncell, mean=0, sd=1)\r\n ##theta <- atan(abs(sin(theta))/pmax(0.000001, abs(cos(theta))))\r\n #cat(i, S, '\\n')\r\n\r\n if ( i%%10 == 0 )\r\n {\r\n theta1 <- theta\r\n theta1 <- atan(abs(sin(theta1))/pmax(0.000001, abs(cos(theta1))))\r\n ecut <- e < ec\r\n theta2 <- theta1 * (1 - ecut) + ecut * runif(Ncell, 0, pi/2)\r\n A1 <- mean(sin(theta1) - cos(theta1))\r\n A2 <- mean(sin(theta2) - cos(theta2))\r\n ES <- mean((1-p0)*cos(theta1) - p0*sin(theta1))\r\n #cat(A1, A2, S, ES, F, sig, thetabar*180/pi, '\\n')\r\n\r\n divtheta <- sample(theta1, 1)\r\n sig1 <- sig1 + cos(divtheta)\r\n sig2 <- sig2 + sin(divtheta)\r\n sig <- ((1-p0)*sig1 - p0*sig2)\r\n\r\n output <- rbind(output, c(sig, F, A1, A2, ES, sig1, sig2, divtheta))\r\n }\r\n}\r\n\r\ncat(A1, A2, S, ES, F, sig, thetabar*180/pi, '\\n')\r\nthetabarList <- c(thetabarList, thetabar*180/pi)\r\n}\r\n##png(\"output1.png\")\r\npdf(\"output1.pdf\")\r\npar(mfrow=c(2, 1))\r\npar(mar=c(4,6,4,4))\r\nplot(output[200:800,1], type='o', col='blue', ylim=c(-6, 6),\r\n xlab=\"Number of divisions\",cex.lab=1.5, cex.axis=1.5,\r\n ylab=\"Signal\")\r\nabline(h=0, lty=2)\r\n##lines(output[,2], col='green', lty=1)\r\n##lines(output[,3], col='blue')\r\nplot(output[200:800,5], col='red')\r\ndev.off()\r\n\r\nsig0 <- mean(output[,1])\r\nF0rec = F0 - alpha * sig0/(1+abs(sig0))\r\n## theta1 <- theta\r\n## theta1 <- atan(abs(sin(theta1))/pmax(0.000001, abs(cos(theta1))))\r\n## ecut <- e < ec\r\n## theta2 <- theta1 * (1 - ecut) + ecut * runif(Ncell, 0, pi/2)\r\n## A1 <- mean(sin(theta1) - cos(theta1))\r\n## A2 <- mean(sin(theta2) - cos(theta2))\r\n## A1List <- c(A1List, A1)\r\n## A2List <- c(A2List, A2)\r\n## cat(A1, A2, S, thetabar*180/pi, '\\n')\r\n\r\n", "meta": {"hexsha": "ac1164e81ababde0d12236135cfa19ee8968b299", "size": 2797, "ext": "r", "lang": "R", "max_stars_repo_path": "MeanFieldModel/regu1.r", "max_stars_repo_name": "hydrays/CellModel", "max_stars_repo_head_hexsha": "c5aef3494869d7438a3b3d8615afb2271f2202d6", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "MeanFieldModel/regu1.r", "max_issues_repo_name": "hydrays/CellModel", "max_issues_repo_head_hexsha": "c5aef3494869d7438a3b3d8615afb2271f2202d6", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "MeanFieldModel/regu1.r", "max_forks_repo_name": "hydrays/CellModel", "max_forks_repo_head_hexsha": "c5aef3494869d7438a3b3d8615afb2271f2202d6", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 29.7553191489, "max_line_length": 98, "alphanum_fraction": 0.5230604219, "num_tokens": 1106, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8221891392358014, "lm_q2_score": 0.6619228825191872, "lm_q1q2_score": 0.544225805018931}} {"text": "#' @title calcCm\n#'\n#' @description Calculates the midship section coefficient (\\code{Cm})\n#' (dimensionless).\n#'\n#' @param shipType Ship type (vector of strings, see \\code{\\link{calcShipType}}). \n#' Must align with \\code{tankerBulkCarrierShipTypes}, \\code{tugShipTypes}, and \n#' \\code{roroPaxShipTypes} groupings \n#' @param Cbw Waterline block coefficient (vector of numericals, dimensionless) \n#' (see \\code{\\link{calcCbw}})\n#' @param maxDraft Maximum summer load line draft (vector of numericals, m)\n#' @param actualDraft Actual draft (vector of numericals, m)\n#' @param CmEquationType Equation type: \\itemize{\n#' \\item\"kristensen\"\n#' \\item\"benford\"\n#' \\item\"schneekluth\"}\n#' This argument is not vectorized, as it takes only a single string\n#'@param tankerBulkCarrierShipTypes Ship types specified in input \\code{shipTypes}\n#'to be modeled as tankers and bulk carriers (vector of strings)\n#'@param tugShipTypes Ship types specified in input \\code{shipTypes} to be\n#'modeled as tugs (vector of strings)\n#'@param roroPaxShipTypes Ship types specified in input \\code{shipTypes} to be\n#'modeled as RORO and passenger ships (vector of strings)\n#'\n#' @details\n#' The midship section coefficient calculation depends on the ship type, in\n#' addition to the actual draft. Actual draft is typically obtained from sources\n#' such as AIS messages or ship records.\n#'\n#' This function can calculate \\code{Cm} using three different methods:\n#' Kristensen, Benford, and Schneekluth. The Kristensen method requires ship\n#' types to be grouped. Use the \\code{tankerBulkCarrierShipTypes},\n#' \\code{tugShipTypes}, and \\code{roroPaxShipTypes} parameters to provide these\n#' ship type groupings. Any ship types not included in these groupings will be\n#' considered as miscellaneous vessels.\n#'\n#' Use the \\code{CmEquationType} parameter to indicate which method to use:\\itemize{\n#'\n#' \\item \"kristensen\" (see Kristensen 2013 & 2017): \\itemize{\n#' \\item Bulk Carriers and Tankers: \\code{Cm} = 0.995\n#' \\item Passenger Vessels: \\code{Cm} = 0.95\n#' \\item Tugboats: \\code{Cm} = 0.92\n#' \\item Miscellaneous Vessels: \\code{Cm} = 0.98\n#' }\n#'\n#' \\item \"benford\" (see Rakke 2016):\n#' \\deqn{Cm=0.977+0.085*(Cbw-0.6)}\n#'\n#' \\item \"schneekluth\" (see Schneekluth 1998):\n#' \\deqn{Cm=1.006-0.0056*Cbw^-3.65}\n#'\n#' }\n#'\n#' @return \\code{Cm} (vector of numericals, dimensionless)\n#'\n#' @references\n#'Kristensen, H. O. and Lutzen, M. 2013. \"Prediction of Resistance and Propulsion\n#'Power of Ships.\"\n#'\n#'\\href{https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}{Kristensen, H. O.\n#'\"Ship-Desmo-Tool.\" https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}\n#'\n#'Schneekluth, H. and Bertram, V. 1998. \"Ship Design for Efficiency and Economy.\"\n#'2nd ed. Oxford, Boston: Butterworth-Heinemann.\n#'\n#' @seealso \\itemize{\n#' \\item \\code{\\link{calcCbw}}\n#' \\item \\code{\\link{calcShipType}}\n#' }\n#'\n#' @examples\n#' calcCm(c(\"chemical.tanker\",\"container.ship\"),c(0.8,0.75),c(13.6,15.6),c(12.5,14.1),\"kristensen\")\n#' calcCm(c(\"chemical.tanker\",\"container.ship\"),c(0.8,0.75),c(13.6,15.6),c(12.5,14.1),\"kristensen\",\n#' tankerBulkCarrierShipTypes=c(\"other.tanker\"))\n#'\n#' @export\n\n\ncalcCm <-function(shipType,Cbw,maxDraft,actualDraft,CmEquationType,\n tankerBulkCarrierShipTypes=c(\"tanker\",\"chemical.tanker\",\"liquified.gas.tanker\",\"oil.tanker\",\"other.tanker\",\"bulk.carrier\"),\n tugShipTypes=c(\"service.tug\",\"tug\"),\n roroPaxShipTypes=c(\"ferry.pax\",\"ferry.ro.pax\",\"cruise\",\"cruise.ed\",\"yacht\",\"ro.ro\",\"passenger\")\n ){\n\n if(grepl(\"kristensen\",tolower(CmEquationType))==TRUE){\n Cm<- ifelse(#case 1\n shipType %in% tankerBulkCarrierShipTypes\n ,\n 0.995,\n ifelse(#case 2\n shipType %in% tugShipTypes,\n 0.92,\n ifelse(#case 3\n shipType %in% roroPaxShipTypes,\n 0.95,# passenger case\n 0.98 # misc type case\n )#end of case 3\n\n )#end of case 2\n )#end of case 1\n\n #adjust for actual draft\n Cm<-1-maxDraft/actualDraft*(1-Cm)\n\n }else if(grepl(\"benford\",tolower(CmEquationType))==TRUE){\n\n Cm<-0.977+0.085*(Cbw-0.6)\n\n }else if(grepl(\"schneekluth\",tolower(CmEquationType))==TRUE){\n\n Cm<-1.006-0.0056*Cbw^-3.56\n\n }\n return(Cm)\n}\n", "meta": {"hexsha": "35d4fde709521095749c4f3e144aa8e8ce6472de", "size": 4283, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcCm.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcCm.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcCm.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 36.9224137931, "max_line_length": 141, "alphanum_fraction": 0.6754611254, "num_tokens": 1380, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7606506526772884, "lm_q2_score": 0.7154240079185319, "lm_q1q2_score": 0.5441877385642329}} {"text": "## Function to do Wright-Fisher simulation with selection and a finite sample size\n# using priors on Ne and s\n# and specify initial sample allele frequency, f1samp (compare to wfs.r, which sets a prior on initial true allele freq f1)\n# Sample at THREE time points\n\n# f1samp: the initial sample allele frequency. Initial true allele freq will be sampled with this constraint.\n# smin: the minimum for the selection coefficient uniform prior\n# smax:\n# c1: the size of the first sample (in # chromosomes)\n# c2: the size of the second sample (in # chromosomes)\n# c3: the size of the third sample (in # chromosomes)\n# gen1: the number of generations for sample c2\n# gen2: the number of generations for sample c3\n# ne: a vector of possible Ne values. Simulation picks one to use.\n# h: dominance effect\n\nwfs_byf1samp_3samps <- function(f1samp=0.5, smin=-1, smax=1, c1=58, c2=48, c3=48, gen1=10, gen2=20, ne=100, hmin=0.5, hmax=0.5){ \n\ttol <- .Machine$double.eps^0.5 # tolerane for equality. default tolerance in all.equal\n\n\t# make sure f1samp is possible given initial sample size (c1)\n\t# test whether difference from a possible sample size is less than tol\n\tif(!any(abs(f1samp - ((0:c1)/c1)) < tol)) stop(paste('f1samp=', f1samp, 'not possible given c1=', c1))\n\n\t# choose parameters for this simulation\n\ts <- runif(1, min=smin, max=smax) # selection coefficient\n\th <- runif(1, min=hmin, max=hmax) # dominance coefficient\n\tif(length(ne)>1) thisne <- round(sample(ne, 1)) # ne has to be integer for binomial sampling\n\tif(length(ne)==1) thisne <- round(ne)\n\n\t# choose initial population frequency\n\t# brute force method: keep picking possible f1s until one produces the correct f1samp\n\tthisf1 <- -100\n\twhile(abs(thisf1 - f1samp) >= tol){\n\t\tf1 <- runif(1, min=0, max=1) # pick a starting allele frequency\n\t\tthisf1 <- rbinom(1,c1,f1)/c1 # pick a first sample allele frequency that corresponds\n\t}\n\t\n#\tprint(paste(f1, s, thisne))\n\t\n\tp <- f1 # current allele frequency\n\twaa <- 1+s # relative fitness of genotype AA\n\twab <- 1+s*h\n\twbb <- 1\n\tfor(i in 1:gen2){\n\t\tx <- (waa*p^2 + wab*p*(1-p))/(waa*p^2 + wab*2*p*(1-p) + wbb*(1-p)^2) # probability of sampling allele A, given selection\n\t\tp <- rbinom(1,thisne,x)/thisne\n#\t\tprint(paste(x,p))\n\t\tif(i==gen1){\n\t\t\tf2 <- p # take second sample\n\t\t}\n\t}\n\tf3 <- p # third sample\n\n\tf2samp <- rbinom(1,c2,f2)/c2 # second sample allele frequency\n\tf3samp <- rbinom(1,c3,f3)/c3 # third sample allele frequency\n\n\n\t# return values\n#\tout <- c(ne=thisne, f1=f1, s=s, gen=gen, f2=f2, f1samp=f1samp, f2samp=f2samp, fsdprime=stats[1], fsiprime=stats[2])\n\tout <- c(ne=thisne, f1=f1, s=s, h=h, gen1=gen1, gen2=gen2, f2=f2, f3=f3, f1samp=f1samp, f2samp=f2samp, f3samp=f3samp)\n\treturn(out)\n}", "meta": {"hexsha": "4837dddae44f329e44ef5758eca262e1b95d44ff", "size": 2693, "ext": "r", "lang": "R", "max_stars_repo_path": "scripts/wfs_byf1samp_3samps.r", "max_stars_repo_name": "pinskylab/codEvol", "max_stars_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "scripts/wfs_byf1samp_3samps.r", "max_issues_repo_name": "pinskylab/codEvol", "max_issues_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 5, "max_issues_repo_issues_event_min_datetime": "2020-04-11T11:14:18.000Z", "max_issues_repo_issues_event_max_datetime": "2021-02-21T19:57:31.000Z", "max_forks_repo_path": "scripts/wfs_byf1samp_3samps.r", "max_forks_repo_name": "pinskylab/codEvol", "max_forks_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 43.435483871, "max_line_length": 129, "alphanum_fraction": 0.7010768659, "num_tokens": 884, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8688267830311354, "lm_q2_score": 0.6261241772283034, "lm_q1q2_score": 0.5439934546792833}} {"text": "\"\nС клавиатуры ввести целое число N. Рассчитать соответствующий числу день недели\nмесяц и год, исходя из того, что в каждом месяце 30 дней, 1 задает понедельник января\n2 – вторник января и т.д. Вывести результат в виде строки следующего шаблона:\n«Введенное значение N соответствует среде марта, 22-й год».\n\"\n\n{\n number <- as.integer(readline(\"Введите число -> \"))\n year <- number %/% 360 + 1\n number <- number %% 360\n month <- (number %/% 30) + 1\n day <- number %% 7\n monthwrite <- switch(\n month,\n \"Январь\",\n \"Февраль\",\n \"Март\",\n \"Апрель\",\n \"Май\",\n \"Июнь\",\n \"Июль\",\n \"Август\",\n \"Сентябрь\",\n \"Октябрь\",\n \"Ноябрь\",\n \"Декабрь\"\n )\n daywrite <- switch (\n day,\n \"Понедельник\",\n \"Вторник\",\n \"Среда\",\n \"Четверг\",\n \"Пятница\",\n \"Суббота\",\n \"Воскресенье\"\n )\n \n print(paste(\n \"Введенное значение N соответствует\",\n day ,\n daywrite,\n monthwrite,\n year,\n \"год\"\n ))\n}\n", "meta": {"hexsha": "644f704f2e2870e564da851c406179c4eb9ed20b", "size": 950, "ext": "r", "lang": "R", "max_stars_repo_path": "Course II/R/pract/pract5/task1.r", "max_stars_repo_name": "GeorgiyDemo/FA", "max_stars_repo_head_hexsha": "641a29d088904302f5f2164c9b3e1f1c813849ec", "max_stars_repo_licenses": ["WTFPL"], "max_stars_count": 27, "max_stars_repo_stars_event_min_datetime": "2019-08-18T20:54:27.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-22T02:39:45.000Z", "max_issues_repo_path": "Course II/R/pract/pract5/task1.r", "max_issues_repo_name": "GeorgiyDemo/FA", "max_issues_repo_head_hexsha": "641a29d088904302f5f2164c9b3e1f1c813849ec", "max_issues_repo_licenses": ["WTFPL"], "max_issues_count": 217, "max_issues_repo_issues_event_min_datetime": "2019-09-22T14:43:25.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-30T13:49:18.000Z", "max_forks_repo_path": "Course II/R/pract/pract5/task1.r", "max_forks_repo_name": "GeorgiyDemo/FA", "max_forks_repo_head_hexsha": "641a29d088904302f5f2164c9b3e1f1c813849ec", "max_forks_repo_licenses": ["WTFPL"], "max_forks_count": 42, "max_forks_repo_forks_event_min_datetime": "2019-09-18T11:36:28.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-19T18:43:00.000Z", "avg_line_length": 19.387755102, "max_line_length": 86, "alphanum_fraction": 0.5905263158, "num_tokens": 405, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8006920020959544, "lm_q2_score": 0.679178692681616, "lm_q1q2_score": 0.5438129472241561}} {"text": "## Insang Song (isong@uoregon.edu)\n## Spatially Enhanced and Entropy-Derived MATrix (SpEED-MAT)\nspeedmat <- function(sf,\n mode = \"CDE\", # \"CE\"/\"DE\"/\"CDE\"\n input_vars = c(), # Contiguity Distance Entropy\n bandwidth = NULL, # bandwidth: Currently only supports gaussian\n q_jsd = 0.05, # quantile of j-s divergence\n kneigh = NULL, # k-nearest neighbor for point data\n queen = TRUE, # Queen's contiguity for \"C\" mode for polygon data\n cutoff_dist = NULL,\n cutoff_weightd = 0.001,\n cutoff_weightc = 0.001,\n sup_factor = 0.5){\n if (is.null(bandwidth) & grepl(\"D\", mode))\n stop(\"No bandwidth was entered\")\n if (length(input_vars) == 0)\n stop(\"No input variable were specified\")\n if (!grepl(\"E\", mode))\n stop(\"Please use the methods for standard spatial weight matrices for non-entropy matrices\")\n if (!is.null(bandwidth)) {\n cutoff_dist <- bandwidth\n }\n sfp <- sf[1,]\n\n sf_ent <- sf %>%\n dplyr::select(input_vars) %>%\n st_set_geometry(NULL) %>%\n mutate_at(.vars = vars(input_vars),\n .funs = list(~as.vector(scale(.)))) %>%\n mutate_at(.vars = vars(everything()),\n .funs = list(~.+abs(min(.)))) %>%\n as.matrix %>%\n philentropy::JSD(.)\n sf_ent <- sf_ent * (sf_ent <= quantile(sf_ent, q_jsd))\n sf_ent_ex <- exp(-1 * sf_ent)\n sf_ent_f <- (sf_ent_ex * (sf_ent_ex != 1))\n\n if (grepl(\"C\", mode)){\n if (queen){\n pat <- 'F***T****'\n } else {\n pat <- 'F***1****'\n }\n\n if (any(st_is(sfp, 'POLYGON'), st_is(sfp, 'MULTIPOLYGON'), st_is(sfp, \"POLYGON Z\"), st_is(sfp, \"MULTIPOLYGON Z\")))\n {sf_touch <- st_relate(sf, pattern = pat) %>% #poly2nb(sf, queen = queen) %>%\n as(., 'matrix')#nb2mat\n sf_touch_ <- sf_touch\n } else {\n sf_touch <- st_nn(sf, sf, k = kneigh + 1, sparse = FALSE, returnDist = TRUE) #%>%\n sf_touch_ <- sf_touch\n sf_touch_nn <- sf_touch$nn\n diag(sf_touch_nn) <- FALSE\n sf_touch <- sf_touch_nn\n }\n # 122820\n sf_invdv <- ((1-sup_factor)*sf_touch) + (sup_factor * sf_ent_f)\n }\n if (grepl(\"D\", mode))\n {\n if (any(st_is(sfp, 'POLYGON'), st_is(sfp, 'MULTIPOLYGON'), st_is(sfp, \"POLYGON Z\"), st_is(sfp, \"MULTIPOLYGON Z\"))) {\n sf_interd <- as(st_distance(st_centroid(sf)), 'matrix')\n } else {\n sf_interd <- as(st_distance(sf), 'matrix')\n }\n sf_interd <- sf_interd * (sf_interd <= cutoff_dist)\n sf_invd <- exp(-1 * ((sf_interd^2)/(bandwidth^2)))\n sf_invd <- sf_invd * (sf_invd != 1)\n diag(sf_invd) <- 0\n\n if (grepl(\"^C\", mode)) {\n if (!any(st_is(sfp, 'POLYGON'), st_is(sfp, 'MULTIPOLYGON'), st_is(sfp, \"POLYGON Z\"), st_is(sfp, \"MULTIPOLYGON Z\")))\n {\n sf_touch_ <- sf_touch\n sf_touch_d <- do.call(c, sf_touch_$dist)\n sf_touch_d <- sf_touch_d[sf_touch_d != 0]\n sf_touch_nn[sf_touch_nn == TRUE] <- sf_touch_d\n sf_touch <- sf_touch_nn\n }\n\n sf_invdv <- ((1-sup_factor) * (sf_touch * sf_invd)) + (sup_factor * sf_ent_f)\n sf_invdv <- sf_invdv * (sf_invdv >= cutoff_weightc)\n } else {\n sf_invdv <- ((1-sup_factor)* (sf_invd * (sf_invd >= cutoff_weightd))) + (sup_factor * sf_ent_f)\n sf_invdv <- sf_invdv * (sf_invdv >= cutoff_weightc)\n }\n }\n\n # error check\n checksum <- apply(sf_invdv, 1, sum)\n if (any(is.nan(checksum) | checksum == 0)){\n addr_check <- (is.nan(checksum) | checksum==0)\n diag(sf_invdv)[addr_check] <- 1\n }\n speedmat <- sf_invdv / apply(sf_invdv, 1, sum)\n return(speedmat)\n}\n", "meta": {"hexsha": "7fc0c2b8b7f7a4ac325c1a6b2c76102a3ad12299", "size": 3984, "ext": "r", "lang": "R", "max_stars_repo_path": "speedmat/R/speedmat.r", "max_stars_repo_name": "sigmafelix/speed", "max_stars_repo_head_hexsha": "b80e94b8b491e5862a04fa2a3c86511fcc2aeb5e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "speedmat/R/speedmat.r", "max_issues_repo_name": "sigmafelix/speed", "max_issues_repo_head_hexsha": "b80e94b8b491e5862a04fa2a3c86511fcc2aeb5e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "speedmat/R/speedmat.r", "max_forks_repo_name": "sigmafelix/speed", "max_forks_repo_head_hexsha": "b80e94b8b491e5862a04fa2a3c86511fcc2aeb5e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 40.6530612245, "max_line_length": 127, "alphanum_fraction": 0.5188253012, "num_tokens": 1133, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8947894661025424, "lm_q2_score": 0.6076631698328916, "lm_q1q2_score": 0.5437306033049516}} {"text": "#!/usr/bin/r\n##\n## varSimulation.r: Simulation of first-order vector autoregression data\n##\n## Copyright (C) 2011 - 2015 Lance Bachmeier and Dirk Eddelbuettel\n##\n## This file is part of RcppArmadillo.\n##\n## RcppArmadillo is free software: you can redistribute it and/or modify it\n## under the terms of the GNU General Public License as published by\n## the Free Software Foundation, either version 2 of the License, or\n## (at your option) any later version.\n##\n## RcppArmadillo is distributed in the hope that it will be useful, but\n## WITHOUT ANY WARRANTY; without even the implied warranty of\n## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n## GNU General Public License for more details.\n##\n## You should have received a copy of the GNU General Public License\n## along with RcppArmadillo. If not, see .\n\n\n## load Rcpp to be able to use cppFunction() below\nsuppressMessages(library(Rcpp))\n\n\n## parameter and error terms used throughout\na <- matrix(c(0.5,0.1,0.1,0.5),nrow=2)\ne <- matrix(rnorm(10000),ncol=2)\n\n## Let's start with the R version\nrSim <- function(coeff, errors) {\n simdata <- matrix(0, nrow(errors), ncol(errors))\n for (row in 2:nrow(errors)) {\n simdata[row,] = coeff %*% simdata[(row-1),] + errors[row,]\n }\n return(simdata)\n}\n\nrData <- rSim(a, e) # generated by R\n\n\n## Now let's load the R compiler (requires R 2.13 or later)\nsuppressMessages(require(compiler))\ncompRsim <- cmpfun(rSim)\n\ncompRData <- compRsim(a,e) # generated by R 'compiled'\n\nstopifnot(all.equal(rData, compRData)) # checking results\n\n\n## C++ variant: code passed as a text variable ...\ncode <- '\narma::mat rcppSim(const arma::mat& coeff, const arma::mat& errors) {\n int m = errors.n_rows;\n int n = errors.n_cols;\n arma::mat simdata(m,n);\n simdata.row(0) = arma::zeros(1,n);\n for (int row=1; row ## inicio da aula R ##\n> \n\n> ## UNIDADE 1 ##\n\n> ## OPERACOES MATEMATICAS##\n> 3+3\n[1] 6\n> 3-4\n[1] -1\n> 3*3\n[1] 9\n> 4/2\n[1] 2\n> %% 2\nError: unexpected SPECIAL in \"%%\"\n> \n> ## VARIAVEL ##\n> var <- 3*4\n> var\n[1] 12\n> \n> ## VARIAVEL NAO TEM TIPO PODE SER CHARACTER ##\n> VAR <- \"jONATHAN sARTORI bRUZARROSCO\"\n> VAR\n[1] \"jONATHAN sARTORI bRUZARROSCO\"\n\n> ##SALVAR DADOS NO ENVIRONMENT##\n> idade <- 34\n> nome <- \"Jonathan Sartori Bruzarrosco\"\n> logica <- TRUE\n> salario <- 1000.15\n> var <- \"qualquer coisa\"\n> \n> ## MOSTRAR A KISTA DAS VARIAVEIS SALVAS ##\n> ls()\n[1] \"idade\" \"logica\" \"nome\" \"salario\"\n[5] \"var\" \n> \n> ## REMOVER VARIAVEL SALVA ##\n> rm(var)\n> \n> ## REMOVER A LISTA TODA ##\n> rm(list = ls())\n> \n> \n\n> ## UNIDADE 2 ##\n\n## ADICIONANDO VARIAVEIS ##\n> nome <- \"Jonathan Sartori Bruzarrosco\"\n> idade <- 34\n> salario <- 1000.15\n> logica <- TRUE\n> \n> ## PARA SABER O TIPO DE DADO ##\n> class(idade)\n[1] \"numeric\"\n> class(salario)\n[1] \"numeric\"\n> class(nome)\n[1] \"character\"\n> class(logica)\n[1] \"logical\"\n> ## LOGICAL ACEITA VERDADEIRO OU FALSO ##\n\n> ## character vai estar sempre entre\"\"##\n> \n> ## EXEMPLO ##\n> idade <- \"5\"\n> class(idade)\n[1] \"character\"\n> \n> ## TESTAR O TIPO DO DADO ##\n> is.numeric(idade)\n[1] FALSE\n> is.character(nome)\n[1] TRUE\n> is.logical(logica)\n[1] TRUE\n> \n> ## CONVERTER TIPO ##\n> var <- idade + 5\nError in idade + 5 : non-numeric argument to binary operator\n> as.integer(idAde)\nError: object 'idAde' not found\n> as.integer(idade)\n[1] 5\n> var <- as.integer(idade) + 5\n> var\n[1] 10\n> ## ou seja is verifica e as converte ##\n> ## nem sempre se consegue exemplo ##\n> as.integer(nome)\n[1] NA\nWarning message:\nNAs introduced by coercion \n\n> ## INICIANDO CONDICIONAL ##\n> numero <- 10\n> \n> ##SHIFT + ENTER PARA PASSAR LINHA SEM RODAR CODIGO##\n> \n> if(class(numero) != \"numeric\"){\n+ print(\"não é numerico\")\n+ }else{\n+ if(numero %% 2 == 0){\n+ print(\"o numero é par\")\n+ }else{\n+ print(\"o numero é impar\")\n+ }\n+ }\n[1] \"o numero é par\"\n> \n> numero <- 11\n> if(class(numero) != \"numeric\"){\n+ print(\"não é numerico\")\n+ }else{\n+ if(numero %% 2 == 0){\n+ print(\"o numero é par\")\n+ }else{\n+ print(\"o numero é impar\")\n+ }\n+ }\n[1] \"o numero é impar\"\n> \n> numero <- \"jonathan\"\n> if(class(numero) != \"numeric\"){\n+ print(\"não é numerico\")\n+ }else{\n+ if(numero %% 2 == 0){\n+ print(\"o numero é par\")\n+ }else{\n+ print(\"o numero é impar\")\n+ }\n+ }\n[1] \"não é numerico\"\n\n> ## != diferente ##\n> ## %% 2 == 0 (resto da divisao é igual a 0 ou nao)##\n> ## %% = mod ##\n> \n> \n> ##ESTRUTURA DE REPETIÇÃO##\n> \n> professores <- c(\"Gobbato\", \"Maito\", \"alcides\")\n> \n> ##saber a quantidade de elementos dentro do conjunto##\n> \n> length(professores)\n[1] 3\n> professores[1]\n[1] \"Gobbato\"\n> professores[2]\n[1] \"Maito\"\n> professores[0]\ncharacter(0)\n> ##diferente de outras linguagens R comeca no 1##\n> \n> print(paste(\"o nome do professor é: \",professores[1] ))\n[1] \"o nome do professor é: Gobbato\"\n> \n> for(pos in 1:length(professores)){\n+ print(paste(\"o nome do professor é:\", professores[pos]))\n+ }\n[1] \"o nome do professor é: Gobbato\"\n[1] \"o nome do professor é: Maito\"\n[1] \"o nome do professor é: alcides\"\n\n> posicao <- 1\n\n> while(posicao <= length(professores)){\n+ print(paste(\"o nome do professor é:\", professores[posicao]))\n+ posicao <- posicao + 1 #incremento\n+ }\n[1] \"o nome do professor é: Gobbato\"\n[1] \"o nome do professor é: Maito\"\n[1] \"o nome do professor é: alcides\"\n\n> ## UNIDADE 3 ##\n\n\n", "meta": {"hexsha": "ea9fba6d561057abf0b04460703a0b4c30d640bd", "size": 3496, "ext": "r", "lang": "R", "max_stars_repo_path": "Rscript.r", "max_stars_repo_name": "jonathansartorib/ProjectR-teste", "max_stars_repo_head_hexsha": "ecf07755bc64ee2ea77d580fb4c0c250c42bcffc", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2021-09-19T19:24:04.000Z", "max_stars_repo_stars_event_max_datetime": "2021-09-19T19:24:04.000Z", "max_issues_repo_path": "Rscript.r", "max_issues_repo_name": "jonathansartorib/ProjectR-teste", "max_issues_repo_head_hexsha": "ecf07755bc64ee2ea77d580fb4c0c250c42bcffc", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Rscript.r", "max_forks_repo_name": "jonathansartorib/ProjectR-teste", "max_forks_repo_head_hexsha": "ecf07755bc64ee2ea77d580fb4c0c250c42bcffc", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 19.1038251366, "max_line_length": 66, "alphanum_fraction": 0.5855263158, "num_tokens": 1279, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6959583376458152, "lm_q2_score": 0.7799929053683038, "lm_q1q2_score": 0.5428425657956544}} {"text": "#' @title\n#' Predict the response trajectory\n#'\n#' @description\n#' This function generates prediction of the functional response based on the fitting\n#' results of the NRRR model and the given predictor trajectory. The functional\n#' response is predicted at a given sequence of time points \\code{tseq}.\n#'\n#' @usage\n#' NRRR.pred(tseq, X, sseq, Ag, Bg, Al, Bl, phi)\n#'\n#' @param tseq a sequence of time points at which the prediction of the response\n#' trajectory is generated.\n#' @param sseq a sequence of time points at which the predictor trajectory is observed.\n#' @param X an array of dimension \\code{(n, p, length(sseq))} where n is\n#' the sample size and p is the number of components in the\n#' multivariate predictor. It is the predictor trajectory observed at\n#' discrete time points \\code{sseq}.\n#' @param phi a matrix of dimension \\code{length(sseq)}-by-jx. It is the set of\n#' basis functions to expand the predictor trajectory.\n#' @param Ag the estimated matrix U in a NRRR model.\n#' @param Bg the estimated matrix V in a NRRR model.\n#' @param Al the estimated matrix A in a NRRR model.\n#' @param Bl the estimated matrix B in a NRRR model.\n#'\n#'\n#' @return This function returns a list:\n#' \\item{Cstar}{the estimated coefficient matrix in equation (7) of the NRRR paper,\n#' i.e., \\eqn{(U \\otimes I_jy)A* B*^T(V \\otimes I_jx)^T}.}\n#' \\item{Ypred}{the predicted response trajectory which is an array of\n#' dimension \\code{(n, d, length(tseq))}}\n#'\n#'\n#' @references Liu, X., Ma, S., & Chen, K. (2020).\n#' Multivariate Functional Regression via Nested Reduced-Rank Regularization.\n#' arXiv: Methodology.\n#'\n#' @importFrom splines bs\n#' @export\n#' @examples\n#' library(NRRR)\n#' set.seed(3)\n#' # Simulation setting 2 in NRRR paper\n#' simDat <- NRRR.sim(n = 100, ns = 100, nt = 100, r = 3, rx = 3, ry = 3,\n#' jx = 8, jy = 8, p = 20, d = 20, s2n = 1, rho_X = 0.5,\n#' rho_E = 0, Sigma = \"CorrAR\")\n#' fit_nrrr <- with(simDat, NRRR.ic(Yest, Xest, Ag0 = NULL, Bg0 = NULL,\n#' jx = 8, jy = 8, p = 20, d = 20, n = 100,\n#' maxiter = 300, conv = 1e-4, method = c(\"RRR\", \"RRS\")[1],\n#' lambda = 0, ic = c(\"BIC\", \"BICP\", \"AIC\", \"GCV\")[1],\n#' dimred = c(TRUE, TRUE, TRUE), rankfix = NULL))\n#' Ypred_nrrr <- NRRR.pred(simDat$tseq, simDat$X, simDat$sseq,\n#' fit_nrrr$Ag, fit_nrrr$Bg, fit_nrrr$Al,\n#' fit_nrrr$Bl, simDat$phi)\nNRRR.pred <- function(tseq,X,sseq,Ag,Bg,Al,Bl,phi){\n n <- nrow(X)\n\n p <- nrow(Bg)\n d <- nrow(Ag)\n rx <- ncol(Bg)\n ry <- ncol(Ag)\n\n jx <- nrow(Bl)/rx\n jy <- nrow(Al)/ry\n\n nt <- length(tseq)\n ns <- length(sseq)\n\n # Note that you can predict at different time\n psi <- splines::bs(c(0,tseq),df = jy)[-1,]\n Jpsi <- matrix(nrow=jy,ncol=jy,0)\n tdiff <- (tseq - c(0,tseq[-nt]))\n for(t in 1:nt) Jpsi <- Jpsi + psi[t,]%*%t(psi[t,])*tdiff[t]\n eJpsi <- eigen(Jpsi)\n Jpsihalf <- eJpsi$vectors%*%diag(sqrt(eJpsi$values))%*%t(eJpsi$vectors)\n Jpsihalfinv <- eJpsi$vectors%*%diag(1/sqrt(eJpsi$values))%*%t(eJpsi$vectors)\n\n # Compute Cstar from estimated C\n alindex <- rep(1:ry,jy)\n Alstar <- Al[order(alindex),]\n blindex <- rep(1:rx,jx)\n Blstar <- Bl[order(blindex),]\n # Adjust\n Alstar <- kronecker(diag(ry),Jpsihalfinv)%*%Alstar\n Cstar <- kronecker(Ag,diag(jy))%*%Alstar%*%t(Blstar)%*%kronecker(t(Bg),diag(jx))\n\n # Integrated X\n # phi <- splines::bs(c(0,sseq),df = jx)[-1,]\n Xint <- matrix(nrow=p*jx,ncol=n,0)\n sdiff <- (sseq - c(0,sseq[-ns]))\n for(s in 1:ns){\n Xint <- Xint + kronecker(diag(nrow=p,ncol=p),phi[s,])%*%t(X[,,s])*sdiff[s]\n }\n\n\n # Compute predicted values\n Ypred <- array(dim=c(n,d,nt),NA)\n for(t in 1:nt){\n Psit <- kronecker(diag(nrow=d,ncol=d),t(psi[t,]))\n Ypred[,,t] <- t(Psit%*%Cstar%*%Xint)\n }\n\n list(Jpsihalf=Jpsihalf,Cstar=Cstar,Ypred=Ypred)\n\n}\n", "meta": {"hexsha": "f4f2022786d1edbe0065a470da5b2dfc8e6c7080", "size": 3943, "ext": "r", "lang": "R", "max_stars_repo_path": "R/pred_NRRR.r", "max_stars_repo_name": "xliu-stat/NRRR", "max_stars_repo_head_hexsha": "e51d9df7500b287f8c4daa8839f4cc1782525cef", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/pred_NRRR.r", "max_issues_repo_name": "xliu-stat/NRRR", "max_issues_repo_head_hexsha": "e51d9df7500b287f8c4daa8839f4cc1782525cef", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/pred_NRRR.r", "max_forks_repo_name": "xliu-stat/NRRR", "max_forks_repo_head_hexsha": "e51d9df7500b287f8c4daa8839f4cc1782525cef", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 37.1981132075, "max_line_length": 87, "alphanum_fraction": 0.6058838448, "num_tokens": 1374, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8031737963569016, "lm_q2_score": 0.6757646075489392, "lm_q1q2_score": 0.5427564252887132}} {"text": "#!/usr/bin/env Rscript\nlibrary(\"optparse\")\n\ncomputedher<-function(beta, se, af,N){\n#https://journals.plos.org/plosone/article/file?type=supplementary&id=info:doi/10.1371/journal.pone.0120758.s001\nmaf<-af\nmaf[!is.na(af) & af>0.5]<- 1 - maf[!is.na(af) & af>0.5]\nba<-!is.na(beta) & !is.na(se) & !is.na(maf) & !is.na(N)\na<-rep(NA, length(beta))\nb<-rep(NA, length(beta))\na<-2*(beta[ba]**2)*(maf[ba]*(1-maf[ba]))\nb<-2*(se[ba]**2)*N[ba]*maf[ba]*(1-maf[ba])\nres<-rep(NA, length(beta))\nres[ba]<-a[ba]/(a[ba]+b[ba])\nres\n}\n\n\n\n\noption_list = list(\n make_option(c(\"-f\", \"--file\"), type=\"character\", default=NULL, \n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--chro\"), type=\"character\", default=NULL, \n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--chro_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--bp_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--pheno_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--beta_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--ci_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--p_head\"), type=\"character\", default=NULL, \n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--n_head\"), type=\"character\", default=NULL, \n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--freq_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--rs_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--riskall_head\"), type=\"character\", default=NULL,\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--format\"), type=\"character\", default=\"USCS\",\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"--typeformat\"), type=\"character\", default=\"csv\",\n help=\"dataset file name\", metavar=\"character\"),\n make_option(c(\"-o\", \"--out\"), type=\"character\", default=\"out.txt\", \n help=\"output file name [default= %default]\", metavar=\"character\")\n); \n \nopt_parser = OptionParser(option_list=option_list);\nopt = parse_args(opt_parser);\n#bin \t590\tsmallint(5) unsigned \trange \tIndexing field to speed chromosome range queries.\n#chrom \tchr1\tvarchar(255) \tvalues \tReference sequence chromosome or scaffold\n#chromStart \t768252\tint(10) unsigned \trange \tStart position in chromosome\n#chromEnd \t768253\tint(10) unsigned \trange \tEnd position in chromosome\n#name \trs2977608\tvarchar(255) \tvalues \tID of SNP associated with trait\n#pubMedID \t31969693\tint(10) unsigned \trange \tPubMed ID of publication of the study\n#author \tColeman JRI\tvarchar(255) \tvalues \tFirst author of publication\n#pubDate \t2020-01-23\tvarchar(255) \tvalues \tDate of publication\n#journal \tMol Psychiatry\tvarchar(255) \tvalues \tJournal of publication\n#title \tGenome-wide gene-environmen...\tvarchar(1024) \tvalues \tTitle of publication\n#trait \tMajor depressive disorder\tvarchar(255) \tvalues \tDisease or trait assessed in study\n#initSample \t29,475 European ancestry ca...\tlongblob \t \tInitial sample size\n#replSample \tNA\tlongblob \t \tReplication sample size\n#region \t1p36.33\tvarchar(255) \tvalues \tChromosome band / region of SNP\n#genes \tNR\tlongblob \t \tReported Gene(s)\n#riskAllele \trs2977608-A\tlongblob \t \tStrongest SNP-Risk Allele\n#riskAlFreq \t0.259029\tvarchar(255) \tvalues \tRisk Allele Frequency\n#pValue \t8E-6\tvarchar(255) \tvalues \tp-Value\n#pValueDesc \t \tvarchar(255) \tvalues \tp-Value Description\n#orOrBeta \t1.0729614\tvarchar(255) \tvalues \tOdds ratio or beta\n#ci95 \t[1.04-1.1]\tvarchar(255) \tvalues \t95% Confidence Interval\n#platform \tAffymetrix [7791636] (imputed)\tvarchar(255) \tvalues \tPlatform and [SNPs passing QC]\n#cnv \tN\tenum('Y', 'N') \tvalues \tY if Copy Number Variant\n\n\n\n#lisp=\"Type 2 diabetes\";lisc=c(\"22\",\"21\")\nformat=opt[['format']]\nif(format==\"USCS\"){\nData<-read.csv(opt[['file']], header=F, sep='\\t')\nchrohead=\"chrom\";phenhead=\"trait\";poshead=\"chromEnd\";OrBetaHead=\"orOrBeta\";headCi95=\"ci95\";pValueHead=\"pValue\";nvalueHead=\"initSample\";freqHead=\"riskAlFreq\";rsHead=\"name\";riskall=\"riskAllele\"\nnames(Data)<-c(\"bin\", \"chrom\", \"chromStart\", \"chromEnd\", \"name\", \"pubMedID\", \"author\", \"pubDate\", \"journal\", \"title\", \"trait\", \"initSample\",\"replSample\",\"region\", \"genes\", \"riskAllele\", \"riskAlFreq\", \"pValue\", \"pValueDesc\", \"orOrBeta\", \"ci95\",\"platform\", \"cnv\")\n}else{\nchrohead=opt[['chro_head']];phenhead=opt[['pheno_head']];poshead=opt[['bp_head']];OrBetaHead=opt[['beta_head']];headCi95=opt[['ci_head']];pValueHead=opt[['p_head']];nvalueHead=opt[['n_head']];freqHead=opt[['freq_head']];rsHead=opt[['rs_head']];riskall=opt[['riskall_head']]\nif(opt[['typeformat']]=='csv')Data<-read.csv(opt[['file']])\nelse Data<-read.table(opt[['file']], header=T, sep='\\t')\n}\n\nData[,chrohead]<-gsub('chr', '',as.character(Data[,chrohead]))\n\nData2Sub<-Data #[, c(chrohead,poshead, OrBetaHead,headCi95,pValueHead,phenhead,nvalueHead,freqHead)]\n\nIC<-t(sapply(strsplit(gsub(\"[\", \"\", sapply(strsplit(as.character(Data2Sub[,headCi95]), split=']',fixed=T),function(x)x[1]),fixed=T), split=\"-\"),function(x){\nif(length(x)==2){return(c(as.numeric(x[1]),as.numeric(x[2])))\n}else{\nreturn(c(-as.numeric(x[2]),as.numeric(x[3])))\n}\n}))\nIC<-data.frame(lower.cat=IC[,1], upper.cat=IC[,2])\nData2Sub<-cbind(Data2Sub,IC)\nData2Sub$nsample.cat<-sapply(strsplit(as.character(Data2Sub[,nvalueHead]),split=\"[ ]\"),function(x)sum(as.integer(gsub(\",\", \"\",grep(\"[0-9]\", x,value=T)))))\nData2Sub$beta.cat<-as.numeric(as.character(Data2Sub[,OrBetaHead]))\nbalisebeta<-!is.na(Data2Sub$beta.cat) & Data2Sub$beta.cat>=1\nData2Sub$lower.cat[balisebeta]<-log2(Data2Sub$lower.cat[balisebeta])\nData2Sub$upper.cat[balisebeta]<-log2(Data2Sub$upper.cat[balisebeta])\n#Data2Sub<-Data2Sub[!is.na(Data2Sub$beta.cat) & !is.na(Data2Sub$lower.cat) & !is.na(Data2Sub$upper.cat) & !is.na(Data2Sub$nsample.cat) ,]\n\nData2Sub$beta.cat[balisebeta]<-log2(Data2Sub$beta.cat[balisebeta])\nData2Sub$risk.allele.af[balisebeta]<-as.numeric(as.character(Data2Sub[balisebeta,freqHead]))\nData2Sub$sd.cat<-(Data2Sub$upper.cat - Data2Sub$beta.cat)/1.96\nData2Sub$sd.cat2<-(Data2Sub$upper.cat - Data2Sub$beta.cat)/1.96*sqrt(Data2Sub$nsample.cat)\nData2Sub$z.cat.v1<-Data2Sub$beta.cat/(Data2Sub$sd.cat)\nData2Sub$z.cat<-qnorm(Data2Sub[,pValueHead],lower.tail=FALSE)\nData2Sub$h2.cat=computedher(Data2Sub$beta.cat, Data2Sub$sd.cat, Data2Sub$risk.allele.af,Data2Sub$nsample.cat)\nData2Sub$pvalue<-as.numeric(as.character(Data2Sub[,pValueHead]))\n#Data2Sub<-Data2Sub[!is.na(Data2Sub$h2.cat) & !is.na(Data2Sub[,pValueHead]),]\n#Data2Sub<-Data2Sub[order(Data2Sub$h2.cat),]\nData2Sub$order<-1:nrow(Data2Sub)\n#Good<-aggregate(as.formula(paste(\"order~\",chrohead,\"+\",poshead,sep=\"\")), data=Data2Sub, min)$order\n#Data2Sub<-Data2Sub[Data2Sub$order %in% Good,]\n\nwrite.csv(Data2Sub, file=paste(opt[['out']], '_all.csv',sep=''), row.names=F)\nwriteLines(unique(as.character(Data2Sub[,phenhead])), con=paste(opt[['out']], '_pheno.csv',sep=''))\nwrite.csv(as.data.frame(table(Data2Sub[,phenhead])), file=paste(opt[['out']], '_phenocount.csv',sep=''))\n\n\n", "meta": {"hexsha": "abf5b892f477885e797247f781ebc136ffc5320e", "size": 7396, "ext": "r", "lang": "R", "max_stars_repo_path": "replication/gwascat/bin/format_gwascat_pheno.r", "max_stars_repo_name": "bioinformatics-lab/h3agwas", "max_stars_repo_head_hexsha": "7470ea6097abeb7d150e7777340468e4ed85dca9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 62, "max_stars_repo_stars_event_min_datetime": "2016-08-29T11:27:35.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-10T17:16:14.000Z", "max_issues_repo_path": "replication/gwascat/bin/format_gwascat_pheno.r", "max_issues_repo_name": "bioinformatics-lab/h3agwas", "max_issues_repo_head_hexsha": "7470ea6097abeb7d150e7777340468e4ed85dca9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 33, "max_issues_repo_issues_event_min_datetime": "2016-12-26T13:48:19.000Z", "max_issues_repo_issues_event_max_datetime": "2021-12-05T13:34:06.000Z", "max_forks_repo_path": "replication/gwascat/bin/format_gwascat_pheno.r", "max_forks_repo_name": "bioinformatics-lab/h3agwas", "max_forks_repo_head_hexsha": "7470ea6097abeb7d150e7777340468e4ed85dca9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 50, "max_forks_repo_forks_event_min_datetime": "2017-04-15T04:17:43.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-30T07:26:01.000Z", "avg_line_length": 56.0303030303, "max_line_length": 273, "alphanum_fraction": 0.694835046, "num_tokens": 2304, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8519528019683106, "lm_q2_score": 0.6370307944803831, "lm_q1q2_score": 0.5427201702976614}} {"text": "#Castigo = 25\r\n\r\nfit = function(observed_I, beta0, beta_min, beta_max, tau10, tau1_min, tau1_max,\r\n tau20, tau2_min, tau2_max,\r\n N00, N0_min, N0_max, Castigo, ignore_beta_diff, max_eval = 1000 ) {\r\n \"\r\n Fits the model to observed daily new case counts.\r\n \r\n Parameters\r\n ----------\r\n observed_I : numeric vector\r\n Daily observed number of new cases. (son los incubadores!! o son los infecciosos nuevos)\r\n beta0 : numeric vector\r\n Initial value for beta; the expected number of cases (incubadores) stemming from a single\r\n person in a single day.\r\n beta_min : numeric\r\n Beta lower bound.\r\n beta_max : numeric\r\n Beta upper bound.\r\n ignore_beta_diff : numeric vector\r\n List of beta indices for which differences should be ignored while\r\n calculating the loss function. This amounts to moments in time in which we\r\n allow the beta series to be discontinuous.\r\n \r\n Returns\r\n -------\r\n beta : numeric vector\r\n Expected number of cases stemming from a single person in a single day.\r\n \"\r\n # concatenate initial parameters into a single vector for optimization\r\n x0 = c(beta0, tau10, tau20, N00)\r\n steps = length(beta0)\r\n lb = c(rep(beta_min, steps), tau1_min, tau2_min, N0_min)\r\n ub = c(rep(beta_max, steps), tau1_max, tau2_max, N0_max)\r\n # set optimization parameters\r\n opts = list(\"algorithm\" = \"NLOPT_LN_BOBYQA\", \"xtol_rel\" = 1.0e-7, \"maxeval\" = max_eval)\r\n print(opts)\r\n # define loss function\r\n loss = function(x) {\r\n # unpack values\r\n beta = x[1:steps]\r\n tau1 = x[steps + 1]\r\n tau2 = x[steps + 2]\r\n N0 = x[steps + 3]\r\n # calculate loss\r\n expected = get_expected(beta, N0, tau1, tau2) # Obtiene el número de casos infectados\r\n beta_diff = diff(beta)\r\n regularization1 = (beta_diff/beta[1:length(beta_diff)]) \r\n regularization = (diff(regularization1)) ^ 2\r\n regularization = regularization[!1:length(regularization) %in% ignore_beta_diff]\r\n observed_I=round(observed_I,0)\r\n factSum=function(x){\r\n return(sum(log(1:round(x,0))))\r\n }\r\n loglikelihood=observed_I*log(expected$NN)-expected$NN-sapply(as.matrix(observed_I), FUN=factSum)\r\n loglikelihood[observed_I==0]=-expected$NN[observed_I==0]\r\n \r\n loss = -mean(loglikelihood) + Castigo * mean(regularization)\r\n #loss=sum((observed_I-expected$I)^2)+ sum(regularization)\r\n loss\r\n }\r\n result = nloptr(x0=x0, eval_f=loss, eval_grad_f=NULL, lb=lb, ub=ub, opts=opts)\r\n model = list()\r\n model$beta = result$solution[1:steps]\r\n model$tau1 = result$solution[steps + 1]\r\n model$tau2 = result$solution[steps + 2]\r\n model$N0 = result$solution[steps + 3]\r\n model$loss = result$objective\r\n beta=model$beta\r\n beta_diff = diff(beta)\r\n regularization1 = (beta_diff/beta[1:length(beta_diff)]) \r\n regularization = (diff(regularization1)) ^ 2\r\n regularization = regularization[!1:length(regularization) %in% ignore_beta_diff]\r\n model$likelihood=(model$loss-Castigo*mean(regularization))*length(beta)\r\n return(model)\r\n}\r\n", "meta": {"hexsha": "bea4ded49897d76837e69289ad924bcc872923b1", "size": 3010, "ext": "r", "lang": "R", "max_stars_repo_path": "fit_confidence_intervals/f_fit.r", "max_stars_repo_name": "secg95/INS_COVID", "max_stars_repo_head_hexsha": "d3e1e9b2d83de9bbfb246c9be93624ffb4df88a1", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "fit_confidence_intervals/f_fit.r", "max_issues_repo_name": "secg95/INS_COVID", "max_issues_repo_head_hexsha": "d3e1e9b2d83de9bbfb246c9be93624ffb4df88a1", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "fit_confidence_intervals/f_fit.r", "max_forks_repo_name": "secg95/INS_COVID", "max_forks_repo_head_hexsha": "d3e1e9b2d83de9bbfb246c9be93624ffb4df88a1", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 39.0909090909, "max_line_length": 101, "alphanum_fraction": 0.6790697674, "num_tokens": 854, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8152324983301568, "lm_q2_score": 0.665410572017153, "lm_q1q2_score": 0.5424643230408424}} {"text": "# purpose: calculate the heterosis (MPH and HPH)\n# Jinliang Yang\n# 2.1.2012\n# updated: 4.12.2012\n\nsetwd(\"/Users/yangjl/Documents/Heterosis_GWAS/pheno2011\");\n#### Heterosis ######################################################\n\ndiallel <- read.csv(\"cache/pheno_diallel_master_BLUE.csv\")\ndim(diallel)\n# 225 8\np <- read.csv(\"cache/pheno_namparents_blue.csv\")\ndim(p)\n# 27 8\n\nnamid <- read.table(\"nam_parents\", header=TRUE)\nnamid <- namid[, c(1,3)]\n\ndiallel$z1 <- NA;\ndiallel$z2 <- NA;\ndiallel$Genotype <- as.character(diallel$Genotype)\nfor(i in 1:nrow(diallel)){\n tem <- unlist(strsplit(diallel$Genotype[i], split=\"x\"))\n diallel$z1[i] <- tem[1];\n diallel$z2[i] <- tem[2];\n}\n\ndiallel <- merge(namid, diallel, by.x=\"pop\", by.y=\"z1\")\ndiallel <- merge(namid, diallel, by.x=\"pop\", by.y=\"z2\")\ndiallel <- diallel[, c(5, 3,1, 4,2, 6:12)]\nnames(diallel)[2:5] <- c(\"z1\", \"z2\", \"p1\", \"p2\")\ndiallel$p1 <- toupper(diallel$p1);\ndiallel$p2 <- toupper(diallel$p2);\n\ndiallel$KRN_HPH <- NA\ndiallel$KRN_MPH <- NA\ndiallel$CL_HPH <- NA\ndiallel$CL_MPH <- NA\ndiallel$CW_HPH <- NA\ndiallel$CW_MPH <- NA\ndiallel$CD_HPH <- NA\ndiallel$CD_MPH <- NA\ndiallel$AKW_HPH <- NA\ndiallel$AKW_MPH <- NA\ndiallel$TKW_HPH <- NA\ndiallel$TKW_MPH <- NA\ndiallel$KC_HPH <- NA\ndiallel$KC_MPH <- NA\n\n##########################################################################################################\np$Genotype <- as.character(p$Genotype)\nfor (i in 1:nrow(diallel)){\n p1 <- as.character(diallel$p1[i]);\n p2 <- as.character(diallel$p2[i]);\n diallel$KRN_HPH[i] <- diallel$KRN[i] - max(p[p$Genotype==p1, ]$KRN, p[p$Genotype==p2, ]$KRN);\n diallel$KRN_MPH[i] <- diallel$KRN[i] - mean(c(p[p$Genotype==p1, ]$KRN, p[p$Genotype==p2, ]$KRN), na.rm=T);\n diallel$CL_HPH[i] <- diallel$CL[i] - max(p[p$Genotype==p1, ]$CL, p[p$Genotype==p2, ]$CL);\n diallel$CL_MPH[i] <- diallel$CL[i] - mean(c(p[p$Genotype==p1, ]$CL, p[p$Genotype==p2, ]$CL), na.rm=T);\n diallel$CW_HPH[i] <- diallel$CW[i] - max(p[p$Genotype==p1, ]$CW, p[p$Genotype==p2, ]$CW);\n diallel$CW_MPH[i] <- diallel$CW[i] - mean(c(p[p$Genotype==p1, ]$CW, p[p$Genotype==p2, ]$CW), na.rm=T);\n diallel$CD_HPH[i] <- diallel$CD[i] - max(p[p$Genotype==p1, ]$CD, p[p$Genotype==p2, ]$CD);\n diallel$CD_MPH[i] <- diallel$CD[i] - mean(c(p[p$Genotype==p1, ]$CD, p[p$Genotype==p2, ]$CD), na.rm=T);\n diallel$AKW_HPH[i] <- diallel$AKW[i] - max(p[p$Genotype==p1, ]$AKW, p[p$Genotype==p2, ]$AKW);\n diallel$AKW_MPH[i] <- diallel$AKW[i] - mean(c(p[p$Genotype==p1, ]$AKW, p[p$Genotype==p2, ]$AKW), na.rm=T);\n diallel$TKW_HPH[i] <- diallel$TKW[i] - max(p[p$Genotype==p1, ]$TKW, p[p$Genotype==p2, ]$TKW);\n diallel$TKW_MPH[i] <- diallel$TKW[i] - mean(c(p[p$Genotype==p1, ]$TKW, p[p$Genotype==p2, ]$TKW), na.rm=T);\n diallel$KC_HPH[i] <- diallel$KC[i] - max(p[p$Genotype==p1, ]$KC, p[p$Genotype==p2, ]$KC);\n diallel$KC_MPH[i] <- diallel$KC[i] - mean(c(p[p$Genotype==p1, ]$KC, p[p$Genotype==p2, ]$KC), na.rm=T);\n}\n\nwrite.table(diallel, \"cache/pheno_diallel_master_BLUE_heterosis.csv\", sep=\",\", quote=FALSE, row.names=FALSE)\n\n", "meta": {"hexsha": "44773ca0a5f006c9a984c39a63d3414cdca2093f", "size": 3015, "ext": "r", "lang": "R", "max_stars_repo_path": "profiling/pheno2011/02-C.pheno_diallel_heterosis.r", "max_stars_repo_name": "yangjl/Heterosis-GWAS", "max_stars_repo_head_hexsha": "454208509c22b1269f17ba63452ef19a9c3d13f8", "max_stars_repo_licenses": ["RSA-MD"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2021-04-16T08:27:37.000Z", "max_stars_repo_stars_event_max_datetime": "2021-10-31T13:00:43.000Z", "max_issues_repo_path": "profiling/pheno2011/02-C.pheno_diallel_heterosis.r", "max_issues_repo_name": "yangjl/Heterosis-GWAS", "max_issues_repo_head_hexsha": "454208509c22b1269f17ba63452ef19a9c3d13f8", "max_issues_repo_licenses": ["RSA-MD"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "profiling/pheno2011/02-C.pheno_diallel_heterosis.r", "max_forks_repo_name": "yangjl/Heterosis-GWAS", "max_forks_repo_head_hexsha": "454208509c22b1269f17ba63452ef19a9c3d13f8", "max_forks_repo_licenses": ["RSA-MD"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2020-01-03T14:35:43.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-03T01:34:08.000Z", "avg_line_length": 41.301369863, "max_line_length": 108, "alphanum_fraction": 0.6072968491, "num_tokens": 1337, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104789086703224, "lm_q2_score": 0.6688802603710086, "lm_q1q2_score": 0.5421133434566161}} {"text": "################################### INTERPOLACAO VIA TRIANGULACAO ##################################\n\n#' Triangulacao De \\code{curvacolina}\n#' \n#' Funcao interna executada quando \\code{metodo = \"triangulacao\"} em \\code{interpolador}. O \n#' argumento \\code{tessfunc} permite controlar como o espaco projetado sera tesselado em triangulos. \n#' Atualmente ha duas opcoes implementadas no pacote:ang\n#' \n#' \\itemize{\n#' \\item{\\code{\\link{tessdelaunay}}}\n#' \\item{\\code{\\link{tessradial}}}\n#' }\n#' \n#' A primeira executa a tesselacao pelo metodo de Delaunay atraves da funcao \n#' \\code{\\link[geometry]{delaunayn}}. A segunda e uma versao alternativa quase igual a primeira, com\n#' excessao dos triangulos definidos entre a ultima curva e o maximo. Pelo metodo de Delaunay e \n#' possivel que acontecam platos, isto e, triangulos cujos tres vertices pertencem a mesma curva.\n#' \\code{tessradial} interfere apenas nessa regiao, forcando para que todos os triangulos entre a \n#' ultima curva e o maximo tenham este ponto como um vertice, deixando os dois restantes como pontos\n#' adjacentes na ultima curva.\n#' \n#' Esta funcao nao deve ser chamada pelo usuario diretamente na maioria dos casos\n#' \n#' @param colina objeto \\code{curvacolina} retornado pelas funcoes de leitura\n#' @param tessfunc funcao ou string com nome da funcao pela qual executar a tesselacao do espaco.\n#' Ver Detalhes\n#' @param ... nao possui funcao, so existe para compatibilizacao com a chamada generica de \n#' \\code{\\link{interpolador}}\n#' \n#' @return objeto da classe \\code{triangulacao} contendo a tesselacao da curva colina\n#' \n#' @export\n\ntriangulacao <- function(colina, tessfunc = tessdelaunay, ...) {\n hl <- pot <- NULL\n\n if(is.character(tessfunc)) tessfunc <- as.name(tessfunc)\n tri <- eval(as.call(list(tessfunc, colina)))\n\n new_triangulacao(tri, colina)\n}\n\nnew_triangulacao <- function(tri, colina) {\n\n obj <- list(triangulos = tri, colina = colina)\n\n class(obj) <- c(\"triangulacao\", \"interpolador\")\n attr(obj, \"ntri\") <- nrow(tri)\n\n return(obj)\n}\n\n#' @export\n\nprint.triangulacao <- function(x, ...) {\n cat(\"* Tesselacao \", \"\\n\")\n cat(\"Numero de triangulos: \", attr(x, \"ntri\"), \"\\n\")\n cat(\"-----\\n\")\n cat(\"* Curva colina \\n\")\n summary(x$colina)\n}\n\n# FUNCOES DE TRIANGULACAO --------------------------------------------------------------------------\n\n#' Triangulacao Delauney\n#' \n#' Realiza triangulacao do espaco pelo metodo de Delauney via \\code{geometry::delaunayn}\n#' \n#' @param colina objeto \\code{curvacolina} contendo curva a tesselar\n#' \n#' @return matriz com tres colunas indicando, em cada linha, o indice em \\code{colina$CC} dos pontos\n#' correspondentes aos vertices de cada triangulo gerado\n\ntessdelaunay <- function(colina) {\n hl <- pot <- NULL\n geometry::delaunayn(colina$CC[, list(hl, pot)])\n}\n\n#' Triangulacao Radial\n#' \n#' Define os triangulos sempre usando o maximo como um dos vertices\n#' \n#' Este e um metodo de triangulacao especifico para uso em torno do maximo. Cada triangulo e \n#' definido usando sempre o maximo como um dos vertices. Os dois vertices restantes sao pontos \n#' adjacentes entre si na curva de rendimento imediatamente inferior ao maximo, ou seja, o espaco \n#' entre ultima curva e ponto maximo e fatiado em triangulos com arestas radiais, para todos os \n#' pontos da ultima curva.\n#' \n#' @param colina um data.frame ou data.table contendo queda liquida, potencia e rendimento\n#' \n#' @return matriz de tres colunas indicando o indicie em \\code{dat} dos pontos correspondentes aos\n#' vertices de cada triangulo. Cada linha corresponde a um triangulo\n#' \n#' @importFrom utils tail\n\ntessradial <- function(colina) {\n\n rend <- NULL\n\n tri <- tessdelaunay(colina)\n\n # identifica triangulos da ultima curva de rend para dentro\n ultrends <- tail(attr(colina, \"rends\"), 2)\n innertri <- sapply(seq(ncol(tri)), function(i) colina$CC$rend[tri[, i]] %in% ultrends)\n innertri <- rowMeans(innertri) == 1\n tri <- tri[!innertri, ]\n\n dat <- copy(colina$CC)\n chl <- dat[rend == ultrends[2]]$hl\n cpot <- dat[rend == ultrends[2]]$pot\n dat <- dat[rend == ultrends[1]]\n angord <- orderpoly(dat, chl, cpot)\n\n N1 <- nrow(dat)\n N2 <- nrow(colina$CC[!(rend %in% ultrends)])\n N3 <- nrow(colina$CC)\n\n out <- cbind(angord, c(angord[-1], angord[1])) + N2\n out <- cbind(out, N3)\n out <- rbind(tri, out)\n\n return(out)\n}\n\n# METODOS ------------------------------------------------------------------------------------------\n\n#' @rdname getcolina\n\ngetcolina.triangulacao <- function(object) object$colina\n\n#' Amostragem De Pontos Na Triangulacao\n#' \n#' Realiza interpolacao baricentrica de \\code{pontos} nos triangulos da tesselacao\n#' \n#' @param object objeto da classe \\code{triangulacao} retornado pela funcao homonima\n#' @param pontos data.frame ou matriz contendo coordenadas \\code{(hl, pot)} dos pontos onde \n#' interpolar\n#' @param as.gradecolina booleano -- se \\code{FALSE} (padrao) retorna apenas o vetor de rendimentos\n#' interpolados nas coordenadas \\code{pontos}; se \\code{TRUE} um objeto \\code{gradecolina}. Veja\n#' \\code{\\link{gradecolina}}\n#' @param ... existe somente para consistencia de metodos. Nao possui utilidade\n#' \n#' @return se \\code{as.gradecolina = FALSE}, vetor de rendimentos interpolados, do contrario um \n#' objeto \\code{\\link{gradecolina}}\n#' \n#' @export\n\npredict.triangulacao <- function(object, pontos, as.gradecolina = FALSE, ...) {\n\n pontos <- pontos[complete.cases(pontos), ]\n\n if(nrow(pontos) == 0) return(numeric(0))\n\n npontos <- nrow(pontos)\n pontos <- data.matrix(pontos)\n\n triangulos <- object$triangulos\n colina <- data.matrix(object$colina$CC)\n\n barycoord <- geometry::tsearchn(colina[, c(\"hl\", \"pot\")], triangulos, pontos)\n\n rends <- sapply(seq(npontos), function(i) {\n indtri <- barycoord$idx[i]\n vertices <- triangulos[indtri, ]\n rends <- colina[vertices, \"rend\"]\n sum(barycoord$p[i, ] * rends)\n })\n\n if(as.gradecolina) {\n out <- new_gradecolina(pontos, rends, object)\n } else {\n out <- as.numeric(rends)\n }\n\n return(out)\n}", "meta": {"hexsha": "598be05e11b73c4bb78164fa8bc0ceaef3c8870b", "size": 6171, "ext": "r", "lang": "R", "max_stars_repo_path": "R/mod-triangulacao.r", "max_stars_repo_name": "lkhenayfis/gtdp-curvacolina", "max_stars_repo_head_hexsha": "a3583255f8dfeb9ef14a6195055deac2b8938ba4", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/mod-triangulacao.r", "max_issues_repo_name": "lkhenayfis/gtdp-curvacolina", "max_issues_repo_head_hexsha": "a3583255f8dfeb9ef14a6195055deac2b8938ba4", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2022-01-29T15:10:52.000Z", "max_issues_repo_issues_event_max_datetime": "2022-01-30T16:13:23.000Z", "max_forks_repo_path": "R/mod-triangulacao.r", "max_forks_repo_name": "lkhenayfis/gtdp-curvacolina", "max_forks_repo_head_hexsha": "a3583255f8dfeb9ef14a6195055deac2b8938ba4", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.0625, "max_line_length": 101, "alphanum_fraction": 0.6608329282, "num_tokens": 1800, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7826624789529375, "lm_q2_score": 0.6926419767901475, "lm_q1q2_score": 0.5421048865814398}} {"text": "#' bootstrap.face\n#'\n#' Apply the bootstrap test for testing a quadratic polynomial covariance\n#'\n#' @param data \"data frame with three arguments:\n#' (1) \"argvals\": observation times;\n#' (2) \"subj\": subject indices;\n#' (3) \"y\": values of observations;\n#' Note that: we only handle complete data, so missing values are not allowed at this moment\n#' @param nbs number of bootstrap samples, default = 1000\n#' @param argvals.new \"argvals.new\" if we want the estimated covariance function at \"argvals.new\"; if NULL,\n#' then 100 equidistant points in the range of \"argvals\" in \"data\"\n#' @param semi.iter indicator whether to use semi_iterative for bootstrap\n#' @param tune.bs indicator whether to use tunning for bootstrap\n#' @param center \"center\" means if we want to compute population mean\n#' @param knots number of interior knots for B-spline basis functions to be used;\n#' @param p the degrees of B-splines; defaults to 3.\n#' @param m the order of differencing penalty; defaults to 2.\n#' @param lambda the value of the smoothing parameter for covariance smoothing; defaults to NULL.\n#' @param lambda_mean the value of the smoothing parameter for mean smoothing; defaults to NULL.\n#' @param search.length the number of equidistant (log scale) smoothing parameters to search; defaults to 14.\n#' @param upper,lower bounds for log smoothing parameter for first step of estimation; defaults are -3 and 10, respectively.\n#' @param pve Defaults 0.99. To select the number of eigenvalues by percentage of variance.\n#'\n#'\n#' @import nlme\n#' @return an object \"bootstrap.face\" contain:\n#' fit.alt: Alternative model fit (functional principal components analysis)\n#' fit.null: Null model fit (linear random effects)\n#' C.alt: Covariance matrix under alternative model\n#' C.null: Covariance matrix under null model\n#' tn: Test statistic\n#' p: p-value for test statistic based on the bs.approx\n#' bs.approx: List of values from the null distribution of Tn\n#'\n#' @references modified from face.sparse.inner from face package\n#' and bootstrap.test.R written by Stephanie\nbootstrap.face <- function(data, nbs = 1000, argvals.new = NULL,\n trunc.eig = 1,\n semi.iter = F, center.bs = F,\n center = TRUE,\n knots = 7, knots.option = \"equally-spaced\",\n p = 3, m = 2, lambda = NULL, lambda_mean = NULL,\n search.length = 14,\n lower = -3, upper = 10,\n pve = 0.99, off_diag = F, gam.mgcv = T, no.pen = F, dense.grid = 8e5) {\n #########################\n #### step 0: read in data\n #########################\n check.data(data)\n nb <- knots + p\n\n y <- data$y\n t <- data$argvals\n subj <- data$subj\n tnew <- argvals.new\n if (is.null(tnew)) tnew <- seq(min(t), max(t), length = 100)\n\n fit_mean <- NULL\n\n knots.initial <- knots\n #########################\n #### step 1: Test statistics\n #########################\n\n ###### a. Initialize C, X, Q\n r <- y\n if (center) {\n if (gam.mgcv) {\n fit_mean <- mgcv::gam(as.vector(y) ~ s(t, k = nb))\n } else {\n fit_mean <- pspline(data, argvals.new = tnew, knots = knots.initial, lambda = lambda_mean)\n }\n r <- y - fit_mean$fitted.values\n }\n\n raw <- raw.construct(data.frame(\"argvals\" = t, \"subj\" = subj, \"y\" = as.vector(r)))\n C <- raw$C # C\n st <- raw$st\n N2 <- raw$N2\n n0 <- raw$n0\n # indicator where ti = tj\n delta <- Matrix((st[, 1] == st[, 2]) * 1) # sparse\n delta <- c(as.matrix(delta))\n\n knots <- construct.knots(t, knots, knots.option, p)\n ## construct design & penalty on column dimension\n List <- pspline.setting(st[, 1], knots = knots, p, m, type = \"simple\", knots.option = knots.option)\n B1 <- List$B\n B1 <- Matrix(B1)\n DtD <- List$P\n ## construct design on row dimension\n B2 <- spline.des(knots = knots, x = st[, 2], ord = p + 1, outer.ok = TRUE, sparse = TRUE)$design\n c <- dim(B1)[2]\n c2 <- c * (c + 1) / 2\n ## combine tensor product design\n B <- Matrix(t(KhatriRao(Matrix(t(B2)), Matrix(t(B1)))))\n G <- Matrix(duplication.matrix(c))\n ### double the weight for off-diagonal\n # v_idx <- which(delta == 0)\n # B[v_idx, ] <- sqrt(2) * B[v_idx, ]\n if (off_diag) {\n B <- B * (sqrt(2) * (1 - delta) + delta)\n C <- C * (sqrt(2) * (1 - delta) + delta)\n }\n\n # BtWB <- matrix(0, nrow = c^2, ncol = c^2)\n # Wdelta <- c()\n # WC <- c()\n # for (i in 1:n0) {\n # # select combos for object i\n # seq <- (sum(N2[1:i]) - N2[i] + 1):(sum(N2[1:i]))\n # B3 <- Matrix(matrix(B[seq, ], nrow = length(seq)))\n # BtWB <- BtWB + crossprod(B3, B3)\n # Wdelta <- c(Wdelta, as.matrix(delta[seq]))\n # WC <- c(WC, as.matrix(C[seq]))\n # }\n BtB <- crossprod(B)\n BG <- B %*% G # sparse\n\n GtBtBG <- crossprod(G, BtB %*% G)\n detde <- crossprod(delta) # detWde = sum(delta)\n GtBtdelta <- crossprod(BG, delta)\n XtX <- rbind(cbind(GtBtBG, GtBtdelta), cbind(t(GtBtdelta), detde))\n ## design X\n X <- cbind(BG, delta)\n #XtX <- crossprod(X)\n ## penalty\n P <- crossprod(G, Matrix(suppressMessages(kronecker(diag(c), DtD)))) %*% G\n Q <- bdiag(P, 0)\n\n ###### b. Pre-calculate s, F, B^*, Li, g, f, G (2 eigens)\n eSig <- eigen(XtX, symmetric = TRUE)\n V <- eSig$vectors\n E <- eSig$values\n E <- E + 0.000001 * max(E)\n Sigi_sqrt <- matrix.multiply(V, 1 / sqrt(E)) %*% t(V)\n\n tUQU <- crossprod(Sigi_sqrt, (Q %*% Sigi_sqrt))\n Esig <- eigen(tUQU, symmetric = TRUE)\n # s, f, F(m_F here)\n U <- Esig$vectors\n s <- Esig$values\n A0 <- Sigi_sqrt %*% U\n m_F <- as.matrix(X %*% A0)\n\n m_FtF <- crossprod(m_F) # FTF\n f <- crossprod(m_F, C) # f=FTC\n\n c2 <- c2 + 1\n g <- rep(0, c2)\n G1 <- matrix(0, c2, c2)\n mat_list <- list()\n # G(G1 here), g, Li\n for (i in 1:n0) {\n seq <- (sum(N2[1:i]) - N2[i] + 1):(sum(N2[1:i]))\n Fi <- matrix(m_F[seq, ], nrow = length(seq))\n Li <- crossprod(Fi)#Li <- FitFi\n\n fi <- crossprod(Fi, C[seq]) # t(Fi)Ci\n\n g <- g + fi * fi\n G1 <- G1 + Li * (fi %*% t(f))\n\n LList <- list()\n LList[[1]] <- Li\n #LList[[2]] <- Li\n mat_list[[i]] <- LList\n }\n # B^*\n st.construct <- function(tnew) {\n m1 <- length(tnew)\n if (m1 > 1) {\n st <- cbind(vech(kronecker(tnew, t(rep(1, m1)))),\n vech(kronecker(rep(1, m1), t(tnew))))\n } else if (m1 == 1) {\n st <- rbind(st, c(tnew, tnew))\n }\n st\n }\n #stnew <- st.construct(tnew)\n #Bnew1 <- spline.des(knots = knots, x = stnew[, 1], ord = p + 1, outer.ok = TRUE, sparse = TRUE)$design\n #Bnew2 <- spline.des(knots = knots, x = stnew[, 2], ord = p + 1, outer.ok = TRUE, sparse = TRUE)$design\n #Bstar <- Matrix(t(KhatriRao(Matrix(t(Bnew2)), Matrix(t(Bnew1)))))\n\n\n #Xstar <- Bstar %*% G\n #delta_star <- which(stnew[, 1] != stnew[, 2])\n #Xstar[delta_star, ] <- sqrt(2) * Xstar[delta_star, ]\n dense.t <- seq(min(tnew), max(tnew), length.out = dense.grid)\n Xstar <- spline.des(knots = knots, x = dense.t, ord = p + 1, outer.ok = TRUE, sparse = TRUE)$design\n Xstar <- crossprod(Xstar) / dense.grid\n Eigen1 <- eigen(Xstar)\n Xstar.half <- Eigen1$vectors %*% diag(sqrt(Eigen1$values)) %*% t(Eigen1$vectors)\n Xstar.invhalf <- Eigen1$vectors %*% diag(1 / sqrt(Eigen1$values)) %*% t(Eigen1$vectors)\n\n Bnew <- spline.des(knots = knots, x = tnew, ord = p + 1, outer.ok = TRUE, sparse = TRUE)$design\n\n ###### c. Null estimate Ctilde0, f0 (2 eigens)\n fitNull <- function(data) {\n try(nlme::lme(y ~ 1, random = list(subj = pdSymm(~ 1 + argvals)),\n data = data), silent = T)\n }\n fit.null <- fitNull(data.frame(\"argvals\" = t, \"subj\" = subj, \"y\" = as.vector(r)))\n if (\"try-error\" %in% class(fit.null)) {# issue with null fit\n stop(fit.null)\n }\n calc.R0 <- function(fit.null, times) {\n var.mat <- VarCorr(fit.null)\n sigsq0 <- as.numeric(var.mat[1, 1])\n sigsq1 <- as.numeric(var.mat[2, 1])\n cov01 <- as.numeric(var.mat[2, 3]) * sqrt(sigsq0 * sigsq1) # corr->cov\n Rbar0 <- sigsq0 + cov01 * (times[, 1] + times[, 2]) + sigsq1 * (times[, 1] * times[, 2])\n if (off_diag) {\n Rbar0 <- Rbar0 * (sqrt(2) * (1 - delta) + delta)\n }\n list(Rbar0 = Rbar0, coef.null = c(sigsq0, cov01, sigsq1))\n }\n\n Rbar0.fit <- calc.R0(fit.null, st)\n C0 <- Rbar0.fit$Rbar0\n\n ###### d. tunning(s, F, F^*, Li, g, f, G) to get lambda^*\n Lambda <- seq(lower, upper, length = search.length)\n Gcv <- 0 * Lambda\n gcv <- function(x) {\n lambda <- exp(x)\n d <- 1 / (1 + lambda * s)\n f_d <- f * d\n cv0 <- -2 * sum(f_d * f)\n cv1 <- sum(f_d * (m_FtF %*% f_d))\n cv2 <- 2 * sum(d * g)\n cv3 <- -4 * sum(d * (G1 %*% d))\n cv4 <- sum(unlist(sapply(mat_list, function(x) {\n a <- x[[1]] %*% f_d\n 2 * sum(a * a * d)\n })))\n cv <- cv0 + cv1 + cv2 + cv3 + cv4\n return(cv)\n }\n if (is.null(lambda)) {\n Lambda <- seq(lower, upper, length = search.length) # construct lambda grid\n Length <- length(Lambda)\n Gcv <- rep(0, Length)\n for (i in 1:Length) {\n Gcv[i] <- gcv(Lambda[i])\n }\n i0 <- which.min(Gcv)\n lambda <- exp(Lambda[i0]) # lambda^*\n }\n\n ###### e. calculate estimated covariance function and test statistics\n Xxstar <- crossprod(Bnew)\n m_est_sigma <- (t(A0[c2, ]) * t(1 / (1 + lambda * s))) %*% t(m_F)\n sigsq <- m_est_sigma %*% C\n if (sigsq <= 0.000001) {\n # warning(\"error variance cannot be non-positive, reset to 1e-6!\")\n sigsq <- 0.000001\n }\n if (no.pen || lambda == 0) {\n m_est <- A0[-c2, ] %*% t(m_F)\n } else {\n m_est <- matrix.multiply(A0[-c2, ], 1 / (1 + lambda * s)) %*% t(m_F)\n }\n\n trun_mat <- function(C) {\n alpha <- m_est %*% C\n Theta <- G %*% alpha\n Theta <- matrix(Theta, c, c)\n # sigma2 <- alpha[c2]\n # if (sigma2 <= 0.000001) {\n # # warning(\"error variance cannot be non-positive, reset to 1e-6!\")\n # sigma2 <- 0.000001\n # }\n if (trunc.eig != 0) {\n # make sure Theta positive definite(2 eigens)\n Eigen <- eigen(Theta, symmetric = TRUE)\n Eigen$values[Eigen$values < 0] <- 0\n npc <- sum(Eigen$values > 0) # which.max(cumsum(Eigen$values)/sum(Eigen$values)>pve)[1]\n if (npc > 1) {\n pc <- Bnew %*% Eigen$vectors[, 1:npc]\n C <- tcrossprod(matrix.multiply(pc, Eigen$values[1:npc]), pc)\n }\n if (npc == 1) {\n C <- Eigen$values[1] * tcrossprod(Bnew %*% Eigen$vectors[, 1])\n }\n } else {\n C <- as.matrix(tcrossprod(Bnew %*% Matrix(Theta), Bnew))\n }\n list(C = C, Theta = Theta)\n }\n\n trunc.mat.inner <- function(Theta, trunc.eig = 1) {\n if (trunc.eig == 1) {\n eigen.fit <- eigen(Theta, symmetric = T)\n efuncs <- eigen.fit$vectors\n } else if (trunc.eig == 2) {\n Theta <- as.matrix(Xstar.half %*% Matrix(Theta) %*% Xstar.half)\n eigen.fit <- eigen(Theta, symmetric = T)\n efuncs <- Xstar.invhalf %*% eigen.fit$vectors\n }\n evals <- as.numeric(eigen.fit$values)\n evals[evals < 1e-5] <- 0\n return(matrix.multiply(efuncs, evals) %*% t(efuncs))\n}\n\n\n # sigsq <- l_mat$sigma2\n C.alt <- trun_mat(C)\n C.null <- trun_mat(C0)\n\n if (trunc.eig == 0) {\n Tn <- C.alt$Theta - C.null$Theta\n } else {\n Theta.alt <- trunc.mat.inner(C.alt$Theta, trunc.eig)\n Theta.null <- trunc.mat.inner(C.null$Theta, trunc.eig)\n Tn <- Theta.alt - Theta.null\n }\n\n Tn <- Tn %*% Xstar\n Tn <- sqrt(sum(Tn * t(Tn)))\n\n #########################\n #### step 2: Bootstrap\n #########################\n\n\n raw.C <- function(data) {\n y <- data$y\n subj <- data$subj\n subj_unique <- unique(subj)\n n <- length(subj_unique)\n C <- c()\n for (i in 1:n) {\n r1 <- y[subj == subj_unique[i]]\n m1 <- length(r1)\n if (m1 > 1) {\n C <- c(C, vech(tcrossprod(r1)))\n } ## for if(m1>1)\n if (m1 == 1) {\n C <- c(C, r1^2)\n }\n } ## for i\n if (off_diag) {\n C <- C * (sqrt(2) * (1 - delta) + delta)\n }\n C\n }\n\n bs.stats <- c()\n bs.success <- 0\n if (center.bs) {\n mean.bs <- fit_mean$fitted.values\n this.bs <- data\n } else {\n mean.bs <- 0\n }\n\n if (semi.iter) {\n C0.bs <- matrix(NA, nrow = length(C), ncol = 0)\n C.bs <- C0.bs\n y.bs <- resample(data, mean.bs, Rbar0.fit$coef.null, sigsq, L = nbs)\n\n while (bs.success < nbs) { #(0.15s)\n\n ###### a. generate Y_ij^(l) (0.01s)## fit_mean$fitted.values\n s_iterator <- bs.success + 1\n if (ncol(y.bs) < s_iterator) {\n y.bs <- cbind(y.bs, resample(data, mean.bs, Rbar0.fit$coef.null, sigsq, L = 5))\n }\n y <- y.bs[, s_iterator]\n\n ###### b. center Y_ij^(l) (0.01s)\n r <- y\n if (center && center.bs) {\n this.bs$y <- r\n if (gam.mgcv) {\n fit_mean.bs <- mgcv::gam(as.vector(y) ~ s(t, k = nb))\n } else {\n fit_mean.bs <- pspline(this.bs, argvals.new = tnew, knots = knots.initial, lambda = lambda_mean)\n }\n r <- y - fit_mean.bs$fitted.values\n }\n data.demean.bs <- data.frame(\n \"argvals\" = t,\n \"subj\" = subj, \"y\" = as.vector(r)\n )\n\n ###### c. Initialize C0^(l) (0.08s)\n fit.null.bs <- fitNull(data.demean.bs) # null fit\n if (\"try-error\" %in% class(fit.null.bs)) { # issue with null fit\n y.bs <- y.bs[, -s_iterator]\n next # if problem\n }\n Rbar0.fit.bs <- calc.R0(fit.null.bs, st)\n # C0.bs <- cbind(C0.bs, Rbar0.fit.bs$Rbar0)\n\n ###### d. Initialize C^(l) (0.05s)\n C.bs <- cbind(C.bs, Rbar0.fit.bs$Rbar0 - raw.C(data.demean.bs))\n bs.success <- bs.success + 1\n\n }\n# ptm <- proc.time()\n # Compute delta.bs (0.17 s)\n delta.bs <- m_est %*% Matrix(C.bs)\n# print(proc.time() - ptm)\n # compute bs statistics (0 s)\n bs.stats <- (Xstar %*% delta.bs) ^ 2\n bs.stats <- sqrt(colSums(bs.stats))\n\n } else {\n while (bs.success < nbs) { #(0.6s)\n#ptm <- proc.time()\n ###### a. generate Y_ij^(l) (0.01s)\n y <- c(resample(data, mean.bs, Rbar0.fit$coef.null, sigsq))\n\n ###### b. center Y_ij^(l) (0.01s)\n r <- y\n if (center && center.bs) {\n this.bs$y <- r\n if (gam.mgcv) {\n fit_mean.bs <- mgcv::gam(as.vector(y) ~ s(t, k = nb))\n } else {\n fit_mean.bs <- pspline(this.bs, argvals.new = tnew, knots = knots.initial, lambda = lambda_mean)\n }\n r <- y - fit_mean.bs$fitted.values\n }\n data.demean.bs <- data.frame(\n \"argvals\" = t,\n \"subj\" = subj, \"y\" = as.vector(r)\n )\n\n ###### c. Initialize C0^(l) (0.04s)\n fit.null.bs <- fitNull(data.demean.bs) # null fit\n if (\"try-error\" %in% class(fit.null.bs)) { # issue with null fit\n next # if problem\n }\n\n C0.bs <- calc.R0(fit.null.bs, st)$Rbar0\n\n ###### d. Initialize C^(l) (0.00s)\n C.bs <- raw.C(data.demean.bs) #(0.00s)\n\n\n\n\n ###### e. calculate estimated covariance function and test statistics\n # (0.00 s)\n\n if (trunc.eig == 0) {\n Tn.bs <- matrix(G %*% (m_est %*% (C.bs - C0.bs)), c)\n } else {\n Theta.alt <- trunc.mat.inner(matrix(G %*% (m_est %*% C.bs), c), trunc.eig)\n Theta.null <- trunc.mat.inner(matrix(G %*% (m_est %*% C0.bs), c), trunc.eig)\n Tn.bs <- Theta.alt - Theta.null\n }\n Tn.bs <- Tn.bs %*% Xstar\n Tn.bs <- sqrt(sum(Tn.bs * t(Tn.bs)))\n\n bs.stats <- c(bs.stats, Tn.bs) # save bs stats\n bs.success <- bs.success + 1\n#print(proc.time() - ptm)\n\n }\n }\n\n\n #########################\n #### step 3: P-values\n #########################\n p.bs <- function(stat, bs.stat) {\n p <- mean(stat <= bs.stat)\n list(p = p, mean = mean(bs.stat), var = var(bs.stat))\n }\n Tn.stats <- p.bs(Tn, unlist(bs.stats))\n\n\n list(\n mu = fit_mean$fitted.values,\n C.alt = C.alt$C,\n C.null = C.null$C,\n sigma2 = sigsq,\n Tn = Tn, p = Tn.stats$p, p.var = Tn.stats$var,\n bs.approx = bs.stats\n )\n}\n\n\n\n\n############################\n# Written by: Stephanie Chen (stchen3@ncsu.edu)\n# Purpose: Generate bootstrap resamples\n# Updated: Aug 4, 2018\n# Modified on July 30, 2021\n\nresample <- function(data, mu, coef.null, sigsq, L = 1) {\n nsubj <- length(unique(data$subj))\n cov.mat <- matrix(c(coef.null[1], coef.null[2], coef.null[2], coef.null[3]), nrow = 2)\n b.mat <- matrix(rnorm(2 * nsubj * L), ncol = 2) %*% chol(cov.mat) # random effects\n\n # r.slope & r.int by subject\n par.random <- sapply(seq_len(L), function(i) {\n s_offset <- (i-1) * nsubj\n sapply(seq_len(nsubj), function(x) {\n b.mat[s_offset + x, 1] + b.mat[s_offset + x, 2] * subset(data, subj == x)$argvals\n })\n })\n\n if (is.list(par.random)) { # if a list\n par.random <- unlist(par.random)\n }\n # mean.null + null.par r.int & r.slope*t + alt.par residual\n matrix(mu + par.random + rnorm(nrow(data) * L, sd = sqrt(sigsq)), ncol = L)\n}\n\n# resample(data, 0, c(1,0,1), 2, L = 2)\n# sapply(1:2, function(i) {\n# sapply(6:8, function(x) {\n# i*x\n# })\n# })\n\n\n", "meta": {"hexsha": "80c6c52b345b8702e9c4e86890d9675ed732eecc", "size": 16822, "ext": "r", "lang": "R", "max_stars_repo_path": "R/bootstrap.face.r", "max_stars_repo_name": "ZhuolinSong/Ftesting", "max_stars_repo_head_hexsha": "d0ecb36b44d377ab012a6325220f52110aeaf8d3", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/bootstrap.face.r", "max_issues_repo_name": "ZhuolinSong/Ftesting", "max_issues_repo_head_hexsha": "d0ecb36b44d377ab012a6325220f52110aeaf8d3", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/bootstrap.face.r", "max_forks_repo_name": "ZhuolinSong/Ftesting", "max_forks_repo_head_hexsha": "d0ecb36b44d377ab012a6325220f52110aeaf8d3", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 32.0419047619, "max_line_length": 124, "alphanum_fraction": 0.5464272976, "num_tokens": 5639, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8459424295406088, "lm_q2_score": 0.6406358617010351, "lm_q1q2_score": 0.5419410572982152}} {"text": "subroutine delout(delsum,nadj,madj,x,y,ntot,npd,nerror)\n\n# Put a summary of the Delaunay triangles with a vertex at point i,\n# for i = 1, ..., npd, into the array delsum. Do this in the original\n# order of the points, not the order into which they have been\n# bin-sorted.\n# Called by master.\n\nimplicit double precision(a-h,o-z)\ndimension nadj(-3:ntot,0:madj), x(-3:ntot), y(-3:ntot)\ndimension delsum(npd,4)\n\ndo i = 1,npd {\n area = 0. # Initialize area of polygon consisting of triangles\n # with a vertex at point i.\n # Get the coordinates of the point and the number of\n # (real) triangles emanating from it.\n np = nadj(i,0)\n\txi = x(i)\n\tyi = y(i)\n npt = np\n do k = 1,np {\n kp = k+1\n if(kp>np) kp = 1\n if(nadj(i,k)<=0|nadj(i,kp)<=0) npt = npt-1\n }\n\n # For each point in the adjacency list of point i, find its\n # successor, and the area of the triangle determined by these\n # three points.\n do j1 = 1,np {\n j = nadj(i,j1)\n if(j<=0) next\n\t\txj = x(j)\n\t\tyj = y(j)\n call succ(k,i,j,nadj,madj,ntot,nerror)\n\t\tif(nerror > 0) return\n if(k<=0) next\n\t\txk = x(k)\n\t\tyk = y(k)\n call triar(xi,yi,xj,yj,xk,yk,tmp)\n # Downweight the area by 1/3, since each\n # triangle eventually appears 3 times over.\n area = area+tmp/3.\n }\n\tdelsum(i,1) = xi\n\tdelsum(i,2) = yi\n\tdelsum(i,3) = npt\n\tdelsum(i,4) = area\n}\n\nreturn\nend\n", "meta": {"hexsha": "ec28f551c955417d9bb02817e90091d353c2e772", "size": 1568, "ext": "r", "lang": "R", "max_stars_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/delout.r", "max_stars_repo_name": "hyeongmokoo/SAAR_beta1", "max_stars_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2018-08-23T15:35:47.000Z", "max_stars_repo_stars_event_max_datetime": "2020-11-24T12:20:59.000Z", "max_issues_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/delout.r", "max_issues_repo_name": "hyeongmokoo/SAAR_beta1", "max_issues_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2020-08-17T15:14:11.000Z", "max_issues_repo_issues_event_max_datetime": "2020-09-23T21:55:49.000Z", "max_forks_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/delout.r", "max_forks_repo_name": "hyeongmokoo/SAAR_beta1", "max_forks_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-11-04T05:34:16.000Z", "max_forks_repo_forks_event_max_datetime": "2020-11-04T05:34:16.000Z", "avg_line_length": 29.037037037, "max_line_length": 72, "alphanum_fraction": 0.5427295918, "num_tokens": 487, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8056321889812554, "lm_q2_score": 0.672331705744791, "lm_q1q2_score": 0.5416520638206772}} {"text": "library(\"rgl\")\n\ncrossdot <- function(a, b, c) {\n\treturn (c[1]*(a[2]*b[3]-a[3]*b[2]) + c[2]*(a[3]*b[1]-a[1]*b[3]) + c[3]*(a[1]*b[2]-a[2]*b[1]) )\n}\n\ncross <- function(a, b) {\n\treturn (c( a[2]*b[3]-a[3]*b[2] , a[3]*b[1]-a[1]*b[3] , a[1]*b[2]-a[2]*b[1] ))\n}\n\narrow3d <- function(start, end, cl) {\n\tx <- (cross(end-start, c(0,0,1))/10)\n\tlines3d(c(start[1], end[1]), c(start[2], end[2]), c(start[3], end[3]), col=cl, lwd=3)\n\thead <- start+(end-start)*0.8\n\tlines3d(c(end[1], head[1]+x[1]), c(end[2], head[2]+x[2]), c(end[3], head[3]+x[3]), col=cl, lwd=3)\n\tlines3d(c(end[1], head[1]-x[1]), c(end[2], head[2]-x[2]), c(end[3], head[3]-x[3]), col=cl, lwd=3)\n\tlines3d(c(head[1]-x[1], head[1]+x[1]), c(head[2]-x[2], head[2]+x[2]), c(head[3]-x[3], head[3]+x[3]), col=cl, lwd=3)\n}\n\nif (rerun==0) {\n\tgas <- read.table(\"halo_gas.txt\", head=TRUE)\n\t#n <- intersect(which(gas$m>2e5), which(sqrt(gas$x^2 + gas$y^2 + gas$z^2)<0.01))\n\tn <- which(sqrt(gas$x^2 + gas$y^2 + gas$z^2)<0.04)\n\tgas <- gas[n,]\n\n\tstar <- read.table(\"halo_star.txt\", head=TRUE)\n\tn <- which(sqrt(star$x^2 + star$y^2 + star$z^2)<0.04)\n\tstar <- star[n,]\n\tn <- which(star$m > quantile(star$m, probs=0.8))\n\t#n <- sample(seq(length(star[,1])), 10000)\n\tstar <- star[n,]\n\n\tphi <- read.table(\"halophi.txt\", head=TRUE)\n\tphi$pot[which(is.nan(phi$pot))] <- 0\n\tphifunc <- splinefun(phi$r, phi$pot)\n\trealv <- sqrt(phifunc(phi$r, deriv=1)*phi$r)\n\n\tj0 <- c(0,0,0)\n\tfor (i in seq(length(gas$x))) {\n\t \tj0 <- j0 + cross(c(gas$x[i], gas$y[i], gas$z[i]), c(gas$vx[i], gas$vy[i], gas$vz[i]))\n\t}\n\tj0 <- j0/sqrt(sum(j0^2))\n\n\tratio <- numeric(length(gas$x))\n\tfor (i in seq(length(gas$x))) {\n\t\tradius <- sqrt(gas$x[i]^2 + gas$y[i]^2 + gas$z[i]^2)\n\t\tratio[i] <- crossdot(c(gas$x[i], gas$y[i], gas$z[i]), c(gas$vx[i], gas$vy[i], gas$vz[i]), j0)/radius\n\t\tradbin <- which.min(abs(phi$r - radius))\n\t\tratio[i] <- ratio[i]/realv[radbin]\n\t}\n\n\tpcols <- ratio\n\tpcols[which(is.infinite(pcols))] = 0\n\tpcols <- pcols/max(pcols)\n\tpcols[which(pcols<0)] = 0\n\n\tsratio <- numeric(length(star$x))\n\tfor (i in seq(length(star$x))) {\n\t\tradius <- sqrt(star$x[i]^2 + star$y[i]^2 + star$z[i]^2)\n\t\tsratio[i] <- crossdot(c(star$x[i], star$y[i], star$z[i]), c(star$vx[i], star$vy[i], star$vz[i]), j0)/radius\n\t\tradbin <- which.min(abs(phi$r - radius))\n\t\tsratio[i] <- sratio[i]/realv[radbin]\n\t}\n\n\tspcols <- sratio\n\tspcols[which(is.infinite(spcols))] = 0\n\tspcols <- spcols/max(spcols)\n\tspcols[which(spcols<0)] = 0\n\n\tforce <- read.table(\"forcemap.txt\", head=TRUE)\n}\n\nif (type==0) {\n\tn <- which(ratio>=1.0)\n\tplot3d(gas$x[n], gas$y[n], gas$z[n], col=rgb(pcols[n], 0, 0), xlim=c(-0.01, 0.01), ylim=c(-0.01, 0.01), zlim=c(-0.01, 0.01))\n\tn <- which(ratio<1.0)\n\tpoints3d(gas$x[n], gas$y[n], gas$z[n], col=rgb(0, 0, pcols[n]))\n}\n\nif (type==1) {\n\tn <- which(ratio>=1.0)\n\tplot3d(gas$x[n], gas$y[n], gas$z[n], col=rgb(pcols[n], 0, 0), xlim=c(-0.01, 0.01), ylim=c(-0.01, 0.01), zlim=c(-0.01, 0.01))\n\tn <- which(ratio<1.0)\n\tpoints3d(gas$x[n], gas$y[n], gas$z[n], col=rgb(0, 0, pcols[n]))\n\tarrow3d(-j0/500,j0/500, \"green\")\n\tfor (i in seq(length(force$x))) {\n\t\tfscale <- 4e10\n\t\tarrow3d(c(force$x[i], force$y[i], force$z[i]), c(force$x[i], force$y[i], force$z[i])+c(force$Fx[i], force$Fy[i], force$Fz[i])/fscale, \"black\")\n\t}\n}\n\nif (type==2) {\n\tn <- which(sratio>=1.0)\n\tplot3d(star$x[n], star$y[n], star$z[n], col=rgb(spcols[n], 0, 0), xlim=c(-0.01, 0.01), ylim=c(-0.01, 0.01), zlim=c(-0.01, 0.01))\n\tn <- which(sratio<1.0)\n\tpoints3d(star$x[n], star$y[n], star$z[n], col=rgb(0, 0, spcols[n]))\n\tarrow3d(-j0/500,j0/500, \"green\")\n}\n\nif (type==3) {\n\tn <- which(sratio>=1.0)\n\tplot3d(star$x[n], star$y[n], star$z[n], col=rgb(spcols[n], 0, 0), xlim=c(-0.01, 0.01), ylim=c(-0.01, 0.01), zlim=c(-0.01, 0.01))\n\tn <- which(sratio<1.0)\n\tpoints3d(star$x[n], star$y[n], star$z[n], col=rgb(0, 0, spcols[n]))\n\ttr <- read.table(\"trcurve.txt\", head=TRUE)\n\ttheta <- seq(0,2*pi,by=pi/100)\n\tfor (i in seq(length(tr$r)/2)) {\n\t\txhat <- cross(c(tr$Lx[i], tr$Ly[i], tr$Lz[i]), c(1,0,0))\n\t\tyhat <- cross(c(tr$Lx[i], tr$Ly[i], tr$Lz[i]), xhat)\n\t\txhat <- (xhat*tr$r[i])/sqrt(xhat[1]^2 + xhat[2]^2 + xhat[3]^2)\n\t\tyhat <- (yhat*tr$r[i])/sqrt(yhat[1]^2 + yhat[2]^2 + yhat[3]^2)\n\t\tlines3d(xhat[1]*cos(theta)+yhat[1]*sin(theta), xhat[2]*cos(theta)+yhat[2]*sin(theta), xhat[3]*cos(theta)+yhat[3]*sin(theta), col=\"green\", lwd=2)\n\t}\n}\n", "meta": {"hexsha": "a1c5808ef9ff8515bab665e67b96ccaededa9cd6", "size": 4275, "ext": "r", "lang": "R", "max_stars_repo_path": "visualise.r", "max_stars_repo_name": "petehague/galaxyview", "max_stars_repo_head_hexsha": "9202a09c97d66b23213356815f3c6eaeb8958d7f", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "visualise.r", "max_issues_repo_name": "petehague/galaxyview", "max_issues_repo_head_hexsha": "9202a09c97d66b23213356815f3c6eaeb8958d7f", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "visualise.r", "max_forks_repo_name": "petehague/galaxyview", "max_forks_repo_head_hexsha": "9202a09c97d66b23213356815f3c6eaeb8958d7f", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 37.1739130435, "max_line_length": 146, "alphanum_fraction": 0.5656140351, "num_tokens": 1909, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.919642526773001, "lm_q2_score": 0.588889130767832, "lm_q1q2_score": 0.5415674882084852}} {"text": "#autor: Joao Sollari Lopes\n#local: INE, Lisboa\n#Rversion: 3.6.1\n#criado: 10.07.2017\n#modificado: 06.12.2019\n\n#based on https://en.wikipedia.org/wiki/File:Local_maximum.png\nf <- function(x,y){\n exp(1)^(-(x^2+y^2))+2*exp(1)^(-((x-1.5)^2+(y-1.5)^2))\n}\nx <- seq(-2,4,len=50)\ny <- seq(-2,4,len=50)\nz <- outer(x,y,f)\n\nfnam <- \"../images/concept_3.tiff\"\ntiff(file=fnam,units=\"in\",width=5,height=4,res=300,compression=\"lzw\")\npar(mar=c(0.2,0.2,0.2,0.2))\npersp(x,y,z,phi=20,theta=30,d=5,expand=0.5,box=FALSE)\ndev.off()\n", "meta": {"hexsha": "4c6e15e2e1c698a6ef0ccdf000ff548d1be6f2fb", "size": 528, "ext": "r", "lang": "R", "max_stars_repo_path": "bin/pic_localmax.r", "max_stars_repo_name": "jsollari/DAA2019", "max_stars_repo_head_hexsha": "6b4e0764f6570bef4652a73a773f3b9ee8af92da", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "bin/pic_localmax.r", "max_issues_repo_name": "jsollari/DAA2019", "max_issues_repo_head_hexsha": "6b4e0764f6570bef4652a73a773f3b9ee8af92da", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "bin/pic_localmax.r", "max_forks_repo_name": "jsollari/DAA2019", "max_forks_repo_head_hexsha": "6b4e0764f6570bef4652a73a773f3b9ee8af92da", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-01-17T11:11:10.000Z", "max_forks_repo_forks_event_max_datetime": "2020-01-17T11:11:10.000Z", "avg_line_length": 26.4, "max_line_length": 69, "alphanum_fraction": 0.615530303, "num_tokens": 236, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.8311430394931455, "lm_q2_score": 0.6513548646660542, "lm_q1q2_score": 0.5413690620071907}} {"text": "### Calculate matrix using Plucinski's Unsupervized Naive Bayes classifier\r\n\r\n# calculate allele frequencies\r\n\r\n# calculate likelihood of no relation and likelihood of relation\r\n\r\ncalculate_loglikelihood = function(v1,v2,p1,p2,ploid){\r\n\t# p1 is vector of allele frequencies in sample 1\r\n\t# p2 is vector of allele frequencies in sample 2\r\n\tn1 = length(p1)\r\n\tn2 = length(p2)\r\n\tloglikelihood0 = sum(log(p1))+sum(log(p2))\r\n\tloglikelihood1 = log(sum(sum(sapply(1:n1, function (i) sapply(1:n2, \r\n\t\t\t\tfunction (j) 1/n1/n2 * (v1[i] == v2[j])*exp((sum(log(p1[-i]))+sum(log(p2)))))))))\r\n\tif (length(v1) > 1 & length(v2) > 1) {\r\n\tpairs1 = combn(v1,2,simplify = FALSE)\r\n\tpairs2 = combn(v2,2,simplify = FALSE)\r\n\tnpairs1 = length(pairs1)\r\n\tnpairs2 = length(pairs2)\r\n\r\n\tloglikelihood2 = log(sum(sum(sapply(1:npairs1, function (i) sapply(1:npairs2, \r\n\t\t\t\tfunction (j) 1/npairs1/npairs2 * (sum(sort(pairs1[[i]]) == sort(pairs2[[j]]))==2)*\r\n\t\t\t\t\t\t\texp((sum(log(p1[-match(pairs1[[i]],v1)]))+sum(log(p2)))))))))\r\n\t} else { loglikelihood2 = NA}\r\n\tif (ploid == 1) {\r\n\t\tloglikelihood2 = loglikelihood1 \r\n\t}\r\n\tc(loglikelihood0,loglikelihood1,loglikelihood2)\r\n}\r\n\r\n##### modify epsilon value here\r\n\t\t\t\t\t \r\nepsilon = 0.3072 #rate of missed alleles, estimated by Joel May 2 2019\r\n\r\ncalculate_loglikelihood2 = function(v1,v2,p1,p2,ploid){\r\n\t# p1 is vector of allele frequencies in sample 1\r\n\t# p2 is vector of allele frequencies in sample 2\r\n\tn1 = length(p1)\r\n\tn2 = length(p2)\r\n\tloglikelihood0 = sum(log(p1))+sum(log(p2))\r\n\tloglikelihood1 = log(max(sapply(1:n1, function (i) sapply(1:n2, \r\n\t\t\t\tfunction (j) (v1[i] == v2[j])*exp((sum(log(p1[-i]))+sum(log(p2)))))),na.rm=TRUE))\r\n\tif (length(v1) > 1 & length(v2) > 1) {\r\n\tpairs1 = combn(v1,2,simplify = FALSE)\r\n\tpairs2 = combn(v2,2,simplify = FALSE)\r\n\tnpairs1 = length(pairs1)\r\n\tnpairs2 = length(pairs2)\r\n\r\n\tloglikelihood2 = log(max(sapply(1:npairs1, function (i) sapply(1:npairs2, \r\n\t\t\t\tfunction (j) (sum(sort(pairs1[[i]]) == sort(pairs2[[j]]))==2)*\r\n\t\t\t\t\t\t\texp((sum(log(p1[-match(pairs1[[i]],v1)]))+sum(log(p2)))))),na.rm=TRUE))\r\n\t} else { loglikelihood2 = NA}\r\n\r\n\tif (loglikelihood1 == -Inf) {\r\n\t\tloglikelihood1 = log(epsilon * min(c(p1,p2)))\r\n\t}\r\n\tif (loglikelihood2 == -Inf & ploid > 1) {\r\n\t\tnsharedalleles = max(sapply(1:npairs1, function (i) sapply(1:npairs2, \r\n\t\t\t\t\t\tfunction (j) (length(intersect(pairs1[[i]], pairs2[[j]])))))) ## modified line - used to be: \r\n\t\t\t\t\t\t\t\t\t ## function (j) (sum(sort(pairs1[[i]]) == sort(pairs2[[j]]))))))\r\n\t\tloglikelihood2 = log((epsilon*min(c(p1,p2))) ^ (2-nsharedalleles))\r\n\t}\r\n\tif (ploid == 1) {\r\n\t\tloglikelihood2 = loglikelihood1 \r\n\t}\r\n\tc(loglikelihood0,loglikelihood1,loglikelihood2)\r\n}\r\n\r\nalleles = list()\r\nfrequencies = list()\r\n\r\nfor (j in 1:nloci) {\r\n\tlocicolumns = grepl(paste(locinames[j],\"\",sep=\"\"),colnames(data))\r\n\traw_alleles = c(as.matrix(data[,locicolumns]))\r\n\traw_alleles[raw_alleles == \"NA\"] = NA\r\n\traw_alleles[raw_alleles == 0] = NA\r\n\talleles[[j]] = unique(raw_alleles[!is.na(raw_alleles)])\r\n\tfrequencies[[j]] = sapply(alleles[[j]], function(x) sum(raw_alleles == x,na.rm=TRUE))\r\n\tfrequencies[[j]] = frequencies[[j]] / sum(frequencies[[j]])\r\n\r\n}\r\n\r\n\r\nobserveddatamatrix = list()\r\nfor (j in 1:nloci) {\r\n\tlocus = locinames[j]\r\n\tlocicolumns = grepl(paste(locus,\"\",sep=\"\"),colnames(data))\r\n\toldalleles = as.vector(data[,locicolumns])\r\n\toldalleles [oldalleles == \"NA\"] = NA\r\n\toldalleles [oldalleles == 0] = NA\r\n\tif (length(dim(oldalleles)[2]) == 0) {\r\n\t\toldalleles = matrix(oldalleles,length(oldalleles),1)\r\n\t}\r\n\tobserveddatamatrix[[j]] = oldalleles \r\n}\r\n\r\npairwisedistance = function(isolate1,isolate2){\r\n\tprint(((isolate2-1)*nids+isolate1)/ (nids*nids))\r\n\tprint(isolate1)\r\n\tprint(isolate2)\r\n\tloglik = matrix(NA,nloci,3)\r\n\tfor (j in 1:nloci) {\r\n\t\tv1 = observeddatamatrix[[j]][isolate1,]\r\n\t\tv1 = v1[!is.na(v1)]\r\n\t\tp1 = frequencies[[j]][match(v1,alleles[[j]])]\r\n\t\tv2 = observeddatamatrix[[j]][isolate2,]\r\n\t\tv2 = v2[!is.na(v2)]\r\n\t\tp2 = frequencies[[j]][match(v2,alleles[[j]])]\r\n\t\tif (length(v1) > 0 & length(v2) > 0) {\r\n\t\t\tloglik[j,] = calculate_loglikelihood2(v1,v2,p1,p2,ploidy[j])\r\n\t\t} else { loglik[j,] = c(NA,NA,NA)}\r\n\t}\r\n\tloglik_allloci = (colSums(matrix(loglik,ncol=3),na.rm=TRUE))\r\n\tlik_allloci = exp(loglik_allloci) / sum(exp(loglik_allloci))\r\n\tsum(lik_allloci*c(0,1,2))\r\n}\r\n\r\nallpossiblepairs = expand.grid(1:nids,1:nids)\r\nallpossiblepairs = unique(allpossiblepairs[allpossiblepairs[,1] <= allpossiblepairs[,2],])\r\n\r\n# pairwisedistancevector = unlist(lapply(1:dim(allpossiblepairs)[1], function (x) pairwisedistance(allpossiblepairs[x,1],allpossiblepairs[x,2]))) # not parallel\r\n\r\n\t\t\t\t \r\n\t\t\t\t ###### MODIFY NUMBER OF CORES BELOW - mc.cores=###\r\n\t\t\t\t \r\npairwisedistancevector = unlist(mclapply(1:dim(allpossiblepairs)[1], function (x) pairwisedistance(allpossiblepairs[x,1],allpossiblepairs[x,2]),mc.cores=12)) # parallel\r\n\r\npairwisedistancematrix = matrix(NA,nids,nids)\r\nsapply(1:dim(allpossiblepairs)[1], function (x) pairwisedistancematrix[allpossiblepairs[x,1],allpossiblepairs[x,2]] <<- pairwisedistancevector[x])\r\nsapply(1:dim(allpossiblepairs)[1], function (x) pairwisedistancematrix[allpossiblepairs[x,2],allpossiblepairs[x,1]] <<- pairwisedistancevector[x])\r\n\r\ncolnames(pairwisedistancematrix) = ids \r\nrownames(pairwisedistancematrix) = ids\r\n###write.csv(pairwisedistancematrix,\"pairwisedistancematrix_Bayesian.csv\")\r\n\r\nBayesian_pairwisedistancematrix = pairwisedistancematrix \r\n\r\n\r\npairwisedistancematrix\r\n\r\ncolv_clustering = rep(rgb(0,0,0), length(ids))\r\n\r\n\r\nprint(\"Calculation of Bayesian matrix complete\")\r\n", "meta": {"hexsha": "b91f6419d567d17418699844df530f4825b642d9", "size": 5538, "ext": "r", "lang": "R", "max_stars_repo_path": "euk_bayesian_fulldataset_V2.r", "max_stars_repo_name": "Joel-Barratt/Eukaryotyping", "max_stars_repo_head_hexsha": "d1f02d53b7b86627e22d71b513e3afa8d286c36a", "max_stars_repo_licenses": ["CC0-1.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "euk_bayesian_fulldataset_V2.r", "max_issues_repo_name": "Joel-Barratt/Eukaryotyping", "max_issues_repo_head_hexsha": "d1f02d53b7b86627e22d71b513e3afa8d286c36a", "max_issues_repo_licenses": ["CC0-1.0"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2020-03-24T22:42:36.000Z", "max_issues_repo_issues_event_max_datetime": "2020-03-24T22:42:36.000Z", "max_forks_repo_path": "euk_bayesian_fulldataset_V2.r", "max_forks_repo_name": "Joel-Barratt/Eukaryotyping", "max_forks_repo_head_hexsha": "d1f02d53b7b86627e22d71b513e3afa8d286c36a", "max_forks_repo_licenses": ["CC0-1.0"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-03-23T18:10:00.000Z", "max_forks_repo_forks_event_max_datetime": "2020-03-23T18:10:00.000Z", "avg_line_length": 38.1931034483, "max_line_length": 169, "alphanum_fraction": 0.6668472373, "num_tokens": 1852, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8221891305219504, "lm_q2_score": 0.6584174938590246, "lm_q1q2_score": 0.5413437067963931}} {"text": "# --------------------------------------------------------------------------------------\n# Programmer: Jason Thorpe\n# Date 02/24/2011\n# Language: R (Version 2.6.0)\n# Purpose: This package has functions for application of the PEB algorithm\n# Comments:\n# --------------------------------------------------------------------------------------\n\n#' Functions that apply the Parametric Empirical Bayes (PEB) algorithm\n#'\n#' Functions that apply the Parametric Empirical Bayes (PEB) algorithm.\n#' \n#' \\code{zpeb(x,[n,ybar,]...)} returnes the z-score for a marker value after applying the PEB algorithm\n#' \n#' \\code{ppeb(x,[n,ybar,]...)} returnes the p-values for a marker value after applying the PEB algorithm\n#' \n#' \\code{qpeb(p,n,ybar,...,conf.level)} returnes the quantiles for a marker if the PEB is \n#' to algorithm with a specificity \\code{p}. In particular, if the normality \n#' assumptions are satisified, the probability that a marker exceeds \n#' \\code{threshold = qpeb(p,n,ybar,...)} is \\code{p}. \n#' \n#' @name peb\n#' @family peb\n#' \n#' @param x A vector of marker values to be interpreted by the PEB. \\strong{Note}\n#' If n and ybar are not specified, x is interpreted to be a series of observed \n#' marker levels from a single individual\n#' \n#' @param n \\code{n[i]} is the number of prior observations from the same\n#' individual which preceed the i'th observation (\\code{x[i]}). Note that \n#' n and ybar are required paremeters for qpeb and for zpeb and ppeb, must \n#' both be present or both be absent.\n#' \n#' @param ybar \\code{n[i]} is the mean of prior observations from the same \n#' individual which preceed the i'th observation (\\code{x[i]}). Note that \n#' n and ybar are required paremeters for qpeb and for zpeb and ppeb, must \n#' both be present or both be absent.\n#' \n#' @param details If True, a data frame containing resutls from each \n#' sub-calculation is returned,otherwise a vector with the z-scores \n#' or p-values is returned. (Default = FALSE)\n#' \n#' @param na.rm If TRUE and the parameters \\code{n} and \\code{ybar} are not passed, then \n#' mising values in X are ignored when calculationg n and ybar from x. \\strong{Note}\n#' that if \\code{n} and \\code{ybar} are not passed and all values following the first missing \n#' value in x will also be missing. \n#' \n#' @param conf.level (optional) confidence level for the False Positve Rate (FPR)\n#' for the thresholds provided \\code{qpeb()}. If conf.level is specified, \\code{length(p)}\n#' must be 1 and the PEB parameters must be specified by the object returned by \n#' \\code{PEBparams()} using the \\code{'Iterative'} method, as in:\n#' \n#' \\code{params = PEBparams(x,id,method='Iterative',iterations=2);\n#' qpeb(p,params,conf.level=0.95)}\n#' \n#' @inheritParams pebVarArgs\n#' \n#' @note Note that each function requires just 2 of the 4 parameters v,sigma2,tau2, and beta (=beta1)\n#' \\strong{Note} also that mu,sigma2,tau2,v, and beta may be substitured by a the object returned by \\code{\\link{PEBparams}}\n#' \n#' @examples\n#' \n#' n = 10\n#' m = 200\n#' params <- PEBparams(x = rnorm(m*n) + rep(rnorm(m),each=n), id=gl(m,n,m*n))\n#' x <- rnorm(10) + 6\n#' plot(seq(length(x)),ppeb(x,params))\n#' \n#' # interesting facts about the PEB...\n#' b1 = .7\n#' \n#' # this should be uniformly distributed by definition\n#' hist(ppeb(rnorm(5000)*sqrt(1-b1),mu=0,v=1,beta=b1))\n#' \n#' # this should be positive on the outset. \n#' hist(ppeb(rnorm(5)*sqrt(1-b1) + 3,mu=0,v=1,beta=b1)) # 10/sqrt(1-b1) standard deviations above normal!!\n#' \n#' # but even this should be uniformly distributed in the long run\n#' hist(ppeb(rnorm(5000)*sqrt(1-b1) + 10,mu=0,v=1,beta=b1)) # 10/sqrt(1-b1) standard deviations above normal!!\n#' \n\nNULL\n\n\n#' @export\n#' @rdname peb\n#' @usage zpeb(x, ..., [n, ybar,] na.rm = FALSE, details = FALSE)\nzpeb<- function(x,...,n,ybar,na.rm=FALSE,details=FALSE) {\n\tif(missing(n) != missing(ybar))\n\t\tstop(\"parameters 'n' and 'ybar' must be either both present or both absent\")\n\n\tif(missing(n)){\n\t\tif(na.rm){\n\t\t\tx.na <- is.na(x)\n\t\t\tybar <- n <- rep(NA,length(x))\n\t\t\tif(any(!x.na)){\n\t\t\t\t# personal means\n\t\t\t\tybar[!x.na] <- ybarCalc(x[!x.na])\n\t\t\t\t# number of prior results\n\t\t\t\tn[!x.na] <- nCalc(x[!x.na])\n\t\t\t}\n\t\t}else{\n\t\t\t# personal means\n\t\t\tybar <- ybarCalc(x)\n\t\t\t# number of prior results\n\t\t\tn <- nCalc(x)\n\t\t}\n\t}\n\t\n\t# DO THE WORK\n\tout <- with(pebVarArgs(...),{\n\t\t\tbn <- bn(n)\n\t\t\tvn_hat <- sigma2 + (tau2*(1-bn))\n\t\t\tsdn_hat <- sqrt(vn_hat)\n\t\t\tyn_hat <- mu*(1-bn) + ybar*bn\n\t\t\tz <- (x - yn_hat)/sdn_hat\n\t\t\tdata.frame(x=x,\n\t\t\t\t n=n,\n\t\t\t\t ybar=ybar,\n\t\t\t\t bn=bn,\n\t\t\t\t z=z,\n\t\t\t\t mu_n=yn_hat,\n\t\t\t\t sd_n=sdn_hat,\n\t\t\t\t z=z)\n\t\t\t})\n\tif(details)\n\t\treturn(cbind(out \n\t\t\t\t\t,p=pnorm(out[,'z'])\n\t\t\t\t\t))\n\telse\n\t\treturn(out$z)\n}\n\n#' @export\n#' @usage ppeb(x, ..., [n, ybar,] na.rm = FALSE, details = FALSE)\n#' @rdname peb\nppeb<-function(...,details=FALSE) {\n\t# determine the probability of a sereies of marker values\n\tif(details)\n\t\treturn(zpeb(...,details = T))\n\telse\n\t\treturn(pnorm(zpeb(...,details = F)))\n}\n\n#-- \"\n#-- \n#-- source('r:/Urban_N/UrbanGrp/Data Analysis/JThorpe/RTools/PEB/R/peb-parameters.r')\n#-- source('r:/Urban_N/UrbanGrp/Data Analysis/JThorpe/RTools/PEB/R/peb-utils.r')\n#-- source('r:/Urban_N/UrbanGrp/Data Analysis/JThorpe/RTools/PEB/R/peb-functions.r')\n#-- \n#-- params_iter <- PEBparams(x,id,method='iterative',iterations=2)\n#-- params_icc <- PEBparams(x,id,method='ICC')\n#-- params\n#-- \n#-- pebVarArgs(params_iter)\n#-- pebVarArgs(params_icc )\n#-- \n#-- qpeb(p=0.95, params, n=3, ybar=.5,conf.level=0.95)\n#-- \n#-- \"\n\n#' @export\n#' @usage qpeb(p, ..., n, ybar, verbose = FALSE)\n#' @rdname peb\n#' @importFrom CompQuadForm davies\nqpeb<-function(p,...,n,ybar,conf.level,conf.method='davies', verbose = FALSE) {\n\n\tpebArgs <- pebVarArgs(...)\n\twith(pebArgs,{\n\t\t\t\tbn_fun <<- bn(n)\n\t\t\t\tbn <- bn(n)\n\t\t\t\tbn <<- bn\n\t\t\t\ts2 <<- sigma2 \n\t\t\t\tt2 <<- tau2 \n\t\t\t\tv_n <<- sigma2 + (tau2*(1-bn))\n\t\t\t\tsd_n <<- sqrt(sigma2 + (tau2*(1-bn)))\n\t\t\t\tmu_n <<- mu*(1-bn) + ybar*bn\n\t\t\t\t})\n\tif(length(p) == 1){\n\t\tthreshold <- qnorm(p,mean = mu_n,sd=sd_n)\n\t\tif(missing(conf.level))\n\t\t\treturn(threshold)\n\t\tparams <- list(...)[[1]]\n\t\tif((!inherits(params,'PEBparams')) || (params$method != 'iterative'))\n\t\t\tstop(\"If 'conf.level' specified, qpeb must be supplied with a iteragive PEB params object. (See notes in help file)\")\n\n\t\tUN <- unique(n)\n\t\tconf = PEB.conf.int(p,UN,conf.level,params)\n\t\tindx <- match(n,UN)\n \t\treturn(cbind(quantile=threshold,\n\t\t\t\t\t p.lower.CI = conf[,'p.lower.CI'][indx],\n\t\t\t\t\t p.upper.CI = conf[,'p.upper.CI'][indx]))\n\t}\n\n\tif(!missing(conf.level))\n\t\tstop(\"'conf.level' not supported when length(p) > 1\")\n\n\tout <- matrix(NA,length(n),length(p))\n\tfor(i in 1:length(p))\n\t\tout[,i] <- qnorm(p[i],mean = mu_n,sd=sd_n)\n\tdimnames(out) <- list(if(!is.null(names(n)))names(N) else seq(length(n)),\n\t\t\t\t\t\t paste('p =',p))\n\treturn(out)\n}\n\n\n# just a couple of utility functions so that the 'if(mising(n))){...}' block in zpeb() is not to ugly \nybarCalc <- function(x)\n\treturn(if(length(x) > 1)\n\t\t\t\tc(0,cumsum(x[-length(x)])/seq(length(x)-1))\n\t\t\telse 0\n\t\t\t)\n\nnCalc <- function(x)\n\treturn(if(length(x) > 1)\n\t\t\t\tc(0,seq(length(x)-1))\n\t\t\telse 0\n\t\t\t)\n\n\n#' Confidence intervals for personalized PEB thresholds \n#' \n#' Confidence intervals for personalized PEB thresholds.\n#' \n#' @export\n#' @family peb\n#' \n#' @param p probability (specificity) \n#' \n#' @param n Number of previous observations in the individual's history. If \\code{length(n) == 1}, n is expanded to \\code{seq(0,n)}.\n#' \n#' @param conf.level Confidence level of the interval.\n#' \n#' @param params A PEBparams object returned by \\code{PEBparams(...,method='iterative')}\n#' \n#' @examples \n#' PEB.conf.int(p=0.95, n=6, conf.level=0.9, params=params)\n\nPEB.conf.int <- function(p,n,conf.level,params){\n\tif((!inherits(params,'PEBparams')) || (params$method != 'iterative'))\n\t\tstop(\"If 'conf.level' specified, qpeb must be supplied with a iteragive PEB params object. (See notes in help file)\")\n\n\tif(length(n) == 1)\n\t\tn <- seq(0,n)\n\n\twith(pebVarArgs(params),\n\t{\n\t\tbn <<- bn <- bn(n)\n\t\tsd_n <<- sqrt(sigma2 + (tau2*(1-bn)))\n\t})\n\n\tthreshold <- qnorm(p,sd=sd_n)\n\tp.lower.CI <- \n\tp.upper.CI <- numeric(0)\n\n\twith(params,\n\t{\n\t\talpha <- 1 - conf.level\n\t\tcint <- c(alpha/2,1-alpha/2)\n\n\t\tFUN <- function(x,bn,prob){\n\t\t\tCompQuadForm::davies(x,\n\t\t\t\t lambda=c(s2/(N-m),t2*(1-bn)/nstar),\n\t\t\t\t h = c(N-m ,nstar),# degrees of freedom\n\n\t\t\t\t # sigma2 *DOES NOT* ACCOUNT FOR THE STANDARD ERROR IN THE ESTIMATE \n\t\t\t\t # OF THE INDIVIDUALS ESTIMATED MEAN WITH N PRIOR OBSERVATIONS\n\t\t\t\t # B/C THE PEB THRESHOLD IS SET AS THE SUM OF THE MEAN + THE SQRT \n\t\t\t\t # OF THE WEIGHTED AVERAGE OF TWO CHI-SQUARED VARIALBLES (S2 AND T2)\n\t\t\t\t # WHEREAS, THIS sigma2 PARAMETER IS IS ADDED TO THE WEIGHTED AVERAGE \n\t\t\t\t # *WITHIN* THE SQUARE ROOT...\n\t\t\t\t # sigma=(1-bn)*mu.se + bn*(sqrt(s2)/n),\n\n\t\t\t\t lim=10000,\n\t\t\t\t acc=0.0001)$Qq-prob\n\t\t}\n\n\t\tfor(i in 1:length(bn)){\n\t\t\t(INT <- (t2*(1-bn[i])*qchisq(p=cint,df = nstar)/(nstar)) + \n\t\t\t\t\t(s2*qchisq(p=cint,df = N-m )/(N-m )))\n\t\t\tsd.upper.ci <<- sqrt(uniroot(FUN,\n\t\t\t\t\t\t\t\t\t\t interval = INT,\n\t\t\t\t\t\t\t\t\t\t bn=bn[i],\n\t\t\t\t\t\t\t\t\t\t prob=1-alpha/2)$root)\n\t\t\tsd.lower.ci <<- sqrt(uniroot(FUN,\n\t\t\t\t\t\t\t\t\t\t interval = INT,\n\t\t\t\t\t\t\t\t\t\t bn=bn[i],\n\t\t\t\t\t\t\t\t\t\t prob=alpha/2)$root)\n\t\t\tp.lower.CI[i] <<- pnorm(threshold[i], sd=sd.lower.ci) \n\t\t\tp.upper.CI[i] <<- pnorm(threshold[i], sd=sd.upper.ci) \n\t\t}\n\t})\n\treturn(cbind(n, \n\t\t\t\t p.lower.CI = p.lower.CI, \n\t\t\t\t p.upper.CI = p.upper.CI))\n}\n\n", "meta": {"hexsha": "7c345940e037224eaba98a5a65498afb0eba60f3", "size": 9433, "ext": "r", "lang": "R", "max_stars_repo_path": "R/peb-functions.r", "max_stars_repo_name": "jdthorpe/PEB", "max_stars_repo_head_hexsha": "23c0f8d6a6e81d9d2e8e3a74c40027b74e97a97c", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/peb-functions.r", "max_issues_repo_name": "jdthorpe/PEB", "max_issues_repo_head_hexsha": "23c0f8d6a6e81d9d2e8e3a74c40027b74e97a97c", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/peb-functions.r", "max_forks_repo_name": "jdthorpe/PEB", "max_forks_repo_head_hexsha": "23c0f8d6a6e81d9d2e8e3a74c40027b74e97a97c", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 31.6543624161, "max_line_length": 132, "alphanum_fraction": 0.6186791053, "num_tokens": 3055, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7853085708384736, "lm_q2_score": 0.6893056167854461, "lm_q1q2_score": 0.5413176087887113}} {"text": "envelop <- function(x, np=3) {\n for (k in 1:np) {\n#\n# Fill minima\n#\n dm1 = x[1]\n dz = x[2]\n for (j in 2:(length(x)-1)) {\n dp1 = x[j+1]\n if (dm1 > dz & dp1 > dz)\n x[j] <- 0.5 * (dm1 + dp1)\n dm1 = dz\n dz = dp1\n }\n#\n# Smooth non-maxima\n#\n dm1 = x[1]\n dz = x[2]\n for (j in 2:(length(x)-1)) {\n dp1 = x[j+1]\n if (dm1 >= dz || dp1 >= dz)\n x[j] <- 0.25*dm1 + 0.5*dz + 0.25*dp1\n dm1 = dz\n dz = dp1\n }\n }\n x\n}\n\npolevne <- function(x, np=3) {\n for (k in 1:np) {\n #\n # Degrade maxima\n #\n dm1 = x[1]\n dz = x[2]\n for (j in 2:(length(x)-1)) {\n dp1 = x[j+1]\n if (dm1 < dz & dp1 < dz)\n x[j] <- 0.5 * (dm1 + dp1)\n dm1 = dz\n dz = dp1\n }\n #\n # Smooth non-maxima\n #\n dm1 = x[1]\n dz = x[2]\n for (j in 2:(length(x)-1)) {\n dp1 = x[j+1]\n if (dm1 >= dz || dp1 >= dz)\n x[j] <- 0.25*dm1 + 0.5*dz + 0.25*dp1\n dm1 = dz\n dz = dp1\n }\n }\n x\n}\n", "meta": {"hexsha": "aeb4ff1f17f2fb448703feb9258c3ac2232317eb", "size": 990, "ext": "r", "lang": "R", "max_stars_repo_path": "R/envelop.r", "max_stars_repo_name": "NemoursResearch/FormantTracking", "max_stars_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2021-09-01T14:22:23.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-25T03:47:04.000Z", "max_issues_repo_path": "R/envelop.r", "max_issues_repo_name": "NemoursResearch/FormantTracking", "max_issues_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/envelop.r", "max_forks_repo_name": "NemoursResearch/FormantTracking", "max_forks_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-08-31T18:20:28.000Z", "max_forks_repo_forks_event_max_datetime": "2021-08-31T18:20:28.000Z", "avg_line_length": 16.5, "max_line_length": 44, "alphanum_fraction": 0.3787878788, "num_tokens": 445, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8175744761936437, "lm_q2_score": 0.6619228891883799, "lm_q1q2_score": 0.541171259408773}} {"text": "### fimd8.R\n### R code from Van Buuren, S. (2012). \n###\t\tFlexible Imputation of Missing Data. \n###\t\tCRC/Chapman & Hall, Boca Raton, FL.\n### (c) 2012 Stef van Buuren, www.multiple-imputation.com\n### Version 1, 22mar2012\n### Version 2, 4nov2015 tested with mice 2.23\n### Tested with Mac OS X 10.7.3, R2.14-2, mice 2.12\n\nif (packageVersion(\"mice\")<'2.12') stop(\"This code requires mice 2.12.\")\n\nlibrary(\"mice\")\nlibrary(\"lattice\")\n\n### Section 8.1 Correcting for selective drop-out\n\ndata <- pops\npred <- pops.pred\nif (!is.data.frame(data)) stop(\"The code for section 8.1 requires access to the POPS data.\")\n\n\n### Section 8.1.4 A degenerate solution (TIME CONSUMING (30 MINUTES))\n\nimp1 <- mice(data, pred=pred, maxit=20, seed=51121)\n\n### Figure 8.2\n\ntp82 <- plot(imp1, c(\"a10u\",\"a10b\",\"adhd\"),col=mdc(5),lty=1:5)\nprint(tp82)\n\n\n### Section 8.1.5 A better solution\n\n## TIME CONSUMING (30 MINUTES)\npred2 <- pred\npred2[61:86,61:86] <- 0\nimp2 <- mice(data, pred=pred2, maxit=20, seed=51121)\n\n### Figure 8.3\n\ntp83 <- bwplot(imp2, iq+e_tot+l19_sd+b19_sd+coping+seffz~.imp,\n layout=c(2,3))\nprint(tp83)\n\n\n### Section 8.1.6 Results\n\n### Full responders\nsummary(lm(as.numeric(a10u)~1,data,na.action=na.omit))\nsummary(lm(as.numeric(a10b)~1,data,na.action=na.omit))\nsummary(lm(as.numeric(adhd)~1,data,na.action=na.omit))\n\n### All children\nsummary(pool(with(imp2,lm(as.numeric(a10u)~1))))\nsummary(pool(with(imp2,lm(as.numeric(a10b)~1))))\nsummary(pool(with(imp2,lm(as.numeric(adhd)~1))))\n\n\n### Section 8.2 Correcting for nonresponse\n\nlibrary(\"mice\")\ndata <- fdgs\n\n### Section 8.2.4 Augmenting the sample\n\nnimp <- c(400, 600, 75, 300, 200, 400)\nregcat <- c(\"North\",\"City\",\"North\",\"East\",\"North\",\"City\")\nreg <- rep(regcat, nimp)\n\nnimp2 <- floor(rep(nimp, each=2)/2)\nnimp2[5:6] <- c(38,37)\nsex <- rep(rep(c(\"boy\",\"girl\"),6), nimp2)\n\nminage <- rep(c(0, 0, 10, 10, 14, 14), nimp)\nmaxage <- rep(c(10, 10, 14, 14, 21, 21), nimp)\nset.seed(42444)\nage <- runif(length(minage), minage, maxage)\n\nid <- 600001:601975\n\npad <- data.frame(id, reg, age, sex, hgt=NA, wgt=NA, hgt.z=NA, wgt.z=NA)\ndata2 <- rbind(data, pad)\n\n\n### Figure 8.4\n\n## inspect the age by region pattern\nmeans <- aggregate(data$hgt.z, by=list(reg=data$reg, \n age=floor(data$age)), mean, na.rm=TRUE)\ntp84 <- xyplot(x~age, means, group=reg, type=c(\"g\",\"l\"),\n lty=1:5,col=mdc(4), \n xlim=c(-1,22), ylim=c(-0.6, 0.8), \n ylab=\"Height (SDS)\", xlab=\"Age (years)\",\n key=list(\n text=list(levels(means$reg)), \n lines=list(lty=1:5, col=mdc(4)), \n x=0.1, y=0.98, background=\"white\",\n columns=3, between.columns=0),\n scales=list(x=list(tck=c(1,0)), \n y=list(tck=c(1,0))))\nprint(tp84)\n\n\n### Section 8.2.5 Imputation model\n\n## add interaction terms\nna.opt <- options(na.action=na.pass)\nint <- model.matrix(~I(age-10)*hgt.z+I(age-10)*wgt.z+age*reg, data=data2)[,-(1:9)]\noptions(na.opt)\ndata3 <- cbind(data2, int)\n\n### define the imputation model\n\nini <- mice(data3, maxit=0)\n\nmeth <- ini$meth\nmeth[\"hgt\"] <- \"\"\nmeth[\"wgt\"] <- \"\"\nmeth[\"hgt.z\"] <- \"norm\"\nmeth[\"wgt.z\"] <- \"norm\"\nmeth[\"I(age - 10):hgt.z\"] <- \"~I(I(age-10)*hgt.z)\"\nmeth[\"I(age - 10):wgt.z\"] <- \"~I(I(age-10)*wgt.z)\"\n\npred <- ini$pred\npred[,c(\"hgt\",\"wgt\")] <- 0\npred[\"hgt.z\", c(\"id\",\"I(age - 10):hgt.z\")] <- 0\npred[\"wgt.z\", c(\"id\",\"I(age - 10):wgt.z\")] <- 0\n\nvis <- ini$vis[c(3,5,4,6)]\n\n### run the imputation model\n\nimp <- mice(data3, meth=meth, pred=pred, vis=vis, m=10, maxit=20, seed=28107)\n\n### Figure 8.5\n\ncda <- complete(imp, \"long\", include=TRUE)\nmeans2 <- aggregate(cda$hgt.z, by=list(reg=cda$reg, age=floor(cda$age), imp=cda$.imp), mean, na.rm=TRUE)\ntp85 <- xyplot(x~age|reg,means2, group=imp, subset=(reg==\"North\"|reg==\"City\"),\n type=c(\"g\",\"l\"),lwd=c(4,rep(1,imp$m)),\n lty=1:5,col=c(mdc(4),rep(mdc(6),imp$m)), \n ylab=\"Height (SDS)\", xlab=\"Age (years)\",\n ylim=c(-0.5,0.8), xlim=c(-2,23),\n scales=list(x=list(alternating=FALSE, tck=c(1,0)),\n y=list(tck=c(1,0))),\n strip = strip.custom(bg=\"grey95\"))\nprint(tp85)\n\n", "meta": {"hexsha": "f4b487fa1e80a193a0a195cfff66a08305dcadd3", "size": 4146, "ext": "r", "lang": "R", "max_stars_repo_path": "packrat/lib/x86_64-pc-linux-gnu/3.2.5/mice/doc/fimd8.r", "max_stars_repo_name": "Chicago-R-User-Group/2017-n3-Meetup-RStudio", "max_stars_repo_head_hexsha": "71a3204412c7573af2d233208147780d313430af", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "packrat/lib/x86_64-pc-linux-gnu/3.2.5/mice/doc/fimd8.r", "max_issues_repo_name": "Chicago-R-User-Group/2017-n3-Meetup-RStudio", "max_issues_repo_head_hexsha": "71a3204412c7573af2d233208147780d313430af", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2021-11-12T14:06:52.000Z", "max_issues_repo_issues_event_max_datetime": "2021-12-10T23:26:27.000Z", "max_forks_repo_path": "packrat/lib/x86_64-pc-linux-gnu/3.2.5/mice/doc/fimd8.r", "max_forks_repo_name": "Chicago-R-User-Group/2017-n3-Meetup-RStudio", "max_forks_repo_head_hexsha": "71a3204412c7573af2d233208147780d313430af", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 27.8255033557, "max_line_length": 104, "alphanum_fraction": 0.5984081042, "num_tokens": 1521, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8128673087708699, "lm_q2_score": 0.665410558746814, "lm_q1q2_score": 0.5408904901162435}} {"text": "H02BBJ Example Program Results\n\n\n *** IP solver\n\n Parameters\n ----------\n\n Linear constraints...... 3 First integer solution.. OFF\n Variables............... 2 Max depth of the tree... 4\n\n Feasibility tolerance... 1.05E-08 Print level............. 10\n Infinite bound size..... 1.00E+20 EPS (machine precision). 1.11E-16\n\n Integer feasibility tol. 1.00E-05 Iteration limit......... 50\n Max number of nodes..... NONE\n\n ** Workspace provided with MAXDPT = 4: LRWORK = 84 LIWORK = 137\n ** Workspace required with MAXDPT = 4: LRWORK = 84 LIWORK = 137\n\n\n *** Optimum LP solution *** -17.50000\n\n\n Varbl State Value Lower Bound Upper Bound Lagr Mult Residual\n\n V 1 FR 3.92857 0.00000 None 0.000 3.929\n V 2 FR 1.42857 0.00000 None 0.000 1.429\n\n\n L Con State Value Lower Bound Upper Bound Lagr Mult Residual\n\n L 1 UL 15.0000 None 15.0000 -1.000 0.000\n L 2 UL 5.00000 None 5.00000 -0.5000 -8.8818E-16\n L 3 FR 14.6429 5.00000 None 0.000 9.643\n\n\n *** Start of tree search ***\n\n\n Node Parent Obj Varbl Value Lower Upper Value Depth\n No Node Value Chosen Before Bound Bound After\n 2 1 No Feas Soln 1 3.93 4.00 None 4.00 1\n 3 1 -16.2 1 3.93 0.00 3.00 3.00 1\n 4 3 -15.5 2 1.80 2.00 None 2.00 2\n 5 3 -13.0 2 1.80 0.00 1.00 1.00 2\n *** Integer solution ***\n\n\n Node Parent Obj Varbl Value Lower Upper Value Depth\n No Node Value Chosen Before Bound Bound After\n 6 4 No Feas Soln 1 2.50 3.00 3.00 3.00 3\n 7 4 -14.8 1 2.50 0.00 2.00 2.00 3\n 8 7 -12.0 CO 2 2.20 3.00 None 3.00 4\n 9 7 -14.0 2 2.20 2.00 2.00 2.00 4\n *** Integer solution ***\n\n *** End of tree search ***\n\n\n Total of 9 nodes investigated.\n\n Exit IP solver - Optimum IP solution found.\n\n Final IP objective value = -14.00000\n\n\n\n Varbl State Value Lower Bound Upper Bound Lagr Mult Residual\n\n V 1 UL 2.00000 0.00000 2.00000 -3.000 0.000\n V 2 EQ 2.00000 2.00000 2.00000 -4.000 0.000\n\n\n L Con State Value Lower Bound Upper Bound Lagr Mult Residual\n\n L 1 FR 14.0000 None 15.0000 0.000 1.000\n L 2 FR 0.00000 None 5.00000 0.000 5.000\n L 3 FR 10.0000 5.00000 None 0.000 5.000\n", "meta": {"hexsha": "64136ef93fe66d060bbb8d972a38cc795d437526", "size": 2987, "ext": "r", "lang": "R", "max_stars_repo_path": "simple_examples/baseresults/h02bbje.r", "max_stars_repo_name": "numericalalgorithmsgroup/NAGJavaExamples", "max_stars_repo_head_hexsha": "f625b3f043c5c14a88d7ecbc04374acf75d63c82", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2021-07-03T22:53:20.000Z", "max_stars_repo_stars_event_max_datetime": "2021-08-04T01:44:03.000Z", "max_issues_repo_path": "simple_examples/baseresults/h02bbje.r", "max_issues_repo_name": "numericalalgorithmsgroup/NAGJavaExamples", "max_issues_repo_head_hexsha": "f625b3f043c5c14a88d7ecbc04374acf75d63c82", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "simple_examples/baseresults/h02bbje.r", "max_forks_repo_name": "numericalalgorithmsgroup/NAGJavaExamples", "max_forks_repo_head_hexsha": "f625b3f043c5c14a88d7ecbc04374acf75d63c82", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2021-07-03T22:55:16.000Z", "max_forks_repo_forks_event_max_datetime": "2022-02-02T01:00:53.000Z", "avg_line_length": 37.3375, "max_line_length": 80, "alphanum_fraction": 0.4519584868, "num_tokens": 1021, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7662936430859597, "lm_q2_score": 0.7057850154599562, "lm_q1q2_score": 0.5408385707322902}} {"text": "rmnl <- function(Data,Z,R=1000,keep=1,ss.b=.4){\n\n ## load libraries\n library(bayesm)\n\n################################################################################\n\n ## internal functions\n\n rmultiregG = function (Y, X, Bbar, A, nu, V) {\n n = nrow(Y)\n m = ncol(Y)\n k = ncol(X)\n RA = chol(A)\n W = rbind(X, RA)\n Z = rbind(Y, RA %*% Bbar)\n IR = backsolve(chol(crossprod(W)), diag(k))\n Btilde = crossprod(t(IR)) %*% crossprod(W, Z)\n S = crossprod(Z - W %*% Btilde)\n rwout = rwishart(nu + n, chol2inv(chol(V + S)))\n B = Btilde + IR %*% matrix(rnorm(m * k), ncol = m) %*% t(rwout$CI)\n return(list(B = B, Sigma = rwout$IW))\n }\n################################################################################\n\n ## Unpack Data\n X = Data[[1]]$X[,,1] # Design differes for most respondents respondents: array(K,A,J)\n Y = Data[[1]]$Y # J*K matrix of ranked respones\n\n################################################################################\n\n ## define constants\n J = length(Y) #number of choice taskts\n K1 = nrow(X) #number of alternatives per choice set\n A = ncol(X) #total number of attributes\n N = length(Data) #number of respondents\n Sig1 = IMat1 = diag(K1)\n sigi1=chol2inv(chol(Sig1))\n\n #ss.b = .07 \n ssB = double(N)\n ssB = ssB + ss.b\n\n################################################################################\n\n ## allocate storage space\n betadraw = array(double((R/keep)*N*A),dim=c(R/keep,A,N))\n bbardraw = matrix(double((R/keep)*A),ncol=A)\n vbetakeep = matrix(double(A*A*R/keep),ncol=A*A)\n nacceptB = matrix(double(N),ncol=1)\n lltemp = matrix(double(N),ncol=1)\n llkeep = matrix(double(N*R/keep),ncol=N)\n\n \n################################################################################\n ## Starting values for beta\n \n ## Uniform [-.01,.01]\n betaM = betaMnew = matrix(runif(N*A,-.01,.01),ncol=A) \n\n ## Initialize variables\n beta = c(double(A))\n bbar = c(double(A))\n ztemp = c(1:K1)\n LB = -50\n UB = 50\n iota = matrix(1,ncol=1,nrow=N)\n\n ## Prior values \n bbar.p = c(double(A)) + 2\n nu = A+3\n A.b = 0.01\n V.b = Sig.b = diag(A)*nu #*.1 pp 74 in text\n rooti.b = backsolve(chol(V.b*5),diag(A))\n\n ## mcmc variables starting values\n llold.b = 0\n pold.b = lndMvn(beta,bbar,rooti.b)\n\n################################################################################\n\n ## Estimation Algorithm\n\n cat(\" \", fill = TRUE)\n cat(\"Starting Estimation Routine for Hierarchical MNL:\", fill = TRUE)\n itime = proc.time()[3]\n cat(\"MCMC Iteration (est time to end - min) \", fill = TRUE)\n flush.console()\n for(rep in 1:R){\n\n ## Unit Loop \n for(nhh in 1:N){\n\n################################################################################\n\n ## unpack data for household nhh\n X1.nhh = Data[[nhh]]$X\n Y1 = Data[[nhh]]$Y\n betaold = betaM[nhh,]\n \n llold = lltemp[nhh,]\n ss.b = ssB[nhh]\n\n################################################################################\n ## Step 1. Draw beta|else - via M-H (evaluate likelihood using GHK)\n\n if(rep==1){\n ## Compute current value of likelihood & prior\n llold = 0\n \n ## Compute for first set of choices (4 alt)\n for(ntask in 1:J){\n Y = Data[[nhh]]$Y[ntask]\n X = X1.nhh[,,ntask]\n mu = X%*%betaold\n ltemp = exp(mu)/sum(exp(mu))\n llold = llold + log(ltemp[Y])\n }\n \n }\n\n pold.b = lndMvn(betaold,as.vector(bbar),rooti.b)\n\n ## draw betanew\n betanew = betaold + rnorm(length(betaold),0,ss.b)\n\n ## Compute likelihood & prior for new values of beta\n llnew = 0 \n for(ntask in 1:J){\n Y = Data[[nhh]]$Y[ntask]\n X = X1.nhh[,,ntask]\n mu = X%*%betanew\n ltemp = exp(mu)/sum(exp(mu))\n llnew = llnew + log(ltemp[Y])\n }\n \n pnew.b = lndMvn(betanew,as.vector(bbar),rooti.b)\n\n ## metropolis step\n ldiff = llnew + pnew.b - llold - pold.b\n\n alpha = min(1,exp(ldiff))\n if (alpha<1) {unif=runif(1)} else {unif=0}\n\n ## update variables and save draws if accepted\n if (unif <= alpha)\n {\n llold = llnew\n betaold = betanew\n nacceptB[nhh,] = nacceptB[nhh,] + 1 } else { }\n\n betaM[nhh,] = betaold\n betaMnew[nhh,] = betanew\n\n lltemp[nhh,] = llold\n\n } ## end unit loop\n\n################################################################################\n ## Step 4. Draw hierarchical priors\n if(N>1){\n\n ## Distribution of heterogeneity for beta\n rmout = rmultiregG(betaM,iota,bbar.p,A.b,nu,V.b)\n bbar = rmout$B\n Sig.b = rmout$Sigma\n rooti.b = backsolve(chol(rmout$Sigma),diag(A))\n\n } ## End N>1 condition\n\n################################################################################\n ## Print Time & Parameter Estimates\n if (rep%%100 == 0) {\n ctime = proc.time()[3]\n timetoend = ((ctime-itime)/rep) * (R-rep)\n cat(\" \", rep, \" \", round(timetoend/60,1), \" \", round(bbar,2),\n \" \", round(mean(nacceptB/10),2),\" \",\n sum(lltemp), \" \", fill = TRUE)\n flush.console()\n }\n \n################################################################################\n ## Save Draws\n if (rep%%keep == 0){\n betadraw[rep/keep,,] = t(betaM) \n bbardraw[rep/keep,] = bbar\n vbetakeep[rep/keep,] = as.vector(rmout$Sigma)\n llkeep[rep/keep,] = as.vector(lltemp)\n }\n\n ## Dynamically update step size\n \n if(rep%%10 == 0){\n if(rep .4) ssB[ii] = ssB[ii]*1.1\n }\n }\n nacceptB = nacceptB*0\n }\n \n } ## end MCMC\n\n return(list(betadraw=betadraw, bbardraw=bbardraw, vbetadraw=vbetakeep,\n llkeep = llkeep))\n}\n\n\n", "meta": {"hexsha": "5d7d21c07f24499304060eb190ce01f61372e512", "size": 5771, "ext": "r", "lang": "R", "max_stars_repo_path": "Code/rhierMNL.r", "max_stars_repo_name": "jeff-dotson/mobile-conjoint", "max_stars_repo_head_hexsha": "361733af7978ed0707aca902f9f005637ffade0c", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Code/rhierMNL.r", "max_issues_repo_name": "jeff-dotson/mobile-conjoint", "max_issues_repo_head_hexsha": "361733af7978ed0707aca902f9f005637ffade0c", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Code/rhierMNL.r", "max_forks_repo_name": "jeff-dotson/mobile-conjoint", "max_forks_repo_head_hexsha": "361733af7978ed0707aca902f9f005637ffade0c", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 27.480952381, "max_line_length": 87, "alphanum_fraction": 0.4713221279, "num_tokens": 1747, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8633916029436189, "lm_q2_score": 0.6261241632752915, "lm_q1q2_score": 0.5405903449719861}} {"text": "#' Computes estimates and ancillary information for diagonal classifiers\n#'\n#' Computes the maximum likelihood estimators (MLEs) for each class under the\n#' assumption of multivariate normality for each class. Also, computes ancillary\n#' information necessary for classifier summary, such as sample size, the number\n#' of features, etc.\n#'\n#' This function computes the common estimates and ancillary information used in\n#' all of the diagonal classifiers in the `sparsediscrim` package.\n#'\n#' The matrix of training observations are given in `x`. The rows of `x`\n#' contain the sample observations, and the columns contain the features for each\n#' training observation.\n#'\n#' The vector of class labels given in `y` are coerced to a `factor`.\n#' The length of `y` should match the number of rows in `x`.\n#'\n#' An error is thrown if a given class has less than 2 observations because the\n#' variance for each feature within a class cannot be estimated with less than 2\n#' observations. If other data have zero variances, these will be removed with\n#' a warning. \n#'\n#' The vector, `prior`, contains the _a priori_ class membership for\n#' each class. If `prior` is NULL (default), the class membership\n#' probabilities are estimated as the sample proportion of observations belonging\n#' to each class. Otherwise, `prior` should be a vector with the same length\n#' as the number of classes in `y`. The `prior` probabilities should be\n#' nonnegative and sum to one.\n#'\n#' @inheritParams lda_diag\n#' @param pool logical value. If TRUE, calculates the pooled sample variances\n#' for each class.\n#' @param est_mean the estimator for the class means. By default, we use the\n#' maximum likelihood estimator (MLE). To improve the estimation, we provide the\n#' option to use a shrunken mean estimator proposed by Tong et al. (2012).\n#' @return named list with estimators for each class and necessary ancillary\n#' information\n#'\n#' @references Tong, T., Chen, L., and Zhao, H. (2012), \"Improved Mean\n#' Estimation and Its Application to Diagonal Discriminant Analysis,\"\n#' Bioinformatics, 28, 4, 531-537.\n#' \\url{http://bioinformatics.oxfordjournals.org/content/28/4/531.long}\ndiag_estimates <- function(x, y, prior = NULL, pool = FALSE,\n est_mean = c(\"mle\", \"tong\")) {\n obj <- list()\n obj$labels <- y\n obj$N <- length(y)\n obj$p <- ncol(x)\n obj$groups <- levels(y)\n obj$num_groups <- nlevels(y)\n\n est_mean <- match.arg(est_mean)\n\n # Error Checking\n if (!is.null(prior)) {\n if (length(prior) != obj$num_groups) {\n rlang::abort(\"The number of 'prior' probabilities must match the number of classes in 'y'.\")\n }\n if (any(prior <= 0)) {\n rlang::abort(\"The 'prior' probabilities must be nonnegative.\")\n }\n if (sum(prior) != 1) {\n rlang::abort(\"The 'prior' probabilities must sum to one.\")\n }\n }\n if (any(table(y) < 2)) {\n rlang::abort(\"There must be at least 2 observations in each class.\")\n }\n\n # By default, we estimate the 'a priori' probabilities of class membership with\n # the MLEs (the sample proportions).\n if (is.null(prior)) {\n prior <- as.vector(table(y) / length(y))\n }\n\n # For each class, we calculate the MLEs (or specified alternative estimators)\n # for each parameter used in the DLDA classifier. The 'est' list contains the\n # estimators for each class.\n obj$est <- tapply(seq_along(y), y, function(i) {\n stats <- list()\n stats$n <- length(i)\n if (est_mean == \"mle\") {\n stats$xbar <- colMeans(x[i, , drop = FALSE])\n } else if (est_mean == \"tong\") {\n stats$xbar <- tong_mean_shrinkage(x[i, , drop = FALSE])\n }\n stats$var <- with(stats, (n - 1) / n * apply(x[i, , drop = FALSE], 2, var))\n stats\n })\n \n # Check to see if any predictors had zero variances\n obj$est <- check_for_zero_vars(obj$est)\n\n # Calculates the pooled variance across all classes.\n if (pool) {\n obj$var_pool <- Reduce('+', lapply(obj$est, function(x) x$n * x$var)) / obj$N\n }\n\n # Add each element in 'prior' to the corresponding obj$est$prior\n for(k in seq_len(obj$num_groups)) {\n obj$est[[k]]$prior <- prior[k]\n }\n obj\n}\n\n\ncheck_for_zero_vars <- function(x, warn = TRUE) {\n var_est <- lapply(x, function(x) x$var == 0)\n is_zv <- do.call(\"rbind\", var_est)\n any_zv <- apply(is_zv, 2, any)\n if (all(any_zv)) {\n rlang::abort(\"All predictors have zero variance.\")\n }\n if (any(any_zv)) {\n if (warn) {\n nms <- paste0(names(any_zv)[any_zv], collapse = \", \")\n nms <- paste(\"The following predictors had zero variance (possibly within \",\n \"a class) and were removed from the analysis:\", nms)\n rlang::warn(nms)\n }\n x <- lapply(x, reduce_elem, retain = names(any_zv)[!any_zv], \"xbar\")\n x <- lapply(x, reduce_elem, retain = names(any_zv)[!any_zv], \"var\")\n }\n x\n}\n\nreduce_elem <- function(x, retain, col) {\n x[[col]] <- x[[col]][retain]\n x\n}\n\n\n#' Computes estimates and ancillary information for regularized discriminant\n#' classifiers\n#'\n#' Computes the maximum likelihood estimators (MLEs) for each class under the\n#' assumption of multivariate normality for each class. Also, computes ancillary\n#' information necessary for classifier summary, such as sample size, the number\n#' of features, etc.\n#'\n#' This function computes the common estimates and ancillary information used in\n#' all of the regularized discriminant classifiers in the `sparsediscrim`\n#' package.\n#'\n#' The matrix of training observations are given in `x`. The rows of `x`\n#' contain the sample observations, and the columns contain the features for each\n#' training observation.\n#'\n#' The vector of class labels given in `y` are coerced to a `factor`.\n#' The length of `y` should match the number of rows in `x`.\n#'\n#' An error is thrown if a given class has less than 2 observations because the\n#' variance for each feature within a class cannot be estimated with less than 2\n#' observations.\n#'\n#' The vector, `prior`, contains the _a priori_ class membership for\n#' each class. If `prior` is NULL (default), the class membership\n#' probabilities are estimated as the sample proportion of observations belonging\n#' to each class. Otherwise, `prior` should be a vector with the same length\n#' as the number of classes in `y`. The `prior` probabilities should be\n#' nonnegative and sum to one.\n#'\n#' @inheritParams lda_diag\n#' @param y vector of class labels for each training observation\n#' @param cov logical. Should the sample covariance matrices be computed?\n#' (Default: yes)\n#' @param prior vector with prior probabilities for each class. If NULL\n#' (default), then the sample proportions are used. See details.\n#' @return named list with estimators for each class and necessary ancillary\n#' information\nregdiscrim_estimates <- function(x, y, cov = TRUE, prior = NULL) {\n obj <- list()\n obj$labels <- y\n obj$N <- length(y)\n obj$p <- ncol(x)\n obj$groups <- levels(y)\n obj$num_groups <- nlevels(y)\n\n # Error Checking\n if (!is.null(prior)) {\n if (length(prior) != obj$num_groups) {\n rlang::abort(\"The number of 'prior' probabilities must match the number of\n classes in 'y'.\")\n }\n if (any(prior <= 0)) {\n rlang::abort(\"The 'prior' probabilities must be nonnegative.\")\n }\n if (sum(prior) != 1) {\n rlang::abort(\"The 'prior' probabilities must sum to one.\")\n }\n }\n if (any(table(y) < 2)) {\n rlang::abort(\"There must be at least 2 observations in each class.\")\n }\n\n # By default, we estimate the 'a priori' probabilities of class membership with\n # the MLEs (the sample proportions).\n if (is.null(prior)) {\n prior <- as.vector(table(y) / length(y))\n }\n\n # For each class, we calculate the MLEs for each parameter used in the\n # 'regdiscrim' classifiers. The 'est' list contains the estimators for each\n # class.\n obj$est <- tapply(seq_along(y), y, function(i) {\n stats <- list()\n stats$n <- length(i)\n stats$xbar <- as.vector(colMeans(x[i, , drop = FALSE]))\n if (cov) {\n stats$cov <- with(stats, cov_mle(x[i, , drop = FALSE]))\n }\n stats\n })\n\n # Calculates the pooled sample covariance matrix.\n if (cov) {\n obj$cov_pool <- Reduce('+', lapply(obj$est, function(x) x$n * x$cov)) / obj$N\n }\n\n # Add each element in 'prior' to the corresponding obj$est$prior\n for(k in seq_len(obj$num_groups)) {\n obj$est[[k]]$prior <- prior[k]\n }\n obj\n}\n", "meta": {"hexsha": "d6710bb95b9e5f032a8f876d7144398505692c4b", "size": 8358, "ext": "r", "lang": "R", "max_stars_repo_path": "R/estimates.r", "max_stars_repo_name": "topepo/sparsediscrim", "max_stars_repo_head_hexsha": "60198a54e0ced0afa3909121eea55321dd04c56f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2020-11-16T08:13:49.000Z", "max_stars_repo_stars_event_max_datetime": "2021-11-28T21:44:00.000Z", "max_issues_repo_path": "R/estimates.r", "max_issues_repo_name": "topepo/sparsediscrim", "max_issues_repo_head_hexsha": "60198a54e0ced0afa3909121eea55321dd04c56f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2021-05-26T12:02:17.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-10T03:00:06.000Z", "max_forks_repo_path": "R/estimates.r", "max_forks_repo_name": "topepo/sparsediscrim", "max_forks_repo_head_hexsha": "60198a54e0ced0afa3909121eea55321dd04c56f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.982300885, "max_line_length": 98, "alphanum_fraction": 0.6776740847, "num_tokens": 2225, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929104825007, "lm_q2_score": 0.6926419831347361, "lm_q1q2_score": 0.5402558363476341}} {"text": "\\name{extreme_deconvolution}\n\\alias{extreme_deconvolution}\n\\title{Density estimation using Gaussian mixtures in the presence of noisy, heterogeneous and incomplete data}\n\\description{We present a general algorithm to infer a d-dimensional distribution function given a set of heterogeneous, noisy observations or samples. This algorithm reconstructs the error-deconvolved or 'underlying' distribution function common to all samples, even when the individual samples have unique error and missing-data properties. The underlying distribution is modeled as a mixture of Gaussians, which is completely general. Model parameters are chosen to optimize a justified, scalar objective function: the logarithm of the probability of the data under the error-convolved model, where the error convolution is different for each data point. Optimization is performed by an Expectation Maximization (EM) algorithm, extended by a regularization technique and 'split-and-merge' procedure. These extensions mitigate problems with singularities and local maxima, which are often encountered when using the EM algorithm to estimate Gaussian density mixtures.}\n\n\\usage{\nextreme_deconvolution(ydata,ycovar,\n xamp,xmean,xcovar,\n projection=NULL,\n weight=NULL,\n fixamp=NULL,fixmean=NULL,fixcovar=NULL,\n tol=1.e-6,maxiter=1e9,w=0,logfile=NULL,\n splitnmerge=0,maxsnm=FALSE,likeonly=FALSE,\n logweight=FALSE)\n}\n\n\\arguments{\n \\item{ydata}{[ndata,dy] matrix of observed quantities}\n \\item{ycovar}{[ndata,dy] / [ndata,dy,dy] / [dy,dy,ndata] matrix, list or 3D array\n of observational error covariances\n (if [ndata,dy] then the error correlations are assumed to vanish)}\n \\item{xamp}{[ngauss] array of initial amplitudes (*not* [1,ngauss])}\n \\item{xmean}{[ngauss,dx] matrix of initial means}\n \\item{xcovar}{[ngauss,dx,dx] list of matrices of initial covariances}\n \\item{projection}{[ndata,dy,dx] list of projection matrices}\n \\item{weight}{[ndata] array of weights to be applied to the data points}\n \\item{logweight}{(bool, default=False) if True, weight is actually\n log(weight)}\n \\item{fixamp}{(default=None) None, True/False, or list of bools}\n \\item{fixmean}{(default=None) None, True/False, or list of bools}\n \\item{fixcovar}{(default=None) None, True/False, or list of bools}\n \\item{tol}{(double, default=1.e-6) tolerance for convergence}\n \\item{maxiter}{(long, default= 10**9) maximum number of iterations to perform}\n \\item{w}{(double, default=0.) covariance regularization parameter\n (of the conjugate prior)}\n \\item{logfile}{basename for several logfiles (_c.log has output from\n the c-routine; _loglike.log has the log likelihood path of\n all the accepted routes, i.e. only parts which increase\n the likelihood are included, during splitnmerge)}\n \\item{splitnmerge}{(int, default=0) depth to go down the splitnmerge path}\n \\item{maxsnm}{(Bool, default=False) use the maximum number of split 'n'\n merge steps, K*(K-1)*(K-2)/2}\n \\item{likeonly}{(Bool, default=False) only compute the total log\n likelihood of the data}\n}\n\n\\value{\n \\item{avgloglikedata}{avgloglikedata after convergence}\n \\item{xamp}{updated xamp}\n \\item{xmean}{updated xmean}\n \\item{xcovar}{updated xcovar}\n}\n\n\\details{\n ...\n}\n\n\\author{Jo Bovy, David W. Hogg, & Sam T. Roweis}\n\n\\references{\nInferring complete distribution functions from\nnoisy, heterogeneous and incomplete observations Jo Bovy, David\nW. Hogg, & Sam T. Roweis, Submitted to AOAS (2009) [arXiv/0905.2979]\n}\n\n\\examples{\nlibrary(\"ExtremeDeconvolution\")\n?extreme_deconvolution\nydata <- c(2.62434536,0.38824359,0.47182825,-0.07296862,1.86540763,-1.30153870,2.74481176,0.23879310,1.31903910,0.75062962,2.46210794,-1.06014071,0.67758280,0.61594565,2.13376944,-0.09989127,0.82757179,0.12214158,1.04221375,1.58281521,-0.10061918,2.14472371,1.90159072,1.50249434,1.90085595,0.31627214,0.87710977,0.06423057,0.73211192,1.53035547,0.30833925,0.60324647,0.31282730,0.15479436,0.32875387,0.98733540,-0.11731035,1.23441570,2.65980218,1.74204416,0.80816445,0.11237104,0.25284171,2.69245460,1.05080775,0.36300435,1.19091548,3.10025514,1.12015895,1.61720311,1.30017032,0.64775015,-0.14251820,0.65065728,0.79110577,1.58662319,1.83898341,1.93110208,1.28558733,1.88514116,0.24560206,2.25286816,1.51292982,0.70190717,1.48851815,0.92442829,2.13162939,2.51981682,3.18557541,-0.39649633,-0.44411380,0.49553414,1.16003707,1.87616892,1.31563495,-1.02220122,0.69379599,1.82797464,1.23009474,1.76201118,0.77767186,0.79924193,1.18656139,1.41005165,1.19829972,1.11900865,0.32933771,1.37756379,1.12182127,2.12948391,2.19891788,1.18515642,0.62471505,0.36126959,1.42349435,1.07734007,0.65614632,1.04359686,0.37999916,1.69803203,0.55287144,2.22450770,1.40349164,1.59357852,-0.09491185,1.16938243,1.74055645,0.04629940,0.73378149,1.03261455,-0.37311732,1.31515939,1.84616065,0.14048406,1.35054598,-0.31228341,0.96130449,-0.61577236,2.12141771,1.40890054,0.97538304,0.22483838,2.27375593,2.96710175,-0.85798186,2.23616403,2.62765075,1.33801170,-0.19926803,1.86334532,0.81907970,0.39607937,-0.23005814,1.55053750,1.79280687,0.37646927,1.52057634,-0.14434139,1.80186103,1.04656730,0.81343023,0.89825413,1.86888616,1.75041164,1.52946532,1.13770121,1.07782113,1.61838026,1.23249456,1.68255141,0.68988323,-1.43483776,2.03882460,3.18697965,1.44136444,0.89984477,0.86355526,0.88094581,1.01740941,-0.12201873,0.48290554,0.00297317,1.24879916,0.70335885,1.49521132,0.82529684,1.98633519,1.21353390,3.19069973,-0.89636092,0.35308331,1.90148689,3.52832571,0.75136522,1.04366899,0.77368576,2.33145711,0.71269214,1.68006984,0.68019840,-0.27255875,1.31354772,1.50318481,2.29322588,0.88955297,0.38263794,1.56276110,1.24073709,1.28066508,0.92688730,2.16033857,1.36949272,2.90465871,2.11105670,1.65904980,-0.62743834,1.60231928,1.42028220,1.81095167,2.04444209)\nydata <- matrix(ydata, length(ydata), 1)\nN <- dim(ydata)[1]\nycovar <- ydata * 0 + 0.01\nxamp <- c(0.5, 0.5)\nxmean <- matrix(c(0.86447943, 0.67078879, 0.322681, 0.45087394), 2, 2)\nxcovar <- list(matrix(c(0.03821028, 0.04014796, 0.04108113, 0.03173839), 2, 2),\n matrix(c(0.06219194,0.09738021, 0.04302473, 0.06778009), 2, 2))\nprojection <- list()\nfor (i in 1:N) projection[[i]] = matrix(c(i\\%\\%2,(i+1)\\%\\%2), 1, 2)\nres <- extreme_deconvolution(ydata, ycovar, xamp, xmean, xcovar, projection=projection, logfile=\"ExDeconDemo\")\nprint(res)\nstopifnot((res$avgloglikedata - (-1.3114744655258121)) **2. < 10.**-8)\nstopifnot((res$xmean[1,1]-2.30368235)**2. < 10.**-5)\nstopifnot((res$xmean[1,2]-1.70701517)**2. < 10.**-5)\nstopifnot((res$xmean[2,1]-1.08009397)**2. < 10.**-5)\nstopifnot((res$xmean[2,2]-0.8888667)**2. < 10.**-5)\nstopifnot((res$xcovar[[1]][1,1]-0.445645987259)**2. < 10.**-5.)\nstopifnot((res$xamp[1]-0.11968415)**2. < 10.**-5)\nstopifnot((res$xamp[2]-0.880315852981)**2. < 10.**-5)\n}\n", "meta": {"hexsha": "abaf275850771d6203bea4c442c090853d305cfa", "size": 7007, "ext": "rd", "lang": "R", "max_stars_repo_path": "r/man/extreme_deconvolution.rd", "max_stars_repo_name": "HaifengWangNAOC/Learn-Bovy-Extreme-deconvolution", "max_stars_repo_head_hexsha": "bc6d58199b17cd5329d72f6af3c7ba7e6d2ae780", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2016-07-11T14:02:03.000Z", "max_stars_repo_stars_event_max_datetime": "2016-07-11T14:02:03.000Z", "max_issues_repo_path": "r/man/extreme_deconvolution.rd", "max_issues_repo_name": "HaifengWangNAOC/Learn-Bovy-Extreme-deconvolution", "max_issues_repo_head_hexsha": "bc6d58199b17cd5329d72f6af3c7ba7e6d2ae780", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "r/man/extreme_deconvolution.rd", "max_forks_repo_name": "HaifengWangNAOC/Learn-Bovy-Extreme-deconvolution", "max_forks_repo_head_hexsha": "bc6d58199b17cd5329d72f6af3c7ba7e6d2ae780", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 77.8555555556, "max_line_length": 2234, "alphanum_fraction": 0.7324104467, "num_tokens": 2635, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8333246118695629, "lm_q2_score": 0.6477982315512489, "lm_q1q2_score": 0.5398262098772337}} {"text": "library(openxlsx)\nlibrary(matlib)\nlibrary(Matrix)\n\nthis.dir<- dirname(parent.frame(2)$ofile)\nparent.dir <- strsplit(this.dir,'scripts')[[1]]\nsetwd(parent.dir)\n\nload_file <- 'efa/missing-loadings-f3-f12-sig-2.xlsx' #only statistically signficant loadings included\n#To compute LSI using all loadings set load_file to 'efa/missing-loadings-f3-f12.csv'\n\nf12_load_mat <- as.matrix(read.xlsx(load_file, sheet = 1, colNames = FALSE))\nf11_load_mat <- as.matrix(read.xlsx(load_file, sheet = 2, colNames = FALSE))\nf10_load_mat <- as.matrix(read.xlsx(load_file, sheet = 3, colNames = FALSE))\nf9_load_mat <- as.matrix(read.xlsx(load_file, sheet = 4, colNames = FALSE))\nf8_load_mat <- as.matrix(read.xlsx(load_file, sheet = 5, colNames = FALSE))\nf7_load_mat <- as.matrix(read.xlsx(load_file, sheet = 6, colNames = FALSE))\nf6_load_mat <- as.matrix(read.xlsx(load_file, sheet = 7, colNames = FALSE))\nf5_load_mat <- as.matrix(read.xlsx(load_file, sheet = 8, colNames = FALSE))\nf4_load_mat <- as.matrix(read.xlsx(load_file, sheet = 9, colNames = FALSE))\nf3_load_mat <- as.matrix(read.xlsx(load_file, sheet = 10, colNames = FALSE))\n\neps <- 1.0e-8\nw_fun <- function(x) {(x^2 + eps)^(10*x^2)}\n\"%^%\" <- function(x, n) with(eigen(x), vectors %*% (values^n * t(vectors)))\nlsi <- function(load_mat) {\n nfac <- ncol(load_mat)\n c_mat <- diag(t(load_mat) %*% load_mat)\n c_mat <- diag(c_mat)\n h_mat <- diag(load_mat %*% inv(c_mat) %*% t(load_mat))\n h_mat <- diag(h_mat)\n b_mat <- (h_mat %^% (-0.5)) %*% load_mat %*% (c_mat %^% (-0.5))\n ww <- sum(sapply( b_mat, w_fun,simplify=\"array\"))*(1/64.0)*(1.0/nfac) \n ee <- ((1.0/nfac) + eps)^(10.0/nfac)\n ind <- (ww-ee)/(1.0-ee)\n return(ind)\n}\n\nprint(paste0(ncol(f12_load_mat),\" factors: LSI =\", lsi(f12_load_mat)))\nprint(paste0(ncol(f11_load_mat),\" factors: LSI =\", lsi(f11_load_mat)))\nprint(paste0(ncol(f10_load_mat),\" factors: LSI =\", lsi(f10_load_mat)))\nprint(paste0(ncol(f9_load_mat),\" factors: LSI =\", lsi(f9_load_mat)))\nprint(paste0(ncol(f8_load_mat),\" factors: LSI =\", lsi(f8_load_mat)))\nprint(paste0(ncol(f7_load_mat),\" factors: LSI =\", lsi(f7_load_mat)))\nprint(paste0(ncol(f6_load_mat),\" factors: LSI =\", lsi(f6_load_mat)))\nprint(paste0(ncol(f5_load_mat),\" factors: LSI =\", lsi(f5_load_mat)))\nprint(paste0(ncol(f4_load_mat),\" factors: LSI =\", lsi(f4_load_mat)))\nprint(paste0(ncol(f3_load_mat),\" factors: LSI =\", lsi(f3_load_mat)))\n\n", "meta": {"hexsha": "090cc04c825844f1e4838422093f893a95bd70c2", "size": 2366, "ext": "r", "lang": "R", "max_stars_repo_path": "scripts/fit-metrics.r", "max_stars_repo_name": "jimioke/mitei-urban-typologies", "max_stars_repo_head_hexsha": "ff60018768b1b9d1a29b7de384f9d9a0adac8f33", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2020-11-18T11:18:04.000Z", "max_stars_repo_stars_event_max_datetime": "2020-11-18T19:11:48.000Z", "max_issues_repo_path": "scripts/fit-metrics.r", "max_issues_repo_name": "jimioke/mitei-urban-typologies", "max_issues_repo_head_hexsha": "ff60018768b1b9d1a29b7de384f9d9a0adac8f33", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "scripts/fit-metrics.r", "max_forks_repo_name": "jimioke/mitei-urban-typologies", "max_forks_repo_head_hexsha": "ff60018768b1b9d1a29b7de384f9d9a0adac8f33", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 47.32, "max_line_length": 102, "alphanum_fraction": 0.690194421, "num_tokens": 771, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8723473680407889, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5397914827715854}} {"text": "model_conductance <- function (vonKarman = 0.42,\n heightWeatherMeasurements = 2.0,\n zm = 0.13,\n zh = 0.013,\n d = 0.67,\n plantHeight = 0.0,\n wind = 124000.0){\n #'- Name: Conductance -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: Conductance Model\n #' * Author: Pierre Martre\n #' * Reference: Modelling energy balance in the wheat crop model SiriusQuality2:\n #' Evapotranspiration and canopy and soil temperature calculations\n #' \n #' * Institution: INRA/LEPSE Montpellier\n #' * Abstract: The boundary layer conductance is expressed as the wind speed profile above the\n #' canopy and the canopy structure. The approach does not take into account buoyancy\n #' effects. \n #' \n #'- inputs:\n #' * name: vonKarman\n #' ** description : von Karman constant\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 1\n #' ** default : 0.42\n #' ** unit : dimensionless\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' ** parametercategory : constant\n #' * name: heightWeatherMeasurements\n #' ** description : reference height of wind and humidity measurements\n #' ** parametercategory : soil\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 10\n #' ** default : 2\n #' ** unit : m\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: zm\n #' ** description : roughness length governing momentum transfer, FAO\n #' ** parametercategory : constant\n #' ** inputtype : parameter\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 1\n #' ** default : 0.13\n #' ** unit : m\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' * name: zh\n #' ** description : roughness length governing transfer of heat and vapour, FAO\n #' ** parametercategory : constant\n #' ** inputtype : parameter\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 1\n #' ** default : 0.013\n #' ** unit : m\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' * name: d\n #' ** description : corresponding to 2/3. This is multiplied to the crop heigth for calculating the zero plane displacement height, FAO\n #' ** inputtype : parameter\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.67\n #' ** min : 0\n #' ** max : 1\n #' ** unit : dimensionless\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547rl\n #' * name: plantHeight\n #' ** description : plant Height\n #' ** datatype : DOUBLE\n #' ** default : 0\n #' ** min : 0\n #' ** max : 1000\n #' ** unit : mm\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' ** variablecategory : auxiliary\n #' * name: wind\n #' ** description : wind\n #' ** variablecategory : auxiliary\n #' ** datatype : DOUBLE\n #' ** default : 124000\n #' ** min : 0\n #' ** max : 1000000\n #' ** unit : m/d\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #'- outputs:\n #' * name: conductance\n #' ** description : the boundary layer conductance\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 10000\n #' ** unit : m/d\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n h <- max(10.0, plantHeight) / 100.0\n conductance <- wind * vonKarman ^ 2 / (log((heightWeatherMeasurements - (d * h)) / (zm * h)) * log((heightWeatherMeasurements - (d * h)) / (zh * h)))\n return (list('conductance' = conductance))\n}", "meta": {"hexsha": "0621d29dd335635e4b7aa0f7c739fb0f51c78c4d", "size": 5698, "ext": "r", "lang": "R", "max_stars_repo_path": "test/Models/energybalance_pkg/src/r/Conductance.r", "max_stars_repo_name": "brichet/PyCrop2ML", "max_stars_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-06-21T18:58:04.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-29T21:32:28.000Z", "max_issues_repo_path": "test/Models/energybalance_pkg/src/r/Conductance.r", "max_issues_repo_name": "brichet/PyCrop2ML", "max_issues_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 27, "max_issues_repo_issues_event_min_datetime": "2018-12-04T15:35:44.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-11T08:25:03.000Z", "max_forks_repo_path": "test/Models/energybalance_pkg/src/r/Conductance.r", "max_forks_repo_name": "brichet/PyCrop2ML", "max_forks_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 7, "max_forks_repo_forks_event_min_datetime": "2019-01-15T04:33:23.000Z", "max_forks_repo_forks_event_max_datetime": "2021-12-09T07:29:46.000Z", "avg_line_length": 55.3203883495, "max_line_length": 164, "alphanum_fraction": 0.3775008775, "num_tokens": 1197, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8670357701094303, "lm_q2_score": 0.6224593312018546, "lm_q1q2_score": 0.539694505590401}} {"text": "#' ddirch_obs_uncertainty\n#' This function takes dirichlet variances at the site, plot and core level.\n#' It also requires a matrix of species abundances to represent \"true\" values at the site level.\n#' It starts at the site level, and draws plots within site from the site mean, and within site-scale variance.\n#' It then draws cores within plots, given the \"true\" plot means, and within plot-scale variance.\n#' It then draws estimates of observed core values, given the \"true\" core mean and within core variance.\n#' Once all this data is drawn, the function hierarchically estimates core, plot and site means (y_obs)\n#' It then regresses y_obs at each scale against the \"true\" values that were drawn, and estimates an R2 value.\n#' This is repeated some number of times to generate a distribution of potential R2 values.\n#' This reflects the maximum predictive accuracy once could expect, given observation uncertainty, in principle a perfect model would predict the \"true\" values generated here.\n#' Predictaibility likely varies as a function of abundance, as well as variance, and their interaction.\n#' Predictiability will also vary as a function scale.\n#' NOTE: larger variance values lead to less variance in the dirichlet.\n#'\n#' @param site_mu #matrix of site-scale relative abundances. Number of rows = number of sites.\n#' @param site.var #within site-scale variance.\n#' @param plot.var #within plot-scale variance.\n#' @param core.var #within core-scale variance.\n#' @param n.sim #numbre of simulations to run.\n#' @param n.plot #Number of plots within site. Default 10 for NEON.\n#' @param n.core #Number of cores within plot. Default 3 for NEON.\n#' @param n.threads #Number of cores to run in parallel. Default will auto-detect.\n#'\n#' @return\n#' @export\n#'\n#' @examples\nddirch_obs_uncertainty <- function(site_mu,\n site.var, plot.var, core.var,\n n.sim = 100, n.plot = 10, n.core = 3, n.threads = NA){\n #function tests to throw errors.----\n check <- \"DirichletReg\" %in% rownames(installed.packages())\n if(check == F){stop('The package DirichletReg is not installed. We need it to be.')}\n check <- \"doParallel\" %in% rownames(installed.packages())\n if(check == F){stop('The package doParallel is not installed. We need it to be.')}\n library(doParallel)\n \n #setup parallel.----\n if( is.na(n.threads)){\n n.cores <- detectCores()\n registerDoParallel(n.cores)\n }\n if(!is.na(n.threads)){\n registerDoParallel(n.threads)\n }\n \n \n #execute loop in parallel.----\n all.output <- \n foreach(j = 1:n.sim) %dopar% {\n #begin try-catch loop (sometimes we fail some type of hessian matrix calc...)----\n attempt = 0\n mod.plot <- NULL\n mod.site <- NULL\n while(is.null(mod.plot) && is.null(mod.site) && attempt <= 10){\n attempt = attempt + 1\n #1. Generate plot and core values.----\n #assign site labels.\n n.site <- nrow(site_mu)\n sites <- letters[1:n.site]\n \n #generate plot values.\n y.plot <- list()\n for(i in 1:nrow(site_mu)){\n plots <- DirichletReg::rdirichlet(n.plot, site_mu[i,]*site.var)\n #kill zeros.\n plots[plots == 0] <- min(plots[plots > 0])\n plot.lab <- paste0(sites[i],c(1:n.plot))\n plots <- data.frame(plot.lab, plots)\n colnames(plots) <- c('plotID',colnames(site_mu))\n y.plot[[i]] <- plots\n }\n y.plot <- do.call(rbind, y.plot)\n y.plot$siteID <- substr(y.plot$plotID, 1,1)\n \n #Generate core values.\n y.core <- list()\n for(i in 1:nrow(y.plot)){\n cores <- DirichletReg::rdirichlet(n.core, as.numeric(y.plot[i,grep('y',colnames(y.plot))]*plot.var))\n #kill zeros\n cores[cores == 0] <- min(cores[cores > 0])\n core.lab <- paste0(y.plot[i,'plotID'], '.', c(1:n.core))\n cores <- data.frame(core.lab, cores)\n colnames(cores) <- c('coreID',colnames(site_mu))\n y.core[[i]] <- cores\n }\n y.core <- do.call(rbind, y.core)\n y.core$coreID <- as.character(y.core$coreID)\n y.core$plotID <- substr(y.core$coreID, 1, nchar(y.core$coreID) - 2)\n y.core$siteID <- substr(y.core$coreID, 1,1)\n \n #Generate observed data by drawing from intra-core variance.\n y.obs <- DirichletReg::rdirichlet(nrow(y.core), as.matrix(y.core[,grep('y',colnames(y.core))]) * core.var)\n y.obs <- y.obs + min(y.obs[y.obs > 0])\n y.obs <- data.frame(y.obs)\n colnames(y.obs) <- colnames(y.core)[grep('y',colnames(y.core))]\n y.obs$coreID <- y.core$coreID\n y.obs$plotID <- substr(y.obs$coreID, 1, nchar(y.obs$coreID) - 2)\n \n #2. Aggregate to plot and site scale.----\n #I should maybe use JAGS here instead...\n #Plot scale aggregation.\n suppressWarnings(\n y.obs$Y <- DR_data(y.obs[,grep('y',colnames(y.obs))])\n )\n try(mod.plot <- DirichletReg::DirichReg(Y ~ plotID - 1, data = y.obs))\n plot.obs <- matrix(mod.plot$coefficients, ncol = ncol(site_mu))\n plot.obs <- exp(plot.obs)\n plot.obs <- data.frame(plot.obs / rowSums(plot.obs))\n colnames(plot.obs) <- colnames(site_mu)\n plot.obs$plotID <- unique(y.obs$plotID)\n plot.obs$siteID <- substr(plot.obs$plotID,1,1)\n \n #Site scale aggregation.\n suppressWarnings(\n plot.obs$Y <- DR_data(plot.obs[,grep('y',colnames(plot.obs))])\n )\n try(mod.site <- DirichletReg::DirichReg(Y ~ siteID - 1, data = plot.obs))\n site.obs <- matrix(mod.site$coefficients, ncol = ncol(site_mu))\n site.obs <- exp(site.obs)\n site.obs <- data.frame(site.obs / rowSums(site.obs))\n colnames(site.obs) <- colnames(site_mu)\n site.obs$siteID <- unique(plot.obs$siteID)\n #end try-catch loop.----\n }\n #3. Calculate r-squared values of 'true' vs. observed at core-plot-site scales.----\n #core-scale rsq values.\n core.rsq <- list()\n for(i in 1:ncol(site_mu)){\n pred <- y.obs[,grep('y',colnames(y.obs))]\n obs <- y.core[,grep('y',colnames(y.core))]\n mod <- lm(obs[,i] ~ pred[,i])\n core.rsq[[i]] <- summary(mod)$r.squared\n }\n core.rsq <- unlist(core.rsq)\n names(core.rsq) <- colnames(site_mu)\n \n #plot-scale rsq values.\n plot.rsq <- list()\n for(i in 1:ncol(site_mu)){\n pred <- plot.obs[,grep('y',colnames(plot.obs))]\n obs <- y.plot[,grep('y',colnames(y.plot))]\n mod <- lm(obs[,i] ~ pred[,i])\n plot.rsq[[i]] <- summary(mod)$r.squared\n }\n plot.rsq <- unlist(plot.rsq)\n names(plot.rsq) <- colnames(site_mu)\n \n #site-scale rsq values.\n site.rsq <- list()\n for(i in 1:ncol(site_mu)){\n pred <- site.obs[,grep('y',colnames(site.obs))]\n mod <- lm(site_mu[,i] ~ pred[,i])\n site.rsq[[i]] <- summary(mod)$r.squared\n }\n site.rsq <- unlist(site.rsq)\n names(site.rsq) <- colnames(site_mu)\n \n #wrap rsq output into list and return.\n rsq.out <- list(core.rsq, plot.rsq, site.rsq)\n names(rsq.out) <- c('core','plot','site')\n return(rsq.out)\n }\n \n #clean up all rsq output.----\n core.output <- list()\n plot.output <- list()\n site.output <- list()\n for(i in 1:length(all.output)){\n core.output[[i]] <- all.output[[i]]$core\n plot.output[[i]] <- all.output[[i]]$plot\n site.output[[i]] <- all.output[[i]]$site\n }\n core.output <- do.call(rbind, core.output)\n plot.output <- do.call(rbind, plot.output)\n site.output <- do.call(rbind, site.output)\n output <- list(core.output,plot.output,site.output)\n names(output) <- c('core','plot','site')\n \n #return output.----\n return(output)\n} #end function.\n", "meta": {"hexsha": "63c464e07e8d62bf68e99b9f3ba9d8c0b1240318", "size": 7832, "ext": "r", "lang": "R", "max_stars_repo_path": "NEFI_functions/ddirch_obs_uncertainty.r", "max_stars_repo_name": "colinaverill/NEFI_microbe", "max_stars_repo_head_hexsha": "e59ddef4aafcefdf0aff61765a8684859daad6e0", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2021-05-13T17:13:54.000Z", "max_stars_repo_stars_event_max_datetime": "2021-05-13T17:13:54.000Z", "max_issues_repo_path": "NEFI_functions/ddirch_obs_uncertainty.r", "max_issues_repo_name": "colinaverill/NEFI_microbe", "max_issues_repo_head_hexsha": "e59ddef4aafcefdf0aff61765a8684859daad6e0", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "NEFI_functions/ddirch_obs_uncertainty.r", "max_forks_repo_name": "colinaverill/NEFI_microbe", "max_forks_repo_head_hexsha": "e59ddef4aafcefdf0aff61765a8684859daad6e0", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 4, "max_forks_repo_forks_event_min_datetime": "2019-02-21T20:26:09.000Z", "max_forks_repo_forks_event_max_datetime": "2021-11-11T16:09:44.000Z", "avg_line_length": 42.7978142077, "max_line_length": 175, "alphanum_fraction": 0.6020173647, "num_tokens": 2114, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.8152324983301568, "lm_q2_score": 0.6619228825191872, "lm_q1q2_score": 0.5396210452180158}} {"text": "#' Sorensen dissimilarity index\n#' \n#' @param comm Community matrix giving sites (rows) by species (columns).\n#' @param partition Logical; if true, the index will be partitioned into turnover and\n#' \t\t\tnestedness.\n#' @references Baselga A. 2010. Partitioning the turnover and nestedness components of \n#' \t\tbeta diversity. Global Ecol. Biogeogr. 19: 134–143\n#' @references Sørensen, T. (1948) A method of establishing groups of equal amplitude in \n#' \t\tplant sociology based on similarity of species content. Kongelige Danske \n#' \t\tVidenskabernes Selskabs Biologiske Skrifter, 5, 1–34.\n#' @return If \\code{partition == TRUE}, list of three matrices giving the turnover,\n#' \t\tnestedness, and total components of dissimilarity. Otherwise, a matrix of\n#' \t\tpairwise dissimilarity values between communities.\n#' @export\nsorensen <- function(comm, partition = FALSE)\n{\n\tif(!is.matrix(comm)) comm <- as.matrix(comm)\n\n\ta <- comm %*% t(comm > 0)\n\tb <- comm %*% t(!comm)\n\tcc <- t(b)\n\tbsor <- (b + cc) / (2*a + b + cc)\n\n\tif(partition) {\n\t\tmbc <- pmin(b, cc)\n\t\tbturn <- mbc / (a + mbc)\n\t\tbnes <- bsor - bturn\n\t\treturn(list(turnover = bturn, nestedness = bnes, sorensen = bsor))\n\t} else\n\t\treturn(bsor)\n}\n\n#' Phylogenetic sorensen dissimilarity index\n#' \n#' @param comm Community matrix giving sites (rows) by species (columns).\n#' @param tree Dendrogram of class \\code{phylo}\n#' @param partition Logical; if true, the index will be partitioned into turnover and\n#' \t\t\tnestedness.\n#' @references Pavoine, S. and Ricotta, C. 2014. Functional and phylogenetic similarity \n#' \t\tamong communities. Methods in Ecology and Evolution 5(7): 666–675.\n#' @return If \\code{partition == TRUE}, list of three matrices giving the turnover,\n#' \t\tnestedness, and total components of dissimilarity. Otherwise, a matrix of\n#' \t\tpairwise dissimilarity values between communities.\n#' @export\nphylosor <- function(comm, tree, partition = FALSE)\n{\n\tif(is.null(tree$edge.length))\n\t\tstop(\"The tree must have branch lengths\")\n\n\tif(any(! comm %in% c(0,1) )) {\n\t\twarning(\"Community matrix can only contain 0 and 1; assigning 1 to all nonzero \n\t\t\tentries\")\n\t\tcomm <- 1 * (comm > 0)\n\t}\n\n\t# node labels and community columns must match\n\tcommDrop <- which(!colnames(comm) %in% tree$tip.label)\n\tif(length(commDrop) > 0)\n\t{\n\t\tif(length(commDrop) == ncol(comm))\n\t\t\tstop(\"No columns in comm match tip labels in tree\")\n\n\t\twarning(\"Dropping \", length(commDrop), \" columns from comm that were \n\t\t\tnot found in tree\")\n\n\t\tcomm <- comm[,-commDrop]\n\t}\n\ttreeDrop <- which(!tree$tip.label %in% colnames(comm))\n\tif(length(treeDrop) > 0)\n\t{\n\t\tif(length(treeDrop) == length(tree$tip.labe))\n\t\t\tstop(\"No tip labels in tree were found in column labels in comm\")\n\n\t\twarning(\"Dropping \", length(treeDrop), \" modes from tree that were \n\t\t\tnot found in comm\")\n\n\t\ttree <- ape::drop.tip(tree, treeDrop)\n\t}\n\n\tif(any(rowSums(comm) == 0))\n\t{\n\t\tdrop <- which(rowSums(comm) == 0)\n\t\twarning(length(drop), \" sites had 0 species and were dropped\")\n\t\tcomm <- comm[-drop,]\n\t}\n\n\tbranch_sp <- branch_adjacency(tree, TRUE)\n\t# save weights\n\twts <- attr(branch_sp, \"branch.lengths\")\n\t# match the margins\n\tbranch_sp <- branch_sp[,match(colnames(comm), colnames(branch_sp))]\n\n\tsite_branch <- comm %*% t(branch_sp)\n\t# convert site by branch to presence absence (instead of sp count)\n\tsite_branch <- 1 * (site_branch > 0)\n\t# now weight by branch lengths\n\tsite_branch <- sweep(site_branch, 2, wts, `*`)\n\tsorensen(site_branch, partition)\n}\n\n\n#' Sorensen dissimilarity index based on a dendrogram\n#' \n#' Produces dissimilarity matrices using a species by species distance matrix; useful\n#' for example for functional diversity.\n#' \n#' @param comm Community matrix giving sites (rows) by species (columns).\n#' @param distmat Distance matrix; margins must match exactly the columns in comm \n#' @param partition Logical; if true, the index will be partitioned into turnover and\n#' \t\t\tnestedness.\n#' @param ... Additional arguments to pass to \\code{\\link{hclust}}.\n#' @references Pavoine, S. and Ricotta, C. 2014. Functional and phylogenetic similarity \n#' \t\tamong communities. Methods in Ecology and Evolution 5(7): 666–675.\n#' @return If \\code{partition == TRUE}, list of three matrices giving the turnover,\n#' \t\tnestedness, and total components of dissimilarity. Otherwise, a matrix of\n#' \t\tpairwise dissimilarity values between communities.\n#' @export\ndendrosor <- function(comm, distmat, partition = TRUE, ...)\n{\n\tif(!inherits(distmat, \"dist\"))\n\t\tdistmat <- as.dist(distmat)\n\n\tif(!attr(distmat, \"Size\") == ncol(comm) & \n\t\tany(attr(distmat, \"Labels\") != colnames(comm)))\n\t\tstop(\"ncol(comm) must equal the number of rows/columns in distmat and the\n\t\t\tmargin labels must be identical\")\n\n\ttree <- hclust(distmat, ...)\n\tphylosor(comm, ape::as.phylo(tree), partition)\n}\n\n#' Branch adjacency\n#' \n#' Produce a branch adjacency matrix (i.e., branch by node)\n#' \n#' @param x Object of class phylo\n#' @param branch.lengths Logical; if true, the values of the adjacency matrix will be the\n#' \t\tbranch lengths, otherwise they will be 1 or 0\n#' @return A branch by node adjacency matrix\n#' @keywords internal\nbranch_adjacency <- function(x, branch.lengths = !is.null(x$edge.length)) {\n m <- matrix(0, ncol=length(x$tip), nrow=nrow(x$edge))\n from <- x$edge[,1]\n to <- x$edge[,2]\n g <- seq_along(x$tip.label)\n while (any(!is.na(g))) {\n i <- match(g, to)\n m[cbind(i, seq_along(i))] <- 1\n g <- from[i]\n }\n rownames(m) <- paste0(\"Branch\", seq.int(nrow(m)))\n colnames(m) <- x$tip.label\n if(branch.lengths)\n {\n \tattr(m, \"branch.lengths\") <- x$edge.length\n }\n return(m)\n}\n\n\n#' Sorensen dissimilarity (without partitioning)\n#' \n#' Compute sorensen dissimilarity index; deprecated method (but faster than new one)\n#' \n#' @param comm Community matrix giving sites (rows) by species (columns).\n#' @details The index is defined as 1 - (2 * (A ∩ B) / (A + B)). \n#' @return Matrix of pairwise dissimilarity values between communities.\n#' @keywords internal\nsor <- function(comm)\n{\n\t# convert to matrix of 0 and 1\t\n\tif(any(! comm %in% c(0,1) )) comm <- 1 * (comm > 0)\n\t# get the intersection of all sites - number of sp in common\n\t# this is the 0.5 * numerator of sorensen similarity\n\tsite_similarity <- comm %*% t(comm)\n\t\n\t# compute species richness and repeat into a matrix\n\trichness <- rowSums(comm)\n\tra <- matrix(richness, nrow=length(richness), ncol=length(richness))\n\t\n\t# denominator - ra is the number of spp at site1, t(ra) is the number at site 2\n\tdenom <- ra + t(ra)\n\tdimnames(denom) <- dimnames(site_similarity)\n\t\n\t1 - ((2 * site_similarity) / denom)\n}", "meta": {"hexsha": "ec2a027fb1a5cbb535e60bb84eef3156a503e3f2", "size": 6603, "ext": "r", "lang": "R", "max_stars_repo_path": "R/sorensen.r", "max_stars_repo_name": "mtalluto/mbmtools", "max_stars_repo_head_hexsha": "67a01b24ba2bc214081085559500e7b16f8489f2", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/sorensen.r", "max_issues_repo_name": "mtalluto/mbmtools", "max_issues_repo_head_hexsha": "67a01b24ba2bc214081085559500e7b16f8489f2", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 10, "max_issues_repo_issues_event_min_datetime": "2017-10-19T11:09:18.000Z", "max_issues_repo_issues_event_max_datetime": "2018-06-27T10:08:48.000Z", "max_forks_repo_path": "R/sorensen.r", "max_forks_repo_name": "mtalluto/mbmtools", "max_forks_repo_head_hexsha": "67a01b24ba2bc214081085559500e7b16f8489f2", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-05-25T20:28:51.000Z", "max_forks_repo_forks_event_max_datetime": "2021-05-25T20:28:51.000Z", "avg_line_length": 35.5, "max_line_length": 89, "alphanum_fraction": 0.6884749356, "num_tokens": 1881, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.7905303186696747, "lm_q2_score": 0.6825737344123242, "lm_q1q2_score": 0.5395952317805245}} {"text": "runLimTREE <- function(line, paramFile, dat = NULL, ...) {\r\n if (is.null(dat)) \r\n\tdat = loadInputData()\r\n \r\n params = read.csv(paramFile, stringsAsFactors=FALSE)[line,]\r\n params = unlist(params)\r\n \r\n out = LimTREE(dat[['MAP']], dat[['RainTerm_Drought']],\r\n dat[['SW1']], dat[['SW2']], \r\n dat[['BurntArea']], dat[['StressTerm_Drought']], dat[['MTWM']],\r\n dat[['MTCM']],\r\n dat[['PopDen']],\r\n dat[['buffalo']], dat[['cattle']], dat[['goat']], dat[['sheep']],\r\n dat[['urban']], dat[['crop']], dat[['pas']],\r\n params['m_drought'],\r\n params['min_mat'], params['max_mat'],\r\n params['trans_d'], params['p_fire'],\r\n params['k_pop'],\r\n params['k_buffalo'], params['k_cattle'], \r\n params['k_goat'], params['k_sheep'],\r\n params['p_drought'],\r\n params['min_maxTemp'], params['max_minTemp'],\r\n params['q10_maxTemp'], params['q10_minTemp'],\r\n params['v_drought'], params['v_maxTemp'], params['v_minTemp'],\r\n params['v_pop'],\r\n params['v_buffalo'], params['v_cattle'],\r\n params['v_goat'], params['v_sheep'],\r\n params['v_crop'] ,params['v_pas'],\r\n params['MAP_x0'], params['MAP_k'], params['MAT_x0'], params['MAT_k'], \r\n params['SW_x0'], params['SW_k'], params['mort_x0'], params['mort_k'],\r\n params['ex_x0'], params['ex_k'],\r\n params['max_T'], ...)\r\n\t\r\n return(out)\r\n}\r\n\r\nLimTREE <- function(MAP, rain_drought, SW1, SW2, fire, stress_drought,\r\n maxTemp, minTemp,\r\n popDen, \r\n buffalo, cattle, goat, sheep,\r\n urban, crop, pas,\r\n\t\t m_drought, min_mat, max_mat,\r\n d, p_fire,\r\n k_popDen, k_buffalo, k_cattle, k_goat, k_sheep,\r\n p_drought, min_maxTemp, max_minTemp,\r\n q10_maxTemp, q10_minTemp,\r\n\t\t v_drought, v_maxTemp, v_minTemp,\r\n v_popDen, v_buffalo, v_cattle, v_goat, v_sheep,\r\n v_crop, v_pas, \r\n\t\t MAP0, MAPk, MAT0, MATk, SW0, SWk, Mort0, Mortk, Exc0, Exck, maxT,\r\n\t\t includeSW = TRUE, ...) {\r\n \r\n f_MAP = LimMAP (MAP, rain_drought, m_drought, MAP0 , MAPk, ...)\r\n #f_MAT = LimMAT (MAT, min_mat, max_mat, MAT0 , MATk, ...)\r\n\t\r\n if (includeSW) f_SW = LimSW(SW1, SW2 , d, SW0 , SWk, ...)\r\n else {\r\n\tf_SW = f_MAP\r\n\tf_SW[!is.na(f_SW)] = 1.0\r\n }\r\n\r\n f_MAT = f_MAP\r\n f_MAT[!is.na(f_MAT)] = 1.0\r\n\t\r\n f_Mort = LimMort (fire, stress_drought, maxTemp, minTemp, popDen,\r\n buffalo, cattle, goat, sheep,\r\n v_drought, v_maxTemp, v_minTemp,\r\n v_popDen, v_buffalo, v_cattle, v_goat, v_sheep,\r\n p_fire, \r\n k_popDen, k_buffalo, k_cattle, k_goat, k_sheep,\r\n p_drought,\r\n min_maxTemp, max_minTemp, q10_maxTemp, q10_minTemp,\r\n Mort0, -Mortk, ...)\r\n\r\n \t\t\t \r\n f_Exc = LimExc (urban, crop, pas, v_crop, v_pas, Exc0, -Exck, ...)\r\n \r\n f_SW = raster::resample(f_SW , f_MAP)\r\n f_Exc = raster::resample(f_Exc, f_MAP)\r\n Tree = f_MAP * f_SW * f_Mort * f_Exc * maxT # * f_MAT\r\n return(addLayer(Tree, f_MAP, f_MAT, f_SW, f_Mort, f_Exc))\r\n}\r\n\r\nlogistic <- function(x, x0, k, sensitivity = FALSE) {\r\n FUN <- function(xi = x, ki = k) 1 / (1 + exp(-ki * (xi - x0)))\r\n dFUN <- function(xi) FUN(xi, k) * FUN(xi, -k)\r\n if (sensitivity)\r\n\tout = dFUN(x)/ dFUN(x0)\r\n else\r\n\tout = FUN()\r\n return(out)\r\n}\r\n\r\nLimMAP <- function(MAP, drought, m_drought,...){\r\n #MAP = MAP + (1-drought) * (1/m_drought) * (exp(-m_drought * MAP)-1)\r\n MAP = log(MAP)\r\n logistic(MAP, ...)\r\n}\r\n\r\n\r\nLimMAT <- function(MAT, min_mat, max_mat, ..., retutnVar = FALSE) {\r\n MAT = MAT - min_mat\r\n max_mat = max_mat - min_mat\r\n\t\r\n MAT[MAT <0] = 0\r\n\t\r\n MAT = logmin(MAT)\r\n if (retutnVar) return(MAT)\r\n logistic(MAT, ...)\r\n}\r\n\r\nLimList <- function(x, v, ..., retutnVar = FALSE) {\r\n v = c(1, v)\r\n x = mapply('*', x, v)\r\n xi = x[[1]]\r\n for (i in x[-1]) xi = xi + i\r\n\r\n xi = xi / sum(v)\r\n if (retutnVar) out = xi else out = try(logistic(xi, ...), silent = TRUE)\r\n if (class(out) ==\"try-error\") out = xi\r\n return(out)\r\n}\r\n\r\nLimSW <- function(SW1, SW2, d,...) {\r\n SW = (SW1 + d * SW2)/(1+d)\r\n #SW = logmin(SW) \r\n \r\n logistic(SW, ...)\r\n}\r\n\r\nLimTREE.convertUnity <- function(x, k) \r\n 1 - exp(x * (-1/k))\r\n\t\r\nLimTREE.Temp <- function(Temp, mn, q10) {\r\n Temp = (q10 + 1.0)^(Temp - mn)\t\r\n return(Temp)\r\n}\r\n\r\n\r\nLimMort <- function(fire, drought, maxTemp, minTemp, popDen,\r\n buffalo, cattle, goat, sheep,\r\n v_drought, v_maxTemp, v_minTemp,\r\n v_popDen, v_buffalo, v_cattle, v_goat, v_sheep,\r\n\t\t p_fire,\r\n k_popDen, k_buffalo, k_cattle, k_goat, k_sheep,\r\n p_drought, min_maxTemp, max_minTemp,\r\n q10_maxTemp, q10_minTemp, ...) {\r\n\t\t\r\n fire = fire^p_fire\r\n popDen0 = popDen\r\n popDen = LimTREE.convertUnity( popDen, k_popDen )\r\n buffalo = LimTREE.convertUnity(buffalo, k_buffalo)\r\n cattle = LimTREE.convertUnity( cattle, k_cattle )\r\n goat = LimTREE.convertUnity( goat, k_goat )\r\n sheep = LimTREE.convertUnity( sheep, k_sheep )\r\n \r\n maxTemp = LimTREE.Temp(maxTemp , min_maxTemp, q10_maxTemp)\r\n minTemp = LimTREE.Temp(minTemp*(-1), -max_minTemp, q10_minTemp)\r\n drought = drought^p_drought\r\n \r\n LimList(c(fire, drought, maxTemp, minTemp,\r\n popDen, buffalo, cattle, goat, sheep),\r\n c(v_drought, v_maxTemp, v_minTemp,\r\n v_popDen, v_buffalo, v_cattle, v_goat, v_sheep), ...)\r\n}\r\n\r\nLimExc <- function(urban, crop, pas, v_crop, v_pas, ...)\r\n LimList(c(urban, crop, pas), c(v_crop, v_pas), ...)\r\n\r\n\r\n\r\n", "meta": {"hexsha": "260fdebdda99ec68af0877bdd8e00aadfb95988d", "size": 6154, "ext": "r", "lang": "R", "max_stars_repo_path": "libs/LimTREE_r/LimTREE.r", "max_stars_repo_name": "douglask3/savanna_fire_feedback_test", "max_stars_repo_head_hexsha": "3c8cedcafd6b841efea6e8813ca674039336f19e", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "libs/LimTREE_r/LimTREE.r", "max_issues_repo_name": "douglask3/savanna_fire_feedback_test", "max_issues_repo_head_hexsha": "3c8cedcafd6b841efea6e8813ca674039336f19e", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "libs/LimTREE_r/LimTREE.r", "max_forks_repo_name": "douglask3/savanna_fire_feedback_test", "max_forks_repo_head_hexsha": "3c8cedcafd6b841efea6e8813ca674039336f19e", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2022-01-13T12:28:00.000Z", "max_forks_repo_forks_event_max_datetime": "2022-01-13T12:28:00.000Z", "avg_line_length": 35.7790697674, "max_line_length": 89, "alphanum_fraction": 0.5069873253, "num_tokens": 1797, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8267118111485244, "lm_q2_score": 0.6513548646660543, "lm_q1q2_score": 0.5384827598684758}} {"text": "# Simulation study: drifting AR(1) model\n\nround_timestamp <- function(ts, scalefactor.sec = 3600*12) {\n as.POSIXct(round(as.numeric(ts)/scalefactor.sec) * scalefactor.sec, origin = '1970-01-01', tz = 'UTC')\n}\n\nread.baro <- function(logger.name) {\n df <- gwloggeR.data::read(logger.name)$df\n if (nrow(df) == 0L) return(df)\n df <- df[!is.na(TIMESTAMP_UTC), ]\n df <- df[!is.na(PRESSURE_VALUE), ]\n df <- df[!duplicated(TIMESTAMP_UTC), ]\n df <- df[, .('PRESSURE_VALUE' = mean(PRESSURE_VALUE)),\n by = .('TIMESTAMP_UTC' = round_timestamp(TIMESTAMP_UTC))]\n\n # Meta data\n df[, 'FILE' := basename(logger.name)]\n df[, 'N' := .N]\n\n data.table::setkey(df, TIMESTAMP_UTC)\n data.table::setattr(df, 'logger.name', logger.name)\n\n df\n}\n\n# from analysis 04\ncompare <- function(df1, df2) {\n if (nrow(df1) == 0L || nrow(df2) == 0L) return(data.table::data.table())\n\n diff.df <- df1[J(df2), .('PRESSURE_DIFF' = x.PRESSURE_VALUE - i.PRESSURE_VALUE, TIMESTAMP_UTC)][!is.na(PRESSURE_DIFF), ]\n\n data.table::setkey(diff.df, TIMESTAMP_UTC)\n\n diff.df\n}\n\nref.compare <- function(df) {\n compare(df, read.baro('KNMI_20200312_hourly'))\n}\n\n# plot(arima.sim(n = 10000, list(ar = c(0.9)), sd = sqrt(23.62)) + 1033.64)\n# plot(forecast:::simulate.Arima(fit.arima, 10000))\n\ndf.drifter <- data.table::data.table(\n TIMESTAMP_UTC = seq(from = as.POSIXct('2010-01-01 00:00:00'), to = as.POSIXct('2020-01-01 00:00:00'), by = '12 hours'),\n PRESSURE_VALUE = arima.sim(n = 7305, list(ar = c(0.9)), sd = sqrt(23.62)) + 1033.64 + c(rep(0, 3000), 1:4305)*0.005,\n key = 'TIMESTAMP_UTC'\n)\n\narima(df.drifter$PRESSURE_VALUE, order = c(1, 0, 0))\nplot(df.drifter$PRESSURE_VALUE)\n\ndf.diff <- ref.compare(df.drifter)\nplot(df.diff$PRESSURE_DIFF)\n\n# AR(1) component in case of difference between two barometers -----------------\nsim <- function(n, mu, sigma, phi1, betas = NULL, xreg = NULL,\n init = mu, xt = NULL, a = rnorm(n, 0, sd = sigma)) {\n\n R <- rep(0, n)\n if (!is.null(betas) || !is.null(xreg)) {\n reg <- xreg[-1,,drop=FALSE] - phi1*xreg[-n,,drop=FALSE]\n R <- c(0, reg %*% betas)\n }\n\n if (is.null(xt)) {\n # xp is the previous x: x_{t-1}\n Reduce(f = function(xp, t) phi1*xp + mu*(1-phi1) + R[t] + a[t], x = 2:n, init = init, accumulate = TRUE)\n } else {\n c(xt[1],\n phi1*xt[-n] + mu*(1-phi1) + R[-1] + a[-1])\n }\n}\n\nset.seed(2020)\ncorrelated.errors <- data.table::as.data.table(\n MASS::mvrnorm(1e5, mu = c('a' = 0, 'b' = 0), Sigma = matrix(c(23, 20, 20, 23), ncol = 2))\n)\ncov(correlated.errors$a, correlated.errors$b)\n\n## all equal: OK (error due to correlation is small --> -2*covariance)\ndf.diff <- data.table::data.table(\n TIMESTAMP_UTC = seq(from = as.POSIXct('2001-01-01 00:00:00'), by = '12 hours', length.out = 1e5),\n BARO_01 = sim(n = 1e5, phi1 = 0.9, mu = 1032, a = correlated.errors$a),\n BARO_02 = sim(n = 1e5, phi1 = 0.9, mu = 1032, a = correlated.errors$b)\n)\ndf.diff[, DIFF := BARO_01 - BARO_02]\nforecast::auto.arima(df.diff$DIFF, trace = TRUE, stationary = TRUE)\n\n## uncorrelated: OK (error = the sum of individual errors)\nset.seed(2020)\ndf.diff <- data.table::data.table(\n TIMESTAMP_UTC = seq(from = as.POSIXct('2001-01-01 00:00:00'), by = '12 hours', length.out = 1e5),\n BARO_01 = sim(n = 1e5, phi1 = 0.9, mu = 1020, sigma = sqrt(23)),\n BARO_02 = sim(n = 1e5, phi1 = 0.9, mu = 1035, sigma = sqrt(23))\n)\ndf.diff[, DIFF := BARO_01 - BARO_02]\nforecast::auto.arima(df.diff$DIFF, trace = TRUE, stationary = TRUE)\n\n## mu different: OK (new mu = difference between the two)\ndf.diff <- data.table::data.table(\n TIMESTAMP_UTC = seq(from = as.POSIXct('2001-01-01 00:00:00'), by = '12 hours', length.out = 1e5),\n BARO_01 = sim(n = 1e5, phi1 = 0.9, mu = 1020, a = correlated.errors$a),\n BARO_02 = sim(n = 1e5, phi1 = 0.9, mu = 1035, a = correlated.errors$b)\n)\ndf.diff[, DIFF := BARO_01 - BARO_02]\nforecast::auto.arima(df.diff$DIFF, trace = TRUE, stationary = TRUE)\n\n## sigma different: OK (new sigma: addition of the two minus 2*cov)\nset.seed(2020)\ncorrelated.errors.different.sigma <- data.table::as.data.table(\n MASS::mvrnorm(1e5, mu = c('a' = 0, 'b' = 0), Sigma = matrix(c(10, 15, 15, 23), ncol = 2))\n)\ncov(correlated.errors.different.sigma$a, correlated.errors.different.sigma$b)\ndf.diff <- data.table::data.table(\n TIMESTAMP_UTC = seq(from = as.POSIXct('2001-01-01 00:00:00'), by = '12 hours', length.out = 1e5),\n BARO_01 = sim(n = 1e5, phi1 = 0.9, mu = 1035, a = correlated.errors.different.sigma$a),\n BARO_02 = sim(n = 1e5, phi1 = 0.9, mu = 1035, a = correlated.errors.different.sigma$b)\n)\ndf.diff[, DIFF := BARO_01 - BARO_02]\nforecast::auto.arima(df.diff$DIFF, trace = TRUE, stationary = TRUE)\n\n## phi different: MA component seems necessary, new sigma larger\ndf.diff <- data.table::data.table(\n TIMESTAMP_UTC = seq(from = as.POSIXct('2001-01-01 00:00:00'), by = '12 hours', length.out = 1e5),\n BARO_01 = sim(n = 1e5, phi1 = 0.5, mu = 1035, a = correlated.errors$a),\n BARO_02 = sim(n = 1e5, phi1 = 0.9, mu = 1035, a = correlated.errors$b)\n)\ndf.diff[, DIFF := BARO_01 - BARO_02]\nforecast::auto.arima(df.diff$DIFF, trace = TRUE, stationary = TRUE)\n\n## phi different & uncorrelated: MA component seems necessary, new sigma larger\nset.seed(2020)\ndf.diff <- data.table::data.table(\n TIMESTAMP_UTC = seq(from = as.POSIXct('2001-01-01 00:00:00'), by = '12 hours', length.out = 1e5),\n BARO_01 = sim(n = 1e5, phi1 = 0.5, mu = 1035, sigma = sqrt(23)),\n BARO_02 = sim(n = 1e5, phi1 = 0.9, mu = 1035, sigma = sqrt(23))\n)\ndf.diff[, DIFF := BARO_01 - BARO_02]\nforecast::auto.arima(df.diff$DIFF, trace = TRUE, stationary = TRUE)\n\n", "meta": {"hexsha": "522ba96f7eec5f42852b7aff9ca03e86ba53389e", "size": 5563, "ext": "r", "lang": "R", "max_stars_repo_path": "src/r/drifts/analysis_24.r", "max_stars_repo_name": "DOV-Vlaanderen/groundwater-logger-validation", "max_stars_repo_head_hexsha": "db9bd59c1644bb298206e717a528b4974e3939d5", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2021-07-16T10:47:56.000Z", "max_stars_repo_stars_event_max_datetime": "2021-07-16T10:47:56.000Z", "max_issues_repo_path": "src/r/drifts/analysis_24.r", "max_issues_repo_name": "DOV-Vlaanderen/groundwater-logger-validation", "max_issues_repo_head_hexsha": "db9bd59c1644bb298206e717a528b4974e3939d5", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 61, "max_issues_repo_issues_event_min_datetime": "2019-05-17T21:14:25.000Z", "max_issues_repo_issues_event_max_datetime": "2022-01-26T13:47:40.000Z", "max_forks_repo_path": "src/r/drifts/analysis_24.r", "max_forks_repo_name": "DOV-Vlaanderen/groundwater-logger-validation", "max_forks_repo_head_hexsha": "db9bd59c1644bb298206e717a528b4974e3939d5", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2019-07-30T10:39:48.000Z", "max_forks_repo_forks_event_max_datetime": "2021-07-16T10:48:04.000Z", "avg_line_length": 39.176056338, "max_line_length": 122, "alphanum_fraction": 0.6381448859, "num_tokens": 2023, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8006920020959543, "lm_q2_score": 0.6723316926137812, "lm_q1q2_score": 0.5383306090314902}} {"text": "# test wfs simulations\n\nrequire(foreach)\nrequire(doParallel)\nsource('scripts/wfs_simp.r')\n\n# make cluster\ncl <- makeCluster(3)\nregisterDoParallel(cl)\n\n\n# probability of fixation of a neutral allele should equal the initial frequency\n\tne <- 100\n\tgen=1000 # run for a long time: likely to fixaton (expected time is 4N)\n\tsmin <- smax <- 0\n\tfs <- seq(1/100, 91/100, by=5/100) # initial allele frequencies\n\tnsims <- 1000\n\n\n\tsims <- data.frame(f1=rep(fs, rep(nsims, length(fs))))\n\tsims$f2 <- sims$f1samp <- sims$f2samp <- NA\n\tfor(j in 1:length(fs)){\n\t\tprint(j)\n\t\tthesesims <- foreach(i=1:nsims, .combine=rbind) %dopar% {\n\t\t\twfs_simp(i, f1=fs[j], s=0, c1=50, c2=50, gen=gen, ne=ne)\n\t\t}\n\t\tsims[((j-1)*1000+1):(j*1000),c('f2', 'f1samp', 'f2samp')] <- as.data.frame(thesesims)\n\t}\n\n\tpropfix <- aggregate(list(prop=sims$f2), by=list(f1=sims$f1), FUN=function(x) sum(x==1)/nsims)\n\n\t# plot of proportion fixed vs. initial frequency\n\tquartz(width=5, height=4)\n\t# png(width=5, height=4, units='in', res=300, file='analysis/figures/wfs_sim_test_s=0.png')\n\tpar(las=1)\n\twith(propfix, plot(f1, prop, xlab='Initial allele frequency', ylab='Proportion fixed', main='Ne=100, s=0, nsims=1000'))\n\tabline(0,1)\n\t\n\tdev.off()\n\n\n\n\n# probability of fixation of a non-neutral allele should equal s if initial frequency is 1/ne\n# (usually written as 2s, but that's when fitness of waa is 1+s2. here it is 1+s)\n# for large N and small s\n\tne <- 100\n\tgen=1000 # run for a long time: likely to fixation in 4N generations with drift\n\tss <- seq(0.05,1,by=0.05)\n\tf1 <- 1/ne # starting allele freq\n\tnsims <- 1000\n\n\tsims2 <- data.frame(s=rep(ss, rep(nsims, length(ss))))\n\tsims2$f2 <- sims2$f1samp <- sims2$f2samp <- NA\n\tlength(ss)\n\tfor(j in 1:length(ss)){\n\t\tprint(j)\n\t\tthesesims <- foreach(i=1:nsims, .combine=rbind) %dopar% {\n\t\t\tif(i %% 100 == 0) cat(i)\n\t\t\twfs_simp(i, f1=f1, s=ss[j], c1=50, c2=50, gen=gen, ne=ne)\n\t\t}\n\t\tsims2[((j-1)*1000+1):(j*1000),c('f2', 'f1samp', 'f2samp')] <- as.data.frame(thesesims)\n\t}\n\n\tsum(sims2$f2 %in% c(0,1))/nsims/length(ss) # fraction that reached fixation: make gen larger until this=1\n\tpropfix2 <- aggregate(list(prop=sims2$f2), by=list(s=sims2$s), FUN=function(x) sum(x==1)/nsims)\n\tpropfix2$s_exp1 <- (1-exp(-propfix2$s))/(1-exp(-2*ne*propfix2$s)) # Eq. 2.21 in Graur & Li\n\tpropfix2$s_exp2 <- propfix2$s/(1-exp(-2*ne*propfix2$s)) # Eq. 2.22 in Graur & Li\n\n\t# plot of proportion fixed vs. s\n\tquartz(width=5, height=4)\n\t# png(width=5, height=4, units='in', res=300, file='analysis/figures/wfs_sim_test_s>0.png')\n\tpar(las=1)\n\twith(propfix2, plot(s, prop, xlab='Selection coefficient (s)', ylab='Proportion fixed at p=1', main='Ne=100, p0=1/Ne, nsims=1000', ylim=c(0,0.7)))\n\twith(propfix2, lines(s, s_exp1, col='green'))\n\tlegend('topleft', col='green', lty=1, legend='(1-exp(-s))/(1-exp(-2*ne*s))', bty='n', cex=0.7)\n\t\n\tdev.off()\n\n\n\n## examine how much drift vs. sampling variance for large ne\n\tne <- 10000 # like NEA cod\n\tgen=11 # like NEA cod\n\tsmin <- smax <- 0\n\tf1min <- f1max <- ne/2/ne # starting allele freq\n\tnsims <- 1000\n\n\n\tsims <- foreach(i=1:nsims, .combine=rbind) %dopar% {\n\t\tif(i %% 100 == 0) cat(i)\n\t\twfs(i, f1min=f1min, f1max=f1max, smin=smin, smax=smax, c1=50, c2=50, gen=gen, ne=ne)\n\t}\n\tsims <- as.data.frame(sims)\n\n\t# examine drift variance and sampling variance\n\tpar(mfrow=c(3,1))\n\tbks <- seq(0,1,length.out=20)\n\thist(sims$f1samp, main='f1samp', breaks=bks, col='grey'); abline(v=f1min, col='red', lwd=2)\n\thist(sims$f2, main='f2', breaks=bks, col='grey'); abline(v=f1min, col='red', lwd=2)\n\thist(sims$f2samp-sims$f2, main='delta f2samp', breaks=seq(-0.5,0.5,length.out=20), col='grey'); abline(v=0, col='red', lwd=2)\n\n\n## examine how much drift to expect for large Ne\n\t# observations\n\tdat <- locnms <- fread('analysis/Frequency_table_Lof07_Lof14.txt', header=TRUE); setnames(locnms, 3:7, c('N_CHR_1', 'Freq_1', 'N_CHR_2', 'Freq_2', 'ABS_DIFF')) # the name and observed frequencies of all the loci, from output by Bastiaan Star\n\n\n\t# simulations\n\tne <- 10000\n\tgen=11 # run for a long time: likely to fixaton (expected time is 4N)\n\tfs <- seq(1/100, 91/100, by=5/100) # initial allele frequencies\n\tnsims <- 100000 # takes 20 min or so\n\n\tsims <- data.frame(f1=rep(fs, rep(nsims, length(fs))))\n\tsims$f2 <- sims$f1samp <- sims$f2samp <- NA\n\tfor(j in 1:length(fs)){\n\t\tprint(j)\n\t\tthesesims <- foreach(i=1:nsims, .combine=rbind) %dopar% {\n\t\t\twfs_simp(i, f1=fs[j], s=0, c1=44, c2=44, gen=gen, ne=ne)\n\t\t}\n\t\tsims[((j-1)*nsims+1):(j*nsims),c('f2', 'f1samp', 'f2samp')] <- as.data.frame(thesesims)\n\t}\n\t\n\txlims <- range(c(dat[,Freq_2-Freq_1], sims$f2samp-sims$f1samp))\n\tcols <- c('black', '#7fcdbb', '#edf8b1', '#2c7fb8') # sims, obs 44, low obs, high obs\n\n\t# plot histogram of diff vs. initial observed frequency\n\tquartz(width=7, height=6)\n\t# png(width=7, height=7, filename='analysis/figures/wfs_sim_test_hist_obsdiff_by_obsf1_with_data.png', units='in', res=300)\n\tf1s <- seq(0,1.1,by=0.1)\n\tbks <- seq(-1,1,by=0.05)\n\tpar(mfrow=c(3,4), las=1, tcl=-0.3, mgp=c(2.4,0.6,0), mai=c(0.5, 0.5, 0.3, 0.05))\n\tfor(i in 2:length(f1s)){\n\t\tinds <- (sims$f1samp >= f1s[i-1]) & (sims$f1samp < f1s[i])\n\t\thst <- hist(sims$f2samp[inds] - sims$f1samp[inds], plot=FALSE, breaks=bks) # returns some 0 values that get eliminated, but appear as holes in the histogram: doesn't change our interpretation\n\t\tj <- which(hst$density>0)\n\t\tplot(hst$mids[j], hst$density[j], type='o', xlab='Sample ∆freq', ylab='log(density)', main=paste('initial ', f1s[i-1], '-', f1s[i], sep=''), pch=16, xlim=xlims, log='y', ylim=c(0.0001,20), col=cols[1])\n\t\tabline(v=c(-0.06, 0.06), col='grey', lty=3) # genome-wide allele frequency change\n\n\t\t\t# low sample size\n\t\tinds3 <- dat[,Freq_1 >= f1s[i-1] & Freq_1 < f1s[i] & (N_CHR_1<44 | N_CHR_2<44)]\n\t\thst3 <- hist(dat[inds3,Freq_2 - Freq_1], plot=FALSE, breaks=bks) # returns some 0 values that get eliminated, but appear as holes in the histogram: doesn't change our interpretation\n\t\tj3 <- which(hst3$density>0)\n\t\tlines(hst3$mids[j3], hst3$density[j3], type='o', pch=16, col=cols[3])\n\n\t\t\t# high sample size\n\t\tinds4 <- dat[,Freq_1 >= f1s[i-1] & Freq_1 < f1s[i] & (N_CHR_1>44 & N_CHR_2>44)]\n\t\thst4 <- hist(dat[inds4,Freq_2 - Freq_1], plot=FALSE, breaks=bks) # returns some 0 values that get eliminated, but appear as holes in the histogram: doesn't change our interpretation\n\t\tj4 <- which(hst4$density>0)\n\t\tlines(hst4$mids[j4], hst4$density[j4], type='o', pch=16, col=cols[4])\n\n\t\t\t# sample size 44\n\t\tinds2 <- dat[,Freq_1 >= f1s[i-1] & Freq_1 < f1s[i] & N_CHR_1==44 & N_CHR_2==44]\n\t\thst2 <- hist(dat[inds2,Freq_2 - Freq_1], plot=FALSE, breaks=bks) # returns some 0 values that get eliminated, but appear as holes in the histogram: doesn't change our interpretation\n\t\tj2 <- which(hst2$density>0)\n\t\tlines(hst2$mids[j2], hst2$density[j2], type='o', pch=16, col=cols[2])\n\n\t}\n\t\n\tlegend('bottomright', col=cols, lty=1, legend=c('Sims n=44', 'Obs n=44', 'Obs low n', 'Obs high n'), cex=0.7, bty='n')\n\n\tdev.off()\n\n\n#######################\n# clean up cluster\n#######################\nstopCluster(cl)\n\n", "meta": {"hexsha": "11d23a81032aa91a9a5dfa8af127c85a850c53e1", "size": 6983, "ext": "r", "lang": "R", "max_stars_repo_path": "scripts/wfs_test.r", "max_stars_repo_name": "pinskylab/codEvol", "max_stars_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "scripts/wfs_test.r", "max_issues_repo_name": "pinskylab/codEvol", "max_issues_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 5, "max_issues_repo_issues_event_min_datetime": "2020-04-11T11:14:18.000Z", "max_issues_repo_issues_event_max_datetime": "2021-02-21T19:57:31.000Z", "max_forks_repo_path": "scripts/wfs_test.r", "max_forks_repo_name": "pinskylab/codEvol", "max_forks_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 40.8362573099, "max_line_length": 242, "alphanum_fraction": 0.6567377918, "num_tokens": 2655, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929002541068, "lm_q2_score": 0.6893056231680121, "lm_q1q2_score": 0.5376534921762822}} {"text": " # copied from library carstm\n stmv_hyperparameters = function( reference_sd, alpha=0.5, reference_mean=0 ) {\n # some generic PC priors, scaled by sd of data\n # pc.prior to median .. minimally info. scale\n\n hyper = list(\n\n iid = list(\n prec = list(\n prior = \"pc.prec\", # exponential decay\n param = c(reference_sd, alpha)\n )\n ),\n\n # means informative, sd marginally diffuse\n # see: inla.set.control.fixed.default() for defaults\n fixed = list(\n mean.intercept = reference_mean,\n prec.intercept = 1e-3,\n mean=0,\n prec=1e-2\n ),\n\n\n # param=c(u, alpha); u=sigma; alpha=prob;\n # see inla.doc(\"pc.prec\") ..prior sd attributable to rw2\n rw2 = list(\n prec = list(\n prior = \"pc.prec\", # exponential decay\n param = c(reference_sd, alpha)\n )\n ),\n\n # see inla.doc(\"ar1\") ; theta0, theta1 are expected\n # param=c(u, alpha); u=sigma; alpha=prob;\n # see inla.doc(\"pc.prec\") ..prior sd attributable to autocor rho\n # param=c(u, alpha); rho = 0.5; u=sqrt(1-rho); alpha=prob; see inla.doc(\"pc.cor1\")\n ar1 = list(\n prec = list(\n prior = \"pc.prec\", # exponential decay\n param = c(reference_sd, alpha)\n ),\n rho = list(\n prior = \"pc.cor0\", # inla.doc(\"pc.cor0\") ..base model: rho = 0 --- expoential; will tend to 0 unless there is info\n param = c(sqrt(1-0.5), 0.1) # rho=0.5; u=sqrt(1-rho) ... 100-10% of probablity weight to rho 0.5 or less .. forces smooth and only goes high if really high\n )\n ),\n\n # naming convention is a bit different in groups .. Must be one of theta theta1 rho logit correlation\n ar1_group = list(\n # theta1 = list(\n # prior = \"pc.prec\", # exponential decay\n # param = c(reference_sd, alpha)\n # ),\n rho = list(\n prior = \"pc.cor0\", # inla.doc(\"pc.cor0\") ..base model: rho = 0 --- expoential; will tend to 0 unless there is info\n param = c(sqrt(1-0.5), 0.1) # rho=0.5; u=sqrt(1-rho) ... 100-10% of probablity weight to rho 0.5 or less .. forces smooth and only goes high if really high\n )\n ),\n\n # param=c(u, alpha); u=phi (proportion spatial); alpha=prob\n bym2 = list(\n prec = list(\n prior = \"pc.prec\",\n param = c(reference_sd, alpha)\n ),\n phi = list(\n prior=\"pc\", # see bottom of inla.doc(\"bym2\")\n param=c(0.5, 0.5) # c(phi=0.5, alpha=0.5)\n )\n )\n )\n\n return(hyper)\n }\n", "meta": {"hexsha": "2d91c2ae85de3fdc0fa0d62e26bebf461d57e44e", "size": 2734, "ext": "r", "lang": "R", "max_stars_repo_path": "R/stmv_hyperparameters.r", "max_stars_repo_name": "jae0/stmv", "max_stars_repo_head_hexsha": "f837151bb59776d5f902c6f5a4b5a43cd272e96d", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/stmv_hyperparameters.r", "max_issues_repo_name": "jae0/stmv", "max_issues_repo_head_hexsha": "f837151bb59776d5f902c6f5a4b5a43cd272e96d", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/stmv_hyperparameters.r", "max_forks_repo_name": "jae0/stmv", "max_forks_repo_head_hexsha": "f837151bb59776d5f902c6f5a4b5a43cd272e96d", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.9736842105, "max_line_length": 169, "alphanum_fraction": 0.5160936357, "num_tokens": 771, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8740772351648677, "lm_q2_score": 0.6150878555160665, "lm_q1q2_score": 0.537634292132971}} {"text": "library(animation)\r\nx <- c(1,2,3,5,6,9)\r\nani.options(interval = 1,ani.height = 1000,ani.width = 1000)\r\nsaveGIF({\r\n for(i in 1:length(x)){\r\n plot(x,rep(0,length(x)), pch = 20,cex = 3, axes = FALSE, xlab = \"\", ylab = \"\")\r\n abline(h = 0)\r\n text(x = x, y = rep(0.1,length(x)), c(expression(x[1]),expression(x[2]),expression(x[3]),\r\n expression(x[4]),expression(x[5]),expression(x[6])),\r\n cex = 2)\r\n text(x = x, y = rep(-0.1,length(x)),x,cex = 2)\r\n arrows(mean(x),-0.75,mean(x),0,length = 0.2,lwd = 2)\r\n text(mean(x),-0.85,expression(paste(mu[x], \"(central value)\")),cex = 1.5)\r\n text(mean(x),-0.95,paste(\"=\",bquote(.(round(mean(x),2)))),cex = 1.5)\r\n arrows(7,0.75,x[4:6],0.2,length = 0.2,lwd = 2,lty = c(2,1,1))\r\n text(7,0.9,\"Possible values of X\",cex = 1.5)\r\n ##legend(\"topleft\",bty = \"n\", legend = expression(paste(\"Var(X) = E[(X -\",mu[X],\")\",\"\"^2,\"]\")),cex = 1.3)\r\n arrows(mean(x),-0.25,x[i],-0.25,length = 0.1,code = 3,lwd = 2) \r\n text(mean(c(x[i],mean(x))),-0.3,bquote(x[.(i)] - mu[X]),cex = 2)\r\n text(mean(c(x[i],mean(x))),-0.4, paste(\"=\",bquote(.(round(x[i] - mean(x),2)))),cex = 1.5)\r\n text(1.5,0.6,bquote((x[.(i)] - mu[X])^2),cex = 2)\r\n text(2.5,0.6, paste(\"=\",bquote(.(round((x[i] - mean(x))^2,2)))),cex = 2)\r\n }}, movie.name = \"var.gif\"\r\n)\r\n", "meta": {"hexsha": "6e5b59124e50ca5ef5cace6541412fc495592039", "size": 1356, "ext": "r", "lang": "R", "max_stars_repo_path": "r_scripts/var.r", "max_stars_repo_name": "statbiscuit/swots", "max_stars_repo_head_hexsha": "005da6fdb980e6cbb40f407e059623256f9c0626", "max_stars_repo_licenses": ["CC0-1.0"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2020-02-22T02:58:01.000Z", "max_stars_repo_stars_event_max_datetime": "2020-02-22T03:01:46.000Z", "max_issues_repo_path": "r_scripts/var.r", "max_issues_repo_name": "cmjt/statbiscuits", "max_issues_repo_head_hexsha": "005da6fdb980e6cbb40f407e059623256f9c0626", "max_issues_repo_licenses": ["CC0-1.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "r_scripts/var.r", "max_forks_repo_name": "cmjt/statbiscuits", "max_forks_repo_head_hexsha": "005da6fdb980e6cbb40f407e059623256f9c0626", "max_forks_repo_licenses": ["CC0-1.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 54.24, "max_line_length": 110, "alphanum_fraction": 0.5051622419, "num_tokens": 551, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.757794360334681, "lm_q2_score": 0.7090191399336402, "lm_q1q2_score": 0.5372907056110585}} {"text": "### RD design functions\n###peter crosta pmcrosta at gmail dot com\n## This file consists of only functions that are used in the \n## RD analysis. It will be sourced\n\n###########MOST FUNCTIONS ARE DEPENDENT ON maxbin AND binwidth (h) ###########################\n\n## Create class binlist for numeric breaks\nsetClass(\"binlist\", representation(breaks=\"numeric\", right=\"numeric\", left=\"numeric\"))\n\ngetbins <- function(binwidth, maxbin) {\n rightbins <- seq(binwidth, maxbin, binwidth)\n leftbins <- -rev(seq(0, maxbin, binwidth))\n\n breakbins <- c(sort(leftbins), rightbins)\n \n z <- new(\"binlist\")\n z@breaks <- breakbins[abs(breakbins)<=maxbin]\n z@right <- rightbins[rightbins<=maxbin]\n z@left <- leftbins[leftbins>=-maxbin]\n\n return(z)\n}\n\n# -244 to 121 might matter, days before and after Sep 1\nmainhist <- function(binwidth, maxbin, yesplot=TRUE, emcp=emc) {\n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n maxbin.hist <- hist(emcp[emcdum==1], breaks=breaks, plot=yesplot, main=paste(\"Histogram of birth date (centered): Binwidth=\",binwidth, \", \", length(breaks)-1, \" bins\", sep=\"\"), xlim=c(min(breaks), max(breaks)), xlab=\"Centered Birth Date\")\n if (yesplot) abline(v=0, lty=2, lwd=2, col=\"darkgrey\")\n binmids <- maxbin.hist$mids \n}\n\n\n####this function is built from McCrary(2008); tests discontinuity of density\nmccrary <- function(binwidth, maxbin, nn, emcp=emc) {\n f.triangle <- function(x) { ifelse(abs(x)<1, 1-abs(x), 0) }\n\n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n maxbin.hist <- hist(emcp[emcdum==1], breaks=breaks, plot=FALSE)\n binmids <- maxbin.hist$mids\n \n totaln <- sum(maxbin.hist$counts)\n Xj <- binmids\n Yj <- maxbin.hist$counts / (totaln*binwidth)\n\n ##set bandwidth (nearest neighbors); #nn <-20; this is argument in func call\n\n ##first step histogram\n plot(Xj, Yj, xlab=\"Bin midpoints\", ylab=\"Normalized cell size\", main=paste(\"First step histogram for McCrary(2008)\\nbinwidth=\",binwidth, \" bandwidth=\", nn, sep=\"\"))\n \n ##local fit right and plot\n right.locfit <- locfit.raw(lp(Xj[Xj > 0], h=nn, deg=1), Yj[Xj > 0], kern='tria', family=\"gaussian\")\n plot(right.locfit, add=TRUE, col=\"red\")\n \n ###weighted reg around boundary point with triang kernel to get intercept term \n wt <- f.triangle((Xj[Xj > 0]-rep(0, length( Xj[Xj > 0])))/nn)\n wt <- (wt/(sum(wt)))*length(wt)\n templm <- lm(Yj[Xj > 0]~Xj[Xj > 0], weight=wt)\n fr <- coef(templm)[1]\n \n #local fit left and plot\n left.locfit <- locfit.raw(lp(Xj[Xj <= 0], h=nn, deg=1), Yj[Xj <= 0], kern='tria', family=\"gaussian\")\n plot(left.locfit, add=TRUE, col=\"red\")\n \n ###weighted reg around boundary point with trian kernel to get intercept term\n wt <- f.triangle((Xj[Xj <= 0]-rep(0,length(Xj[Xj <= 0])))/nn)\n wt <- (wt/(sum(wt)))*length(wt) #like aweights in stata\n templm <- lm(Yj[Xj <= 0]~Xj[Xj <= 0], weight=wt)\n fl <- coef(templm)[1]\n \n abline(v=0, lty=2, lwd=2, col=\"darkgrey\")\n \n thetahat <- log(fr) - log(fl)\n sethetahat <- sqrt( (1/(totaln*nn)) * (24/5) * ((1/fr) + (1/fl)) )\n\n legend(\"topright\", bg=\"white\", cex=.9, pt.cex=.9, c(\"Cutoff\", \"Bin-mid regression with triangular kernel\"), lty=c(2, 1), lwd=c(2, 1), col=c(\"darkgrey\", \"red\"))\n text(-maxbin+30, max(Yj)*.92, paste(\"theta-hat =\", round(thetahat, 4), \"\\nstd. err. = \", round(sethetahat,4)))\n return(cat(binwidth, \"\\t\", maxbin, \"\\t\", nn, \"\\t\", thetahat, \"\\t\", sethetahat, \"\\n\"))\n}\n\n\n#####function to generate prop tests to look at subgroup covariate differences around cutoff (like gifted and atrisk)\nprettyprop <- function(binwidth, maxbin, covariate, emcp=emc, lab=\"\") {\n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n \n covariate <- covariate[emcdum==1]\n cutbybreak <- cut(emcp[emcdum==1], breaks=breaks)\n\n afters<-which(cutbybreak %in% tail(levels(cutbybreak), maxbin))\n befores<-which(cutbybreak %in% head(levels(cutbybreak), maxbin))\n \n x1 <- sum(covariate[befores], na.rm=TRUE)\n x2 <- sum(covariate[afters], na.rm=TRUE)\n n1 <- length(na.omit(covariate[befores]))\n n2 <- length(na.omit(covariate[afters]))\n \n pt <- prop.test(c(x1,x2), c(n1, n2))\n return(paste(lab, round(pt$estimate[1], 3), round(pt$estimate[2], 3), round(diff(pt$estimate), 3), round(pt$p.value, 3), n1, n2, sep=', '))\n}\n\n######## similar to prettyprop, but for variables that take a t-test\nttest <- function(binwidth, maxbin, covariate, emcp=emc, lab=\"\") { \n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n \n covariate <- covariate[emcdum==1]\n cutbybreak <- cut(emcp[emcdum==1], breaks=breaks)\n \n afters<-which(cutbybreak %in% tail(levels(cutbybreak), maxbin))\n befores<-which(cutbybreak %in% head(levels(cutbybreak), maxbin))\n \n x1 <- covariate[befores]\n x2 <- covariate[afters]\n tt <- t.test(x1, x2)\n\n return(paste(lab, round(tt$estimate[1], 3), round(tt$estimate[2], 3), round(-diff(tt$estimate), 3), round(tt$p.value, 3), length(na.omit(x1)), length(na.omit(x2)), sep=\",\"))\n}\n\n\ncovfunc <- function(binwidth, maxbin, covariate, ylabcov, use.loess=FALSE, emcp=emc, YLIM=TRUE, nn=10, g) {\n \n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n \n cutbybreak <- cut(emcp[emcdum==1], breaks=breaks)\n binmids <- hist(emcp[emcdum==1], breaks=breaks, plot=FALSE)$mids\n \n binned.covariate <- tapply(covariate[emcdum==1], cutbybreak, function(x) mean(x, na.rm=TRUE))\n forcing <- ifelse(binmids>0, 1, 0)\n tx <- which(forcing==1)\n forcing.sample <- ifelse(emcp>0, 1, 0)\n \n if (YLIM) { \n plot(binned.covariate~binmids, col=NULL, xlab=\"Birth date (centered)\", ylab=ylabcov, main=paste(ylabcov, \" Grade \",g, \"\\nBinwidth=\",binwidth, \", \", length(breaks)-1, \" bins\", sep=\"\"), xaxt='n', ylim=c(0, 1))\n } else plot(binned.covariate~binmids, col=NULL, xlab=\"Birth date (centered)\", ylab=ylabcov, main=paste(ylabcov, \" Grade \",g, \"\\nBinwidth=\",binwidth, \", \", length(breaks)-1, \" bins\", sep=\"\"), axes=FALSE)\n \n points(binmids[tx], binned.covariate[tx], col=\"blue4\", pch=4)\n points(binmids[-tx], binned.covariate[-tx], col=\"darkorange\", pch=16)\n axis(1, at=seq(-maxbin, maxbin, as.integer(maxbin/10)))\n \n if (!YLIM) axis(2, at=round(seq(min(binned.covariate, na.rm=TRUE), max(binned.covariate, na.rm=TRUE), diff(range(binned.covariate, na.rm=TRUE))/10),2))\n \n abline(v=0, lty=2, lwd=2, col=\"darkgrey\")\n \n if (ylabcov %in% covstlab) {\n pval <- ttest(1, binwidth, covariate, emcp)\n } else {\n pval <- prettyprop(1, binwidth, covariate, emcp)\n }\n pvalsplit <- as.numeric(unlist(strsplit(pval, split=\",\")))\n cat(pvalsplit, \"\\n\")\n \n if (use.loess) { \n f.triangle <- function(x) { ifelse(abs(x)<1, 1-abs(x), 0) }\n Yj <- binned.covariate\n Xj <- binmids[which(!is.na(Yj))]\n Yj <- binned.covariate[which(!is.na(Yj))] \n \n #changed so this is now argument\n #nn<-binwidth*3\n \n ##local fit right and plot\n right.locfit <- locfit.raw(lp(Xj[Xj > 0], h=nn, deg=1), Yj[Xj > 0], kern=\"tria\", family=\"gaussian\")\n plot(right.locfit, add=TRUE, col=\"green\") \n #local fit left and plot\n left.locfit <- locfit.raw(lp(Xj[Xj <= 0], h=nn, deg=1), Yj[Xj <= 0], kern='tria', family=\"gaussian\")\n plot(left.locfit, add=TRUE, col=\"green\") \n \n text(-maxbin/2, quantile(binned.covariate, na.rm=TRUE)[4], cex=0.8, paste(\"1 binwidth difference at cutoff =\", round(pvalsplit[4],3), \"\\np-val=\", round(pvalsplit[5],3)))\n #Add legend.\n legend(\"topright\", bg=\"white\", cex=.9, pt.cex=.9, c(\"Older\", \"Younger\", \"Cutoff\", \"Piecewise loess\"), lty=c(NA, NA, 2, 1), lwd=c(NA, NA, 2, 1), col=c(\"blue4\", \"darkorange\", \"darkgrey\", \"green\"), pch=c(4, 16, NA, NA, NA)) \n } else {\n ##run simple regression model and add curves to plot\n model <- lm(covariate~forcing.sample + emcp + I(emcp^2), subset=emcdum==1)\n coefs <- coefficients(model)\n curve(coefs[1] + coefs[3]*x + coefs[4]*x^2, -maxbin, 0, add=T, col=\"red\")\n curve(coefs[1] + coefs[2] + coefs[3]*x + coefs[4]*x, 0, maxbin, add=T, col=\"red\")\n legend(\"topright\", bg=\"white\", cex=.9, pt.cex=.9, c(\"Older\", \"Younger\", \"Cutoff\", \"Piecewise OLS\"), lty=c(NA, NA, 2, 1), lwd=c(NA, NA, 2, 1), col=c(\"blue4\", \"darkorange\", \"darkgrey\", \"red\"), pch=c(4, 16, NA, NA))\n text(-maxbin/2, quantile(binned.covariate, na.rm=TRUE)[4], cex=0.8, paste(\"OLS difference at cutoff = \", round(coefs[2], 3), \"(\", round(sqrt(vcov(model)[2,2]), 3), \")\",sep=\"\")) \n }\n \n}\n\n\noutvforce <- function(binwidth, maxbin, covariate, ylabcov, use.loess=FALSE, emcp=emc, YLIM=TRUE, nn=10, g, change=FALSE) { \n\n ###Plot of outcome vs. forcing variable\n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n\n cutbybreak <- cut(emcp[emcdum==1], breaks=breaks)\n binmids <- hist(emcp[emcdum==1], breaks=breaks, plot=FALSE)$mids\n\n ##group means with binwidth of ; midpoints for horiz axis supplied by hist\n binned.covariate <- tapply(covariate[emcdum==1], cutbybreak, function(x) mean(x, na.rm=TRUE))\n forcing <- ifelse(binmids>0, 1, 0)\n tx <- which(forcing==1)\n forcing.sample <- ifelse(emcp>0, 1, 0)\n\n if (change) {\n maintit=paste(ylabcov, \" Grade \",g-1, \" to \", g, \"\\nBinwidth=\",binwidth, \", \", length(breaks)-1, \" bins\", sep=\"\")\n } else maintit=paste(ylabcov, \" Grade \",g, \"\\nBinwidth=\",binwidth, \", \", length(breaks)-1, \" bins\", sep=\"\")\n \n if (YLIM) {\n plot(binned.covariate~binmids, col=NULL, xlab=\"Birth date (centered)\", ylab=ylabcov, main=maintit, xaxt='n', ylim=c(0, 1))\n } else plot(binned.covariate~binmids, col=NULL, xlab=\"Birth date (centered)\", ylab=ylabcov, main=maintit, axes=FALSE)\n\n points(binmids[tx], binned.covariate[tx], col=\"blue4\", pch=4)\n points(binmids[-tx], binned.covariate[-tx], col=\"darkorange\", pch=16)\n axis(1, at=seq(-maxbin, maxbin, as.integer(maxbin/10)))\n\n if (!YLIM) axis(2, at=round(seq(min(binned.covariate, na.rm=TRUE), max(binned.covariate, na.rm=TRUE), diff(range(binned.covariate, na.rm=TRUE))/10),2))\n\n abline(v=0, lty=2, lwd=2, col=\"darkgrey\")\n\n\n if (ylabcov %in% covstlab) {\n pval <- ttest(1, binwidth, covariate, emcp)\n } else {\n pval <- prettyprop(1, binwidth, covariate, emcp)\n }\n pvalsplit <- as.numeric(unlist(strsplit(pval, split=\",\")))\n cat(pvalsplit, \"\\n\")\n\n ### LAST MOD IS HERE\n if (use.loess) { \n f.triangle <- function(x) { ifelse(abs(x)<1, 1-abs(x), 0) }\n Yj <- binned.covariate\n Xj <- binmids[which(!is.na(Yj))]\n Yj <- binned.covariate[which(!is.na(Yj))] \n\n #changed so this is now argument\n #nn<-binwidth*3 \n \n ##local fit right and plot\n right.locfit <- locfit.raw(lp(Xj[Xj > 0], h=nn, deg=1), Yj[Xj > 0], kern='tria', family=\"gaussian\")\n plot(right.locfit, add=TRUE, col=\"green\")\n \n #local fit left and plot\n left.locfit <- locfit.raw(lp(Xj[Xj <= 0], h=nn, deg=1), Yj[Xj <= 0], kern='tria', family=\"gaussian\")\n plot(left.locfit, add=TRUE, col=\"green\") \n \n text(-maxbin/2, quantile(binned.covariate, na.rm=TRUE)[4], paste(\"1 binwidth difference at cutoff =\", round(pvalsplit[4],3), \"\\np-val=\", round(pvalsplit[5],3))) \n #Add legend.\n legend(\"topright\", bg=\"white\", cex=.9, pt.cex=.9, c(\"Older\", \"Younger\", \"Cutoff\", \"Piecewise loess\"), lty=c(NA, NA, 2, 1), lwd=c(NA, NA, 2, 1), col=c(\"blue4\", \"darkorange\", \"darkgrey\", \"green\"), pch=c(4, 16, NA, NA, NA)) \n } else {\n ##run simple regression model and add curves to plot\n model <- lm(binned.covariate~forcing + binmids + I(binmids^2), subset=emcdum==1)\n #model <- lm(covariate~forcing.sample + emcp + I(emcp^2), subset=emcdum==1)\n coefs <- coefficients(model)\n curve(coefs[1] + coefs[3]*x + coefs[4]*x^2, -maxbin, 0, add=T, col=\"red\")\n curve(coefs[1] + coefs[2] + coefs[3]*x + coefs[4]*x, 0, maxbin, add=T, col=\"red\")\n legend(\"topright\", bg=\"white\", cex=.9, pt.cex=.9, c(\"Older\", \"Younger\", \"Cutoff\", \"Piecewise OLS\"), lty=c(NA, NA, 2, 1), lwd=c(NA, NA, 2, 1), col=c(\"blue4\", \"darkorange\", \"darkgrey\", \"red\"), pch=c(4, 16, NA, NA))\n text(-maxbin/2, quantile(binned.covariate, na.rm=TRUE)[4], cex=0.8, paste(\"OLS difference\\nat cutoff = \", round(coefs[2], 3), \"(\", round(sqrt(vcov(model)[2,2]), 3), \")\",sep=\"\")) \n }\n}\n\n##********************************************************************COVARIATES******************\n#x11(height=10, width=10)\n\nclx <- function(fm, dfcw, cluster){\n # R-codes (www.r-project.org) for computing\n # clustered-standard errors. Mahmood Arai, Jan 26, 2008.\n \n\t # The arguments of the function are:\n # fitted model, cluster1 and cluster2\n # You need to install libraries `sandwich' and `lmtest'\n\t \n # reweighting the var-cov matrix for the within model\n #library(sandwich);library(lmtest)\n M <- length(unique(cluster)) \n N <- length(cluster) \n K <- fm$rank \n dfc <- (M/(M-1))*((N-1)/(N-K)) \n uj <- apply(estfun(fm),2, function(x) tapply(x, cluster, sum));\n vcovCL <- dfc*sandwich(fm, meat=crossprod(uj)/N)*dfcw\n coeftest(fm, vcovCL) \n}\n\nmclx <- function(fm, dfcw, cluster1, cluster2){\n # R-codes (www.r-project.org) for computing multi-way \n # clustered-standard errors. Mahmood Arai, Jan 26, 2008. \n # See: Thompson (2006), Cameron, Gelbach and Miller (2006)\n # and Petersen (2006).\n\t # reweighting the var-cov matrix for the within model\n \n # The arguments of the function are:\n # fitted model, cluster1 and cluster2\n # You need to install libraries `sandwich' and `lmtest'\n \n library(sandwich);library(lmtest)\n cluster12 = paste(cluster1,cluster2, sep=\"\")\n M1 <- length(unique(cluster1))\n M2 <- length(unique(cluster2)) \n M12 <- length(unique(cluster12))\n N <- length(cluster1) \n K <- fm$rank \n dfc1 <- (M1/(M1-1))*((N-1)/(N-K)) \n dfc2 <- (M2/(M2-1))*((N-1)/(N-K)) \n dfc12 <- (M12/(M12-1))*((N-1)/(N-K)) \n u1j <- apply(estfun(fm), 2, function(x) tapply(x, cluster1, sum)) \n u2j <- apply(estfun(fm), 2, function(x) tapply(x, cluster2, sum)) \n u12j <- apply(estfun(fm), 2, function(x) tapply(x, cluster12, sum)) \n vc1 <- dfc1*sandwich(fm, meat=crossprod(u1j)/N )\n vc2 <- dfc2*sandwich(fm, meat=crossprod(u2j)/N )\n vc12 <- dfc12*sandwich(fm, meat=crossprod(u12j)/N)\n vcovMCL <- (vc1 + vc2 - vc12)*dfcw\n coeftest(fm, vcovMCL)\n}\n\nleecard <- function(binwidth, maxbin, outcome, testtype, emcp=emc) {\n ## outcome = const + B1*age + B2*Cutoff + e\n ## outcome = const + B1*emcp + B2*forcing.sample + e\n\n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n\n cutbybreak <- cut(emcp[emcdum==1], breaks=breaks)\n binmids <- hist(emcp[emcdum==1], breaks=breaks, plot=FALSE)$mids\n\n emcbinmids <- findInterval(emcp[emcdum==1], binmids, rightmost.closed=TRUE)\n emcmids <- binmids[emcbinmids]\n \n #binned.outcome <- tapply(outcome[emcdum==1], cutbybreak, function(x) mean(x, na.rm=TRUE))\n #binned.variance <- tapply(outcome[emcdum==1], cutbybreak, function(x) var(x, na.rm=TRUE))\n \n #binned.test <- tapply(outcome[emcdum==1], cutbybreak, function(x) table(!is.na(x)))\n #Nj <- binned.takers <- unlist(lapply(binned.test, function(x) sum(x[\"TRUE\"])))\n #binned.takers <- binned.takers / (sum(binned.takers) / length(binned.takers))\n #nomissbin <- sapply(names(unlist(binned.takers)), function(x) substr(x, start=1, stop=nchar(x)-5))\n \n forcing <- ifelse(binmids>0, 1, 0)\n tx <- which(forcing==1)\n forcing.sample <- ifelse(emcp>0, 1, 0)\n\n #leftcut <- sort(getbins(binwidth, maxbin)@left)[1]\n #rightcut <- sort(getbins(binwidth, maxbin)@right, decreasing=TRUE)[1]\n #rightobs <- ifelse(emc > 0 & emc <= rightcut, 1, 0)\n #leftobs <- ifelse(emc <= 0 & emc > leftcut, 1, 0)\n #obs <- ifelse(emc >= leftcut & emc <= rightcut, 1, 0)\n test <- outcome[emcdum==1][emcbinmids!=0]\n alpha <- forcing.sample[emcdum==1][emcbinmids!=0]\n\n ##for GOF test\n Unrestricted <- lm(test~alpha+as.factor(emcmids))\n #coef on forcing.sample is NA\n J <- Unrestricted$rank\n N <- dim(Unrestricted$model)[1]\n ESSur <- sum(Unrestricted$residuals^2)\n \n \n ####global polynomial models over microdata with cluster consistent standard errors and a GOF test (Card and Lee 2008)\n #p0.T <- lm(outcome~forcing.sample, subset=emcdum==1)\n #clx(p0.T, 1, clustersub) #cluster consistent standard errors\n #sqrt(vcovHC(p0.T, \"HC\")) #robust Heteroskedasticity-consistent standard errors\n #K <- p0.T$rank\n #ESSr <- sum(p0.T$residuals^2)\n #G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n #print(pval <- pf(G, J-K, N-J)) GOODNESS OF FIT TEST \n \n cat(\"LeeCard GOF Tests\", \"P-values from F-test\", \"Polynomial order\\n\", sep=\", \")\n p1.T <- lm(test~alpha+emcmids)\n clustersub <- emcp[attributes(p1.T$model)$row.names]\n c1<-clx(p1.T, 1, clustersub)\n K <- p1.T$rank \n ESSr <- sum(p1.T$residuals^2)\n G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n cat(pval <- pf(G, J-K, N-J), K-2, \"\\n\", sep=\", \") #\n \n p2.T <- p3.T <- update(p1.T, . ~ . + I(emcmids^2))\n clustersub <- emcp[attributes(p2.T$model)$row.names]\n c2<-clx(p2.T, 1, clustersub)\n K <- p2.T$rank \n ESSr <- sum(p2.T$residuals^2)\n G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n cat(pval <- pf(G, J-K, N-J), K-2, \"\\n\", sep=\", \")\n \n p3.T <- update(p2.T, . ~ . + I(emcmids^3))\n clustersub <- emcp[attributes(p3.T$model)$row.names]\n c3<-clx(p3.T, 1,clustersub)\n K <- p3.T$rank \n ESSr <- sum(p3.T$residuals^2)\n G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n cat(pval <- pf(G, J-K, N-J), K-2, \"\\n\", sep=\", \")\n\n p4.T <- update(p3.T, . ~ . + I(emcmids^4))\n clustersub <- emcp[attributes(p4.T$model)$row.names]\n clx(p4.T, 1, clustersub)\n K <- p4.T$rank \n ESSr <- sum(p4.T$residuals^2)\n G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n cat(pval <- pf(G, J-K, N-J), K-2, \"\\n\", sep=\", \")\n \n p5.T <- update(p4.T, . ~ . + I(emcmids^5))\n clustersub <- emcp[attributes(p5.T$model)$row.names]\n clx(p5.T, 1, clustersub)\n K <- p5.T$rank \n ESSr <- sum(p5.T$residuals^2)\n G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n cat(pval <- pf(G, J-K, N-J), K-2, \"\\n\", sep=\", \") \n \n p6.T <- update(p5.T, . ~ . + I(emcmids^6))\n clustersub <- emcp[attributes(p6.T$model)$row.names]\n clx(p6.T, 1, clustersub)\n K <- p6.T$rank \n ESSr <- sum(p6.T$residuals^2)\n G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n cat(pval <- pf(G, J-K, N-J), K-2, \"\\n\", sep=\", \") \n \n p7.T <- update(p6.T, . ~ . + I(emcmids^7))\n clustersub <- emcp[attributes(p7.T$model)$row.names]\n clx(p7.T, 1, clustersub)\n K <- p7.T$rank \n ESSr <- sum(p7.T$residuals^2)\n G = ((ESSr-ESSur)/(J-K)) / (ESSur / (N-J)) # ~F(J-K, N-J)\n cat(pval <- pf(G, J-K, N-J), K-2, \"\\n\", sep=\", \")\n \n ################################### Global fits end\n\n ## regs run on collapsed to cell level data with heteroskedasticity-consistent standard errors.\n ##Also include variance correction factor as in Card and Lee 2008 equations 12 and 13.\n ## Omit for now\n \n #p0.C <- lm(binned.outcome~forcing, weights=binned.takers)\n #sqrt(vcovHC(p0.C, \"HC\")) #robust Heteroskedasticity-consistent standard errors: about equal to clx(p0.T, 1, clustersub)\n #sigmahat2_a <- 2*(sum(p0.C$residuals^2)/sum(Nj) - sum(binned.variance, na.rm=T)/sum(Nj)) #maybe this one?\n #newstderror <- vcovHC(p0.C, \"HC\")[2,2] + sigmahat2_a\n\n #p1.C <- lm(binned.outcome~forcing*breaks[2:81], weights=binned.takers)\n #sqrt(vcovHC(p1.C, \"HC\"))\n #sigmahat2_a <- 2*(sum(p1.C$residuals^2 * Nj[1:79]) / sum(Nj) - sum(binned.variance, na.rm=T)/sum(Nj)) #maybe this one?\n #newstderror <- vcovHC(p1.C, \"HC\")[2,2] + sigmahat2_a\n\n #p2.C <- lm(outcome~forcing.sample*emc + forcing.sample*I(emc^2), subset=obs==1)\n return(list(c1, c2, c3))\n}\n\n\nsrdregs <- function(binwidth, maxbin, outcome, testtype, emcp=emc, coefs=1, use.coefs=FALSE) {\n ## outcome = const + B1*age + B2*Cutoff + e\n ## outcome = const + B1*emcp + B2*forcing.sample + e\n\n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n\n cutbybreak <- cut(emcp[emcdum==1], breaks=breaks)\n binmids <- hist(emcp[emcdum==1], breaks=breaks, plot=FALSE)$mids\n\n emcbinmids <- findInterval(emcp[emcdum==1], binmids, rightmost.closed=TRUE)\n emcmids <- binmids[emcbinmids]\n \n forcing <- ifelse(binmids>0, 1, 0)\n #tx <- which(forcing==1)\n forcing.sample <- ifelse(emcp>0, 1, 0)\n\n test <- outcome[emcdum==1][emcbinmids!=0]\n alpha <- forcing.sample[emcdum==1][emcbinmids!=0]\n if (use.coefs){\n beta <- coefs[emcdum==1,][emcbinmids!=0,]\n sr.model <- lm(test~alpha+emcmids+I(emcmids^2)+beta)\n } else sr.model <- lm(test~alpha+emcmids+I(emcmids^2))\n \n clustersub <- emcp[attributes(sr.model$model)$row.names]\n adjrsq <- summary(sr.model)$adj.r.squared\n\n ##temporary fix for now until I figure out prob with sandwich estimator\n a<-try(print(clx(sr.model, 1, clustersub)), silent=TRUE)\n if (class(a)==\"try-error\") print(summary(sr.model))\n cat(\"N=\",length(clustersub), \" Adj. R-sq=\", adjrsq, '\\n\\n', sep='')\n \n\n}\n\nfrdregs <- function(binwidth, maxbin, outcome, testtype, emcp=emc, coefs=1, use.coefs=FALSE, instr) {\n ## outcome = const + B1*age + B2*Cutoff + e\n ## outcome = const + B1*emcp + B2*forcing.sample + e\n ## forcing.sample = const+A1*age\n\n breaks <- getbins(binwidth, maxbin)@breaks\n emcdum <- ifelse(emcp >= min(breaks) & emcp <= max(breaks), 1, 0)\n\n cutbybreak <- cut(emcp[emcdum==1], breaks=breaks)\n binmids <- hist(emcp[emcdum==1], breaks=breaks, plot=FALSE)$mids\n\n emcbinmids <- findInterval(emcp[emcdum==1], binmids, rightmost.closed=TRUE)\n emcmids <- binmids[emcbinmids]\n \n forcing <- ifelse(binmids>0, 1, 0)\n #tx <- which(forcing==1)\n forcing.sample <- ifelse(emcp>0, 1, 0)\n\n test <- outcome[emcdum==1][emcbinmids!=0]\n alpha <- forcing.sample[emcdum==1][emcbinmids!=0]\n instrument <- instr[emcdum==1][emcbinmids!=0]\n \n if (use.coefs){\n beta <- coefs[emcdum==1,][emcbinmids!=0,]\n mf <- data.frame(test, alpha, instrument, emcmids, beta)\n form1 <- as.formula(paste(\"test~emcmids+I(emcmids^2)+alpha+\",\n paste(colnames(mf[,5:dim(mf)[2]]), collapse=\"+\"),\n \" | instrument+emcmids+I(emcmids^2)+\",paste(colnames(mf[,5:dim(mf)[2]]), collapse=\"+\"), sep=''))\n sr.model <- ivreg(form1, data=mf)\n } else {\n mf <- data.frame(test, alpha, instrument, emcmids)\n sr.model <- ivreg(test~emcmids+I(emcmids^2)+alpha | instrument+emcmids+I(emcmids^2), data=mf)\n }\n\n clustersub <- emcp[attributes(sr.model$model)$row.names]\n adjrsq <- summary(sr.model)$adj.r.squared\n\n ##temporary fix for now until I figure out prob with sandwich estimator\n a<-try(print(clx(sr.model, 1, clustersub)), silent=TRUE)\n if (class(a)==\"try-error\") print(summary(sr.model))\n cat(\"N=\",length(clustersub), \" Adj. R-sq=\", adjrsq, '\\n\\n', sep='')\n \n\n}\n \n##$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ FUNCTIONS END $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$##\n", "meta": {"hexsha": "9de453e8878dab8cd2e777c13fd042ae07aa443a", "size": 23557, "ext": "r", "lang": "R", "max_stars_repo_path": "rd_func.r", "max_stars_repo_name": "pmcrosta/misc", "max_stars_repo_head_hexsha": "9adb55b2aa3bdcd1d16162ce4946124d51b305d2", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "rd_func.r", "max_issues_repo_name": "pmcrosta/misc", "max_issues_repo_head_hexsha": "9adb55b2aa3bdcd1d16162ce4946124d51b305d2", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "rd_func.r", "max_forks_repo_name": "pmcrosta/misc", "max_forks_repo_head_hexsha": "9adb55b2aa3bdcd1d16162ce4946124d51b305d2", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 45.0420650096, "max_line_length": 240, "alphanum_fraction": 0.6079721527, "num_tokens": 8432, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7718434978390747, "lm_q2_score": 0.6959583187272711, "lm_q1q2_score": 0.5371709030766585}} {"text": "## ----echo=FALSE---------------------------------------------------------------\nrequire(Hmisc)\nknitrSet('continuousY', cache=TRUE)\nknitr::read_chunk('~/doc/rms/continuousY/shared.R')\n\n\n## ----lmassump,h=3.25,w=7,mfrow=c(1,2),ps=10,mgp=c(.5, .365, 0),left=1,cap='Assumptions of the linear model (left panel) and semiparametric ordinal probit or logit (proportional odds) models (right panel). Ordinal models do not assume any shape for the distribution of $Y$ for a given $X$; they only assume parallelism. The linear model can relax the parallelism assumption if $\\\\sigma$ is allowed to vary, but in practice it is difficult to know how to vary it except for the unequal variance two-sample $t$-test.',scap='Assumptions of linear vs.\\\\ semiparametric models',echo=FALSE----\npinv <- expression(paste(Phi^{-1}, '(F(y', '|', 'X))'))\nplot(0, 0, xlim=c(0, 1), ylim=c(-2, 2), type='n', axes=FALSE,\n xlab=expression(y), ylab='')\nmtext(pinv, side=2, line=1)\naxis(1, labels=FALSE, lwd.ticks=0)\naxis(2, labels=FALSE, lwd.ticks=0)\nabline(a=-1.5, b=1)\nabline(a=0, b=1)\narrows(.5, -1.5+.5, .5, 0+.5, code=3, length=.1)\ntext(.525, .5*(-1.5+.5+.5), expression(-Delta*X*beta/sigma), adj=0)\ng <- function(x) -2.2606955+11.125231*x-37.772783*x^2+56.776436*x^3-\n 26.861103*x^4\nx <- seq(0, .9, length=150)\npinv <- expression(atop(paste(Phi^{-1}, '(F(y', '|', 'X))'),\n paste(logit, '(F(y', '|', 'X))')))\nplot(0, 0, xlim=c(0, 1), ylim=c(-2, 2), type='n', axes=FALSE,\n xlab=expression(y), ylab='')\nmtext(pinv, side=2, line=1)\naxis(1, labels=FALSE, lwd.ticks=FALSE)\naxis(2, labels=FALSE, lwd.ticks=FALSE)\nlines(x, g(x))\nlines(x, g(x)+1.5)\narrows(.5, g(.5), .5, g(.5)+1.5, code=3, length=.1)\ntext(.525, .5*(g(.55) + g(.55)+1.5), expression(-Delta*X*beta), adj=0)\n\n\n## ----desc,results='asis',cache=FALSE------------------------------------------\nrequire(rms)\noptions(prType='latex') # for print, summary, anova\ngetHdata(nhgh)\nw <- subset(nhgh, age >= 21 & dx==0 & tx==0, select=-c(dx,tx))\nlatex(describe(w), file='')\ndd <- datadist(w); options(datadist='dd')\n\n\n## ----lookdist,w=5.5,h=5.5,cap='Examination of normality and constant variance assumption, and assumptions for various ordinal models',scap='Examining normality and ordinal model assumptions'----\nf <- ols(gh ~ rcs(age,5) + sex + re + rcs(bmi, 3), data=w)\npgh <- fitted(f)\n\np <- function(fun, row, col) {\n f <- substitute(fun); g <- function(F) eval(f)\n z <- Ecdf(~ gh, groups=cut2(pgh, g=6),\n fun=function(F) g(1 - F),\n ylab=as.expression(f), xlim=c(4.5, 7.75), data=w,\n label.curve=FALSE)\n print(z, split=c(col, row, 2, 2), more=row < 2 | col < 2)\n}\np(log(F/(1-F)), 1, 1)\np(qnorm(F), 1, 2)\np(-log(-log(F)), 2, 1)\np(log(-log(1-F)), 2, 2)\n# Get slopes of pgh for some cutoffs of Y\n# Use glm complementary log-log link on Prob(Y < cutoff) to\n# get log-log link on Prob(Y >= cutoff)\nr <- NULL\nfor(link in c('logit','probit','cloglog'))\n for(k in c(5, 5.5, 6)) {\n co <- coef(glm(gh < k ~ pgh, data=w, family=binomial(link)))\n r <- rbind(r, data.frame(link=link, cutoff=k,\n slope=round(co[2],2)))\n}\nprint(r, row.names=FALSE)\n\n\n## ----comparemany,cache=TRUE,h=6.5,w=6.75,cap='Three estimated quantiles and estimated mean using 6 methods, compared against caliper-matched sample quantiles/means (circles). Numbers are mean absolute differences between predicted and sample quantities using overlapping intervals of age and caliper matching\\\\index{caliper matching}. QR:quantile regression.',scap='Six methods for estimating quantiles or means.'----\nag <- 25:75\nlag <- length(ag)\nq2 <- q3 <- p90 <- means <- numeric(lag)\nfor(i in 1:lag) {\n s <- which(abs(w$age - ag[i]) < 5)\n y <- w$gh[s]\n a <- quantile(y, probs=c(.5, .75, .9))\n q2[i] <- a[1]\n q3[i] <- a[2]\n p90[i] <- a[3]\n means[i] <- mean(y)\n}\nfams <- c('logistic', 'probit', 'loglog', 'cloglog')\nfe <- function(pred, target) mean(abs(pred$yhat - target))\nmod <- gh ~ rcs(age,6)\nP <- Er <- list()\nfor(est in c('q2', 'q3', 'p90', 'mean')) {\n meth <- if(est == 'mean') 'ols' else 'QR'\n p <- list()\n er <- rep(NA, 5)\n names(er) <- c(fams, meth)\n for(family in fams) {\n h <- orm(mod, family=family, data=w)\n fun <- if(est == 'mean') Mean(h)\n else {\n qu <- Quantile(h)\n switch(est, q2 = function(x) qu(.5, x),\n q3 = function(x) qu(.75, x),\n p90 = function(x) qu(.9, x))\n }\n p[[family]] <- z <- Predict(h, age=ag, fun=fun, conf.int=FALSE)\n er[family] <- fe(z, switch(est, mean=means, q2=q2, q3=q3, p90=p90))\n }\n h <- switch(est,\n mean= ols(mod, data=w),\n q2 = Rq (mod, data=w),\n q3 = Rq (mod, tau=0.75, data=w),\n p90 = Rq (mod, tau=0.90, data=w))\n p[[meth]] <- z <- Predict(h, age=ag, conf.int=FALSE)\n er[meth] <- fe(z, switch(est, mean=means, q2=q2, q3=q3, p90=p90))\n\n Er[[est]] <- er\n pr <- do.call('rbind', p)\n pr$est <- est\n P <- rbind.data.frame(P, pr)\n}\n\nxyplot(yhat ~ age | est, groups=.set., data=P, type='l', # Figure (*\\ref{fig:continuousY-comparemany}*)\n auto.key=list(x=.75, y=.2, points=FALSE, lines=TRUE),\n panel=function(..., subscripts) {\n panel.xyplot(..., subscripts=subscripts)\n est <- P$est[subscripts[1]]\n lpoints(ag, switch(est, mean=means, q2=q2, q3=q3, p90=p90),\n col=gray(.7))\n er <- format(round(Er[[est]],3), nsmall=3)\n ltext(26, 6.15, paste(names(er), collapse='\\n'),\n cex=.7, adj=0)\n ltext(40, 6.15, paste(er, collapse='\\n'),\n cex=.7, adj=1)})\n\n\n## ----predobs,w=4.75,h=4,cap='Observed (dashed lines, open circles) and predicted (solid lines, closed circles) exceedance probability distributions from a model using 6-tiles of OLS-predicted \\\\hba. Key shows quantile group intervals of predicted mean \\\\hba.',scap='Observed and predicted distributions'----\nw$pghg <- cut2(pgh, g=6)\nf <- orm(gh ~ pghg, family=loglog, data=w)\nlp <- predict(f, newdata=data.frame(pghg=levels(w$pghg)))\nep <- ExProb(f) # Exceedance prob. functn. generator in rms\nz <- ep(lp)\nj <- order(w$pghg) # puts in order of lp (levels of pghg)\nplot(z, xlim=c(4, 7.5), data=w[j,c('pghg', 'gh')]) # Fig. (*\\ref{fig:continuousY-predobs}*)\n\n\n## ----lookprobit,w=3,h=2.5,bty='l',cap='Estimated intercepts from probit model. Linearity would have indicated Gaussian residuals.',scap='Estimated intercepts from probit model'----\nf <- orm(gh ~ rcs(age,6), family=probit, data=w)\ng <- ols(gh ~ rcs(age,6), data=w)\ns <- g$stats['Sigma']\nyu <- f$yunique[-1]\nr <- quantile(w$gh, c(.005, .995))\nalphas <- coef(f)[1:num.intercepts(f)]\nplot(-yu / s, alphas, type='l', xlim=rev(- r / s), # Fig. (*\\ref{fig:continuousY-lookprobit}*)\n xlab=expression(-y/hat(sigma)), ylab=expression(alpha[y]))\n\n\n## ----htwtcoef,results='asis'--------------------------------------------------\nf <- orm(gh ~ rcs(age,5) + log(ht) + log(wt),\n family=loglog, data=w)\nf\n\n## ----aichtwt------------------------------------------------------------------\naic <- NULL\nfor(mod in list(gh ~ rcs(age,5) + rcs(log(bmi),5),\n gh ~ rcs(age,5) + rcs(log(ht),5) + rcs(log(wt),5),\n gh ~ rcs(age,5) + rcs(log(ht),4) * rcs(log(wt),4)))\n aic <- c(aic, AIC(orm(mod, family=loglog, data=w)))\nprint(aic)\n\n\n## ----coxhtwtcoef,results='asis'-----------------------------------------------\nprint(cph(Surv(gh) ~ rcs(age,5) + log(ht) + log(wt), data=w))\n\n\n## ----redun,cap='Variable clustering for all potential predictors'-------------\nv <- varclus(~ wt + ht + bmi + leg + arml + armc + waist +\n tri + sub + age + sex + re, data=w)\nplot(v) # Figure (*\\ref{fig:continuousY-redun}*)\n# Omit wt so it won't be removed before bmi\nredun(~ ht + bmi + leg + arml + armc + waist + tri + sub,\n data=w, r2=.75)\n\n\n## ----htchange,fig.align='right',cap=\"Estimated median height as a smooth function of age, allowing age to interact with sex, from a proportional odds model\",scap=\"Median height vs.\\\\ age\"----\nf <- orm(ht ~ rcs(age,4)*sex, data=w) # Prop. odds model\nqu <- Quantile(f); med <- function(x) qu(.5, x)\nggplot(Predict(f, age, sex, fun=med, conf.int=FALSE),\n ylab='Predicted Median Height, cm')\n\n\n## ----allocadf,w=4,cap='Generalized squared rank correlations',top=1-----------\ns <- spearman2(gh ~ age + sex + re + wt + leg + arml + armc +\n waist + tri + sub, data=w, p=2)\nplot(s)\n\n\n## ----fitfullcasewise,results='asis'-------------------------------------------\nf <- orm(gh ~ rcs(age,5) + sex + re + rcs(wt,3) + rcs(leg,3) + arml +\n rcs(armc,3) + rcs(waist,4) + tri + rcs(sub,3),\n family=loglog, data=w, x=TRUE, y=TRUE)\nprint(f, coefs=FALSE)\n## Composite test:\nanova(f, leg, arml, armc, waist, tri, sub)\n\n\n## ----casewisemeanmed,h=4,w=5,cap='Estimated mean and 0.5 and 0.9 quantiles from the log-log ordinal model using casewise deletion, along with predictions of 0.5 and 0.9 quantiles from quantile regression (QR). Age is varied and other predictors are held constant to medians/modes.',scap='Estimated mean and quantiles from casewise deletion model.'----\nM <- Mean(f)\nqu <- Quantile(f)\nmed <- function(x) qu(.5, x)\np90 <- function(x) qu(.9, x)\nfq <- Rq(formula(f), data=w)\nfq90 <- Rq(formula(f), data=w, tau=.9)\npmean <- Predict(f, age, fun=M, conf.int=FALSE)\npmed <- Predict(f, age, fun=med, conf.int=FALSE)\np90 <- Predict(f, age, fun=p90, conf.int=FALSE)\npmedqr <- Predict(fq, age, conf.int=FALSE)\np90qr <- Predict(fq90, age, conf.int=FALSE)\nz <- rbind('orm mean'=pmean, 'orm median'=pmed, 'orm P90'=p90,\n 'QR median'=pmedqr, 'QR P90'=p90qr)\nggplot(z, groups='.set.',\n adj.subtitle=FALSE, legend.label=FALSE)\n\n## ----prbw---------------------------------------------------------------------\nprint(fastbw(f, rule='p'), estimates=FALSE)\n\n## ----valbworm,cache=TRUE------------------------------------------------------\nset.seed(13) # so can reproduce results\nv <- validate(f, B=100, bw=TRUE, estimates=FALSE, rule='p')\n\n## ----prval,results='asis'-----------------------------------------------------\n# Show number of variables selected in first 30 boots\nlatex(v, B=30, file='', size='small')\n\n\n## ----sanova,cache=TRUE,results='asis',cap='ANOVA for reduced model, after multiple imputation, with addition of a combined effect for four size variables',scap='ANOVA for reduced model'----\na <- aregImpute(~ gh + wt + ht + bmi + leg + arml + armc + waist +\n tri + sub + age +re, data=w, n.impute=5, pr=FALSE)\ng <- fit.mult.impute(gh ~ rcs(age,5) + re + rcs(leg,3) +\n rcs(waist,4) + tri + rcs(sub,4),\n orm, a, family=loglog, data=w, pr=FALSE)\nprint(g, needspace='1.5in')\nan <- anova(g)\nprint(an, caption='ANOVA for reduced model after multiple imputation, with addition of a combined effect for four size variables')\nb <- anova(g, leg, waist, tri, sub)\n# Add new lines to the plot with combined effect of 4 size var.\ns <- rbind(an, size=b['TOTAL', ])\nclass(s) <- 'anova.rms'\nplot(s)\n\n\n## ----peffects,cache=TRUE,results='asis',cap='Partial effects (log hazard or log-log cumulative probability scale) of all predictors in reduced model, after multiple imputation',scap='Partial effects after multiple imputation',w=6.75,h=4.5,cache=TRUE----\nggplot(Predict(g), abbrev=TRUE, ylab=NULL) # Figure (*\\ref{fig:continuousY-peffects}*)\n\n\n## ----cfmissmeth,cache=TRUE----------------------------------------------------\ngc <- orm(gh ~ rcs(age,5) + re + rcs(leg,3) +\n rcs(waist,4) + tri + rcs(sub,4),\n family=loglog, data=w, x=TRUE, y=TRUE)\ngb <- bootcov(gc, B=300)\n\n## ----peffects2,cap='Partial effect for age from multiple imputation (center red line) and casewise deletion (center blue line) with symmetric Wald 0.95 confidence bands using casewise deletion (gray shaded area), basic bootstrap confidence bands using casewise deletion (blue lines), percentile bootstrap confidence bands using casewise deletion (dashed blue lines), and symmetric Wald confidence bands accounting for multiple imputation (red lines).',scap='Partial effect for age with bootstrap and Wald confidence bands',w=5,h=4,bot=1----\nbootclb <- Predict(gb, age, boot.type='basic')\nbootclp <- Predict(gb, age, boot.type='percentile')\nmultimp <- Predict(g, age)\nplot(Predict(gc, age), addpanel=function(...) {\n with(bootclb, {llines(age, lower, col='blue')\n llines(age, upper, col='blue')})\n with(bootclp, {llines(age, lower, col='blue', lty=2)\n llines(age, upper, col='blue', lty=2)})\n with(multimp, {llines(age, lower, col='red')\n llines(age, upper, col='red')\n llines(age, yhat, col='red')} ) },\n col.fill=gray(.9), adj.subtitle=FALSE) # Figure (*\\ref{fig:continuousY-peffects2}*)\n\n\n## ----meanvs,cap='Predicted mean \\\\hba vs.\\\\ predicted median and 0.9 quantile along with their marginal distributions',scap='Predicted mean, median, and 0.9 quantile of \\\\hba',w=4.5,h=3.5----\nM <- Mean(g)\nqu <- Quantile(g)\nmed <- function(lp) qu(.5, lp)\nq90 <- function(lp) qu(.9, lp)\nlp <- predict(g)\nlpr <- quantile(predict(g), c(.002, .998), na.rm=TRUE)\nlps <- seq(lpr[1], lpr[2], length=200)\npmn <- M(lps)\npme <- med(lps)\np90 <- q90(lps)\nplot(pmn, pme, # Figure (*\\ref{fig:continuousY-meanvs}*)\n xlab=expression(paste('Predicted Mean ', HbA[\"1c\"])),\n ylab='Median and 0.9 Quantile', type='l',\n xlim=c(4.75, 8.0), ylim=c(4.75, 8.0), bty='n')\nbox(col=gray(.8))\nlines(pmn, p90, col='blue')\nabline(a=0, b=1, col=gray(.8))\ntext(6.5, 5.5, 'Median')\ntext(5.5, 6.3, '0.9', col='blue')\nnint <- 350\nscat1d(M(lp), nint=nint)\nscat1d(med(lp), side=2, nint=nint)\nscat1d(q90(lp), side=4, col='blue', nint=nint)\n\n\n## ----nomogram,cap='Nomogram for predicting median, mean, and 0.9 quantile of glycohemoglobin, along with the estimated probability that \\\\hba $\\\\ge 6.5, 7$, or $7.5$, all from the log-log ordinal model',scap='Nomogram of log-log ordinal model for \\\\hba',w=6.75,h=5.75,ps=9----\ng <- Newlevels(g, list(re=abbreviate(levels(w$re))))\nexprob <- ExProb(g)\nnom <-\n nomogram(g, fun=list(Mean=M,\n 'Median Glycohemoglobin' = med,\n '0.9 Quantile' = q90,\n 'Prob(HbA1c >= 6.5)'=\n function(x) exprob(x, y=6.5),\n 'Prob(HbA1c >= 7.0)'=\n function(x) exprob(x, y=7),\n 'Prob(HbA1c >= 7.5)'=\n function(x) exprob(x, y=7.5)),\n fun.at=list(seq(5, 8, by=.5),\n c(5,5.25,5.5,5.75,6,6.25),\n c(5.5,6,6.5,7,8,10,12,14),\n c(.01,.05,.1,.2,.3,.4),\n c(.01,.05,.1,.2,.3,.4),\n c(.01,.05,.1,.2,.3,.4)))\nplot(nom, lmgp=.28) # Figure (*\\ref{fig:continuousY-nomogram}*)\n\n", "meta": {"hexsha": "936e854389df265ecd5a95b7cb235af843f7bb41", "size": 14815, "ext": "r", "lang": "R", "max_stars_repo_path": "RMScode/15_continuous_Y.r", "max_stars_repo_name": "fabarrios/Regression", "max_stars_repo_head_hexsha": "a16af97ccc82d32982b33ab38a2956e64b8b67e7", "max_stars_repo_licenses": ["CC0-1.0"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2022-02-01T18:27:09.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-01T18:27:09.000Z", "max_issues_repo_path": "RMScode/15_continuous_Y.r", "max_issues_repo_name": "fabarrios/Regression", "max_issues_repo_head_hexsha": "a16af97ccc82d32982b33ab38a2956e64b8b67e7", "max_issues_repo_licenses": ["CC0-1.0"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2022-01-23T23:59:10.000Z", "max_issues_repo_issues_event_max_datetime": "2022-01-23T23:59:10.000Z", "max_forks_repo_path": "RMScode/15_continuous_Y.r", "max_forks_repo_name": "fabarrios/Regression", "max_forks_repo_head_hexsha": "a16af97ccc82d32982b33ab38a2956e64b8b67e7", "max_forks_repo_licenses": ["CC0-1.0"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2022-03-24T03:43:53.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-24T03:43:53.000Z", "avg_line_length": 46.1526479751, "max_line_length": 590, "alphanum_fraction": 0.5854876814, "num_tokens": 4850, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.734119526900183, "lm_q2_score": 0.731058584489497, "lm_q1q2_score": 0.536684382181747}} {"text": "## load package deSolve for ode simulation\r\nlibrary(deSolve)\r\n\r\n## types of noise: \r\n## 1. process noise: complementary or essential resources in the main text model. core, tail, or general in the model in Appendix D\r\n## 2. measurement noise\r\nnoisetypes=c('complementary','essential','core','tail','general','measurement')\r\n\r\n## scaling factors: scale noise levels in the kernel\r\nscaling.factors=data.frame(noise=c('complementary','essential','core','tail','general','measurement'),scaling=c(30,1,8,250,2,.4))\r\n\r\n## number of species \r\nS=200 \t\t\t\t\t\t\t\t\r\n\r\n## species niche/trait values\r\nset.seed(0)\r\ntrait=sort(runif(S))\r\nd=as.matrix(dist(trait))\r\nd[d>max(d)/2]=max(d)-d[d>max(d)/2]\r\n\r\n## kernel width\r\nw=.1\r\n\r\n## noise-free competition kernel\r\nA0=exp(-(d/w)^4); A0=A0/mean(A0)*.2\r\n\r\n## Function to generate the competition kernel.\r\n## Inputs: species x-niche/trait values, type of noise, degree of noise, simulation number (run).\r\n## Outputs: A = noisy kernel in the case of process noise, noise-free kernel in the case of measurement noise\r\n## (cont) A0 = noise-free kernel in the case of process noise, noisy kernel in the case of measurement noise; \r\n## (cont) d = distances on x-niche/trait axis (on a circular axis);\r\n## (cont) prox = proxy trait values in the case of measurement noise; traity = y-niche/trait values;\r\n## (cont) spear = Spearman's rho, correlation between kernel and distances on x-axis;\r\n## (cont) pi = P(Atail > Amean) - P(Acore > Amean), quantifies core-tail structure in the kernel\r\nKernel=function(noiselevel,noisetype,traitvalues=trait,run=0){\r\n\tstopifnot(noisetype%in%noisetypes)\r\n\t\r\n\t## tie random number generator to run\r\n\tset.seed(run)\r\n\t\r\n\t## number of species\r\n\tS=length(traitvalues)\r\n\t\r\n\t## scaled noise level\r\n\ta0=noiselevel*with(scaling.factors,scaling[match(noisetype,noise)])\r\n\t\r\n\t## distances on x-niche/trait axis (axis is circular)\r\n\td=as.matrix(dist(traitvalues)); d[d>max(d)/2]=max(d)-d[d>max(d)/2]\r\n\t\r\n\t## A0 = noise-free kernel in the case of process noise, noisy kernel in the case of measurement noise\r\n\tA0=exp(-(d/w)^4); A0=A0/mean(A0)*.2\r\n\t\r\n\t## threshold distance separating core from tail\r\n\tdcoretail=d[which.min(abs(A0-mean(A0)))]\t\r\n\t\r\n\t## traity = y-niche/trait values; will contribute to kernel in proportion to noise level\r\n\ttraity=NULL\r\n\t\r\n\t## A = noisy kernel in the case of process noise, noise-free kernel in the case of measurement noise\r\n\tif(noisetype%in%c('complementary','essential')){ traity=runif(S); dy=as.matrix(dist(traity)); dy[dy>max(dy)/2]=max(dy)-dy[dy>max(dy)/2]}\r\n\tif(noisetype=='complementary'){ wyinverse=a0; A=exp(-sqrt((d/w)^2+(dy*wyinverse)^2)^4)}\t\t## from 1 to 10\r\n\tif(noisetype=='essential'){ wy=a0; A=pmax(exp(-(d/w)^4),exp(-(dy/wy)^4))}\t\t\t\t## from .01 to .15\r\n\t\r\n\tif(noisetype%in%c('core','tail','general')){ \r\n\t\tsigma = if(noisetype=='core') a0*exp(-(d/.1)^4) else if(noisetype=='tail') a0*(exp(d^4)-1) else if(noisetype=='general') a0\r\n\t\tA=matrix(pmax(0,rnorm(S*S,mean=exp(-(d/w)^4),sd=sigma)),S,S)\r\n\t\twhile(TRUE){ i=which(diag(A)==0); if(length(i)==0) break; diag(A)[i]=pmax(0,rnorm(length(i),mean=1,sd=a0))}\r\n\t}\r\n\t\r\n\t## prox = proxy trait values in the case of measurement noise\r\n\tprox=NULL\r\n\tif(noisetype=='measurement'){ \r\n\t\tprox=rnorm(S,mean=traitvalues,sd=a0); d=as.matrix(dist(prox)); d[d>max(d)/2]=max(d)-d[d>max(d)/2]\t\t\r\n\t\tA=A0; A0=exp(-(d/w)^4); A0=A0/mean(A0)*.2\r\n\t}\r\n\t\r\n\t## normalize A\r\n\tA=A/mean(A)*.2\r\n\t\r\n\t## spear = Spearman's rho; pi = P(Atail > Amean) - P(Acore > Amean)\r\n\tspear=cor(as.numeric(d),as.numeric(A),method='spearman')\t\t\t\t\t\t\t\t\t\t\t## -1 (perfect) to 0 (null)\r\n\tpi=sum(A[d>=dcoretail]>mean(A))/sum(d>=dcoretail)-(sum(A[d<=dcoretail]>mean(A))/sum(d<=dcoretail))\t## -1 (perfect) to 0 (null)\r\n\t\r\n\treturn(list(\r\n\t\tA=A,\r\n\t\tA0=A0,\r\n\t\td=d,\r\n\t\tprox=prox,\r\n\t\ttraity=traity,\r\n\t\tspear=spear,\r\n\t\tpi=pi\r\n\t))\r\n}\r\n\r\nRun=function(traits,compkernel,immig,maxtime=1e4,plot=1,logplot=1){\t\r\n\t## initial abundances\r\n\tx0=rep(1e-3,nrow(compkernel))+runif(nrow(compkernel),min=0,max=1e-3/100)\t\r\n\t\r\n\t## time series\r\n\ttimes=seq(0,maxtime,by=.1)\r\n\t\r\n\t## Lotka-Volterra dynamic equations\r\n\tLV=function(times,N,parms) list(pmax(N,0)*(1-as.numeric(parms$Kernel%*%pmax(N,0)))+parms$m)\r\n\t\r\n\t## Simulate model\r\n\tx=ode(y=x0,times=times,func=LV,parms=list(Kernel=compkernel,m=immig),method='lsoda'); n=x[nrow(x),-1]; n[n<1e-5]=0\r\n\t\r\n\t## Calculate richness at each time point\r\n\ts=apply(x[,-1],1,function(v) sum(v>1e-5))\r\n\t\r\n\t## Plot results\r\n\tif(plot) if(!logplot) plot(trait,n,t='h',las=1) else plot(trait[n>0],log(1+n[n>0]/min(n[n>0])),t='h',las=1)\r\n\t\r\n\treturn(list(data=data.frame(trait=traits,abundance=n),richness=data.frame(time=x[,1],S=s)))\t\r\n}\r\n", "meta": {"hexsha": "395ac4107a7bdf1a0d17cfc203581fdd58afb446", "size": 4628, "ext": "r", "lang": "R", "max_stars_repo_path": "NoiseModel.r", "max_stars_repo_name": "rafaeldandrea/Translucent-windows-2018-code", "max_stars_repo_head_hexsha": "1a6ccbdfc2a87396c1adf0f6e39e85acd1a1b07e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "NoiseModel.r", "max_issues_repo_name": "rafaeldandrea/Translucent-windows-2018-code", "max_issues_repo_head_hexsha": "1a6ccbdfc2a87396c1adf0f6e39e85acd1a1b07e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "NoiseModel.r", "max_forks_repo_name": "rafaeldandrea/Translucent-windows-2018-code", "max_forks_repo_head_hexsha": "1a6ccbdfc2a87396c1adf0f6e39e85acd1a1b07e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 39.8965517241, "max_line_length": 138, "alphanum_fraction": 0.6635695765, "num_tokens": 1506, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7981867777396211, "lm_q2_score": 0.672331699179286, "lm_q1q2_score": 0.5366462725401185}} {"text": "\n\n#' Function that optimizes sinusoid parameters to fit d18O data\n#' \n#' The second core function of the ShellChron growth model. Loops\n#' through all data windows and uses the \\code{growth_model} function\n#' to create d18O series that match the input data. This step is\n#' iterated and optimized (minimizing the Sum of Squared Residuals)\n#' through the SCEUA algorithm (by Duan et al., 1992) which finds\n#' the optimal input parameters to the growth rate and Sea Surface\n#' Temperature (SST) sinusoids to simulate d18O data.\n#' @param dat Matrix containing the input data\n#' @param dynwindow Information on the position and length of modeling\n#' windows\n#' @param transfer_function Transfer function used to convert d18Oc to temperature\n#' data.\n#' @param d18Ow Either a single value (constant d18Ow) or a vector of length\n#' equal to the period in SST data (365 days by default) containing information\n#' about seasonality in d18Ow. Defaults to constant d18Ow of 0 permille VSMOW\n#' (the modern mean ocean value)\n#' @param T_per Period of SST sinusoid (in days; default = 365)\n#' @param G_per Period of growth rate sinusoid (in days; default = 365)\n#' @param t_int Time interval (in days; default = 1)\n#' @param t_maxtemp Timing of the warmest day of the year (in julian day; \n#' default = 182.5, or May 26th halfway through the year)\n#' @param SCEUApar Parameters for SCEUA optimization (iniflg, ngs, maxn, kstop\n#' pcento, peps). For details, refer to Duan et al. (1992) in references\n#' @param sinfit Apply sinusoidal fitting to guess initial parameters for SCEUA\n#' optimization? \\code{TRUE/FALSE}\n#' @param MC Number of Monte Carlo simulations to apply for error propagation\n#' Default = 1000\n#' @param plot Should results of modeling be plotted? \\code{TRUE/FALSE}\n#' @return A list containing the \\code{resultarray} which contains the full\n#' result of all simulations on each data window and the \\code{parmat} listing\n#' all optimized growth rate and SST parameters used to model d18O in each data\n#' window\n#' @seealso Duan, Qingyun, Soroosh Sorooshian, and Vijai Gupta. \"Effective and\n#' efficient global optimization for conceptual rainfall runoff models.\" Water\n#' resources research 28.4 (1992): 1015-1031. https://doi.org/10.1029/91WR02985\n#' @references package dependencies: ggplot2 3.2.1; rtop 0.5.14\n#' Function dependencies: sinreg, d18O_model, growth_model\n#' \n#' \\doi{10.1029/91WR02985}\n#' @examples\n#' # Create dummy input data column by column\n#' dat <- as.data.frame(seq(1000, 40000, 1000))\n#' colnames(dat) <- \"D\"\n#' dat$d18Oc <- sin((2 * pi * (seq(1, 40, 1) - 8 + 7 / 4)) / 7)\n#' dat$YEARMARKER <- c(0, rep(c(0, 0, 0, 0, 0, 0, 1), 5), 0, 0, 0, 0)\n#' dat$D_err <- rep(100, 40)\n#' dat$d18Oc_err <- rep(0.1, 40)\n#' # Create dummy dynwindow data\n#' dynwindow <- as.data.frame(seq(1, 29, 2))\n#' colnames(dynwindow) <- \"x\"\n#' dynwindow$y <- rep(12, 15)\n#' # Run model function\n#' \\donttest{resultlist <- run_model(dat = dat,\n#' dynwindow = dynwindow,\n#' transfer_function = \"KimONeil97\",\n#' d18Ow = 0,\n#' T_per = 365,\n#' G_per = 365,\n#' t_int = 1,\n#' t_maxtemp = 182.5,\n#' SCEUApar = c(1, 25, 10000, 5, 0.01, 0.01),\n#' sinfit = TRUE,\n#' MC = 1000,\n#' plot = FALSE)}\n#' @export\nrun_model <- function(dat, # Core function to run the entire model on the data (dat)\n dynwindow, # The window vetor resulting from reading in the data \n transfer_function = \"KimONeil97\",\n d18Ow = 0,\n T_per = 365, # Temperature sinusoid parameters\n G_per = 365, # Growth sinusoid parameters\n t_int = 1, # Default time interval = 1 day\n t_maxtemp = 182.5, # Default time (day) at which maximum temperature is reached is 182.5 (exactly halfway through the year, or 1st of June)\n SCEUApar = c(1, 25, 10000, 5, 0.01, 0.01), # Set parameters for SCEUA optimization (iniflg, ngs, maxn, kstop, pcento, peps)\n sinfit = TRUE, # Apply sinusoidal fitting to guess initial parameters for SCEUA optimization? (TRUE/FALSE)\n MC = 1000, # If errors = TRUE, give the number of iterations for Monte Carlo simulation used in error propagation (default = 1000, if MC = 0 no eror propagation is done)\n plot = FALSE # Should the progress of model fitting be plotted?\n ){\n\n d18Oc <- d18Oc_err <- Omod <- NULL # Predefine variables to circumvent global variable binding error\n\n # Prepare data arrays for storage of modeling results\n resultarray <- array( # Create array to contain all modeling results of overlapping windows\n rep(as.matrix(cbind(dat, matrix(NA, ncol = length(dynwindow$x), nrow = length(dat$D)))), 9), # Replicate matrix of dat + extra columns to contain all variables\n dim = c(length(dat$D), length(dynwindow$x) + length(dat[1,]), 9) # Six variables, being: Modeled d18O, residuals, Day of the Year, Growth between datapoints, Instantaneous growth rate at datapoint and Modeled temperature\n )\n\n parmat <- matrix(NA, nrow = 7, ncol = length(dynwindow$x)) # Matrix in which to store the modeling parameters\n colnames(parmat) <- dynwindow$x\n rownames(parmat) <- c(\"T_amp\", \"T_pha\", \"T_av\", \"G_amp\", \"G_pha\", \"G_av\", \"G_skw\")\n \n # Prepare plot to show model progress\n\n if(plot == TRUE){\n dev.new()\n fitplot <- ggplot2::ggplot(dat, ggplot2::aes(D, d18Oc)) + # Create a plot showing the fit of each window on the original data, start with a plot of the original data\n ggplot2::geom_point() +\n ggplot2::geom_line() +\n ggplot2::geom_errorbar(ggplot2::aes(ymin = d18Oc - d18Oc_err, # Add error bars on model result (1 SD)\n ymax = d18Oc + d18Oc_err),\n width = dat$D_err) +\n ggplot2::geom_errorbarh(ggplot2::aes(xmin = D - D_err,\n xmax = D + D_err),\n height = 0.05) +\n ggplot2::ggtitle(\"Plot of d18Oc values vs. depth. Black = data, Red = model, Errorbars = 1 SD\")\n plot(fitplot)\n }\n\n # Estimate growth rate variability and round up to nearest higher magnitude of 10 for conservative boundary\n GRavest <- max(diff(dat[dat$YEARMARKER == 1,1])) / 365 # Estimate maximum growth rate from yearmarkers\n GRavmax <- 10 ^ (ceiling(log(GRavest, 10))) # Round up to nearest higher magnitude of 10\n\n # Find tailored range of temperatures from data\n d18Oc_range <- range(dat$d18Oc) # Find d18Oc range in data\n if(transfer_function == \"KimONeil97\"){ # Find temperature range (to be superseded with inverse d18O_model function in later updates)\n T_range <- 18.03 * 1000 / (log((d18Oc_range - (0.97002 * rev(range(d18Ow)) - 29.98)) / 1000 + 1) * 1000 + 32.42) - 273.15 # Use Kim and O'Neil (1997) with conversion between VSMOW and VPDB by Brand et al. (2014)\n }else if(transfer_function == \"GrossmanKu86\"){\n T_range <- 20.6 - 4.34 * (d18Oc_range - rev(range(d18Ow)) - 0.2) # Use Grossmann and Ku (1986) modified by Dettmann et al. (1999)\n }else{\n print(\"ERROR: Supplied transfer function is not recognized\")\n }\n T_max <- max(T_range)\n T_amp_max <- 2 * abs(diff(T_range))\n\n # Collate lower boundaries of parameters\n parl <- c(\n T_amp = 0, # Minimum T amplitude in degrees C\n T_pha = 0, # Minimum phase in days\n T_av = -4, # Minimum average T in degrees C\n G_amp = 0, # Minimum seasonal GR range in um/d\n G_pha = 0, # Minimum GR phase in days\n G_av = -1 * GRavmax, # Minimum average GR in um/d.\n G_skw = 0 # Minimum skew factor\n )\n\n # Collate upper boundaries of parameters\n paru <- c(\n T_amp = round(T_amp_max + 0.5, 0), # Maximum T amplitude in degrees C\n T_pha = 365, # Maximum phase in days\n T_av = round(T_max + 0.5, 0), # Maximum average T in degrees C\n G_amp = 2 * GRavmax, # Maximum seasonal GR range in um/d\n G_pha = 365, # Maximum GR phase in days\n G_av = GRavmax, # Maximum average GR in um/d (based on conservative boundaries of YEARMARKER indicators)\n G_skw = 100 # Maximum skew factor \n )\n \n # Set parameters for SCEUA optimization\n iniflg = SCEUApar[1] # Flag for initial parameter array (default = 1; included)\n ngs = SCEUApar[2] # Number of complexes (sub-populations, default = 25)\n maxn = SCEUApar[3] # Maximum number of function evaluations allowed during optimization (default = 10000)\n kstop = SCEUApar[4] # Maximum number of evolution loops before convergency (default = 5)\n pcento = SCEUApar[5] # Percentage change allowed in function value criterion before stop (default = 0.01)\n peps = SCEUApar[6] # Convergence level for parameter set (difference between parameters required for stop; default = 0.01)\n\n # Run the model on all windows\n\n for(i in 1:length(dynwindow$x)){ # Loop over shell record\n print(paste(\"Processing Datawindow:\", i, \"of\", length(dynwindow$x))) # Keep track of progress\n \n # Isolate year of d18O data based on window data\n Dsam <- dat[dynwindow$x[i]:(dynwindow$x[i] + dynwindow$y[i] - 1), 1]\n Osam <- dat[dynwindow$x[i]:(dynwindow$x[i] + dynwindow$y[i] - 1), 2]\n if(MC > 0){\n D_err <- dat[dynwindow$x[i]:(dynwindow$x[i] + dynwindow$y[i] - 1), 4] # Optional: include error on D\n O_err <- dat[dynwindow$x[i]:(dynwindow$x[i] + dynwindow$y[i] - 1), 5] # Optional: include error on d18Oc\n }else{\n D_err <- rep(0, dynwindow$y)\n O_err <- rep(0, dynwindow$y)\n MC <- 0\n }\n\n if(sinfit){ # If sinusoidal fitting is enabled\n sinlist <- sinreg(Dsam, Osam) # Run sinusoidal regression to find initial parameter values\n # Estimate starting parameters from regression results\n O_av_start <- sinlist[[1]][1] # Export starting value for annual d18O average\n O_amp_start <- sinlist[[1]][2] # Export starting value for d18O amplitude\n O_pha_start <- sinlist[[1]][4] %% sinlist[[1]][3] # Estimate position (in depth of the first peak in d18O)\n O_per_start <- sinlist[[1]][3] # Export starting value for period in distance domain\n }else{\n O_av_start <- mean(Osam) # Estimate starting value for annual d18O average by mean of d18O in record\n O_amp_start <- diff(range(Osam)) / 2 # Estimate starting value for d18O amplitude by half the difference between minimum and maximum d18Oc\n O_per_start <- diff(range(Dsam)) # Estimate starting period as thickness of isolated year\n O_pha_start <- 0.25 * O_per_start # Set starting phase to one quarter of a cycle if it cannot be estimated from sinusoidal regression \n }\n\n if(transfer_function == \"KimONeil97\"){\n T_av_start <- 18.03 * 1000 / (1000 * log((O_av_start - (0.97002 * mean(d18Ow) - 29.98)) / 1000 + 1) + 32.42) - 273.15 # Estimate mean temperature. Use Kim and O'Neil (1997) with conversion between VSMOW and VPDB by Brand et al. (2014)\n T_amp_start <- 18.03 * 1000 / (1000 * log((O_av_start - O_amp_start - (0.97002 * mean(d18Ow) - 29.98)) / 1000 + 1) + 32.42) - 273.15 - T_av_start # Estimate temperature amplitude. Use Kim and O'Neil (1997) with conversion between VSMOW and VPDB by Brand et al. (2014)\n }else if(transfer_function == \"GrossmanKu86\"){\n T_av_start <- 20.6 - 4.34 * (O_av_start - mean(d18Ow) - 0.2) # Estimate mean temperature. Use Grossmann and Ku (1986) modified by Dettmann et al. (1999)\n T_amp_start <- 20.6 - 4.34 * (O_av_start - O_amp_start - mean(d18Ow) - 0.2) - T_av_start # Estimate mean temperature. Use Grossmann and Ku (1986) modified by Dettmann et al. (1999)\n }else{\n print(\"ERROR: Supplied transfer function is not recognized\")\n }\n\n O_peak <- O_pha_start + Dsam[1] # Find position of d18O peak in distance domain\n T_pha_start <- ((O_pha_start - 0.5 * O_per_start) %% O_per_start) / O_per_start * T_per # Estimate position of first peak in temperature (low in d18O) relative to annual cycle (days)\n G_av_start <- O_per_start / G_per # Estimate average growth rate in distance/day\n\n years <- 3 # Set default number of years to 3\n\n # Collate starting parameters\n par0 <- c(\n T_amp = T_amp_start,\n T_pha = T_pha_start,\n T_av = T_av_start,\n\n G_amp = G_av_start / 2, # Start by estimating growth rate changes by half the average\n G_pha = T_pha_start, # Start by estimating that the peak in growth rate coincides with the peak in temperature\n G_av = G_av_start,\n G_skw = 50 # Start with a no skew\n )\n\n invisible(capture.output( # Suppress the details on converging SCEUA\n sceua_list <- rtop::sceua(growth_model,\n par0,\n T_per = T_per,\n G_per = G_per,\n years = years,\n t_int = t_int,\n transfer_function = transfer_function,\n d18Ow = d18Ow,\n Dsam = Dsam,\n Osam = Osam,\n t_maxtemp = t_maxtemp,\n parl,\n paru,\n maxn,\n kstop,\n pcento,\n ngs,\n iniflg = iniflg,\n peps = peps,\n implicit = function(pars){sum(pars[4]/2, pars[6]) < 1} # Make sure that the cumulative GR curve is not located below 0 (if G_av < - G_amp / 2)\n )\n ))\n\n par1 <- sceua_list[[1]] # Eport parameters of final model\n names(par1) <- names(par0)\n\n result <- growth_model(par1, T_per, G_per, years, t_int, transfer_function, d18Ow, Dsam, Osam, t_maxtemp, plot = FALSE, MC, D_err, O_err, return = \"result\") # Calculate the end result of the best fit\n \n if(plot == TRUE){\n fitplot <- fitplot + # Add the new model fit to the plot to track progress of the model\n ggplot2::geom_point(data = as.data.frame(result), ggplot2::aes(Dsam, Omod), colour = \"red\") +\n ggplot2::geom_line(data = as.data.frame(result), ggplot2::aes(Dsam, Omod), colour = \"red\") \n print(fitplot)\n }\n \n resultarray[, i + length(dat[1, ]), ] <- rbind(matrix(NA, nrow = i - 1, ncol = 9), result[, 3:11], matrix(NA, nrow = length(dat$D) - i - length(result[,1]) + 1, ncol = 9)) # Add results to result array\n parmat[, i] <- par1 # Add parameters to parameter array\n }\n\n # Provide names for all dimensions in the result array\n dimnames(resultarray) <- list(\n paste(\"sample\", 1:length(resultarray[, 1, 3])),\n c(colnames(dat), paste(\"window\", 1:length(dynwindow$x))),\n c(\"Modeled_d18O\", \"d18O_residuals\", \"Time_of_year\", \"Instantaneous_growth_rate\", \"Modeled temperature\", \"Modeled_d18O_SD\", \"Time_of_Year_SD\", \"Instantaneous_growth_rate_SD\", \"Modeled_temperature_SD\")\n )\n\n colnames(parmat) <- paste(\"window\", 1:length(parmat[1,]))\n return(list(resultarray, parmat))\n}", "meta": {"hexsha": "1cdf4130a9a3df24b4f52ac496bb0890f520daa8", "size": 14912, "ext": "r", "lang": "R", "max_stars_repo_path": "R/Run_model.r", "max_stars_repo_name": "nhoeche/ShellChron.jl", "max_stars_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/Run_model.r", "max_issues_repo_name": "nhoeche/ShellChron.jl", "max_issues_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/Run_model.r", "max_forks_repo_name": "nhoeche/ShellChron.jl", "max_forks_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 55.8501872659, "max_line_length": 279, "alphanum_fraction": 0.6461909871, "num_tokens": 4325, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8519528094861981, "lm_q2_score": 0.6297745935070808, "lm_q1q2_score": 0.5365382342813858}} {"text": "#' LOO-CV, WAIC and Raw LPPD Calculations\n#' \n#' Implements LOO-CV, WAIC and Raw LPPD for the \\code{\\link{model_judgement}} function. Contains\n#' the helper function \"colVars\" which calculates row-wise variances efficiently.\n#' \n#' @param stanfit A Stanfit object fitted on synthetic data.\n#' @param current_model Name of the current model (Currently not in use.)\n#' @param lik_name Name under which the log likelihoods have been saved in the models. Needs to be identical \n#' across all Stanfit objects.\n#' @param impute_inf A boolean which regulates if underflow values should be automatically imputed or not. If\n#' \\code{FALSE}, the models with such values will just be ignored. If \\code{TRUE}, a report will be generated\n#' on how many values were imputed and for which models.\n#' @return A list with LOO-CV, WAIC and Raw LPPD calculations.\nwaic <- function(stanfit, current_model, lik_name, impute_inf){\n #http://kylehardman.com/BlogPosts/View/6 DIC code also from Gelman\n #Modified code of www.stat.columbia.edu/~gelman/research/unpublished/waic_stan.pdf\n #from gist.github.com/ihrke for underflow probs\n \n colVars <- function(a) {\n n <- dim(a)[[1]]; \n c <- dim(a)[[2]]; \n result <- (.colMeans(((a - matrix(.colMeans(a, n, c), \n nrow = n, ncol = c, byrow = TRUE)) ^ 2), n, c) * n / (n - 1))\n return(result)\n }\n \n log_lik <- rstan::extract(stanfit, lik_name)$log_lik\n \n \n if(impute_inf){\n lik_imp <- sum(is.infinite(log_lik))/length(log_lik)\n log_lik[is.infinite(log_lik)] <- mean(log_lik[!is.infinite(log_lik)])\n }\n dim(log_lik) <- if (length(dim(log_lik))==1) c(length(log_lik),1) else\n c(dim(log_lik)[1], prod(dim(log_lik)[2:length(dim(log_lik))]))\n S <- nrow(log_lik)\n n <- ncol(log_lik)\n #log pointwise \n lpd <- log(colMeans(exp(log_lik))) #only when posterior simulations in Stan are correctly made (and possible)\n \n #waic\n p_waic <- colVars(log_lik)\n elpd_waic <- lpd - p_waic\n waic <- -2*elpd_waic\n \n #loo\n loo_weights_raw <- 1/exp(log_lik-max(log_lik))\n \n if(impute_inf){\n loo_imp <- sum(is.infinite(loo_weights_raw))/length(loo_weights_raw)\n loo_weights_raw[is.infinite(loo_weights_raw)] <- mean(loo_weights_raw[!is.infinite(loo_weights_raw)])\n }\n \n loo_weights_normalized <- loo_weights_raw/matrix(colMeans(loo_weights_raw),nrow=S,ncol=n,byrow=TRUE)\n loo_weights_regularized <- pmin (loo_weights_normalized, sqrt(S))\n \n elpd_loo <- log(colMeans(exp(log_lik)*loo_weights_regularized)/colMeans(loo_weights_regularized))\n p_loo <- lpd - elpd_loo\n elpd_loo <- elpd_loo*-2\n pointwise <- cbind(waic,lpd,p_waic,elpd_waic,p_loo,elpd_loo)\n total <- colSums(pointwise)\n se <- sqrt(n*colVars(pointwise))\n ic <- list(waic=total[\"waic\"], elpd_waic=total[\"elpd_waic\"],\n p_waic=total[\"p_waic\"], elpd_loo=total[\"elpd_loo\"], p_loo=total[\"p_loo\"],\n pointwise=pointwise, total=total, se=se)\n \n if(impute_inf){\n imps <- list('lik_imp' = lik_imp, 'loo_imp' = loo_imp)\n return(list('ic' = ic, 'imps' = imps))\n } else{\n return(list('ic' = ic))\n }\n}", "meta": {"hexsha": "6331caa7747834e0103b0c4abd7b9b5a91a7f20b", "size": 3223, "ext": "r", "lang": "R", "max_stars_repo_path": "R/waic.r", "max_stars_repo_name": "josephmbarnby/StanDDM_JMBarnbyEDIT", "max_stars_repo_head_hexsha": "3ebbb6250c41c232d8cf44a1ae16dcd190ba1b0a", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 28, "max_stars_repo_stars_event_min_datetime": "2019-06-04T13:56:15.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-11T15:53:37.000Z", "max_issues_repo_path": "R/waic.r", "max_issues_repo_name": "josephmbarnby/StanDDM_JMBarnbyEDIT", "max_issues_repo_head_hexsha": "3ebbb6250c41c232d8cf44a1ae16dcd190ba1b0a", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2019-07-02T06:45:23.000Z", "max_issues_repo_issues_event_max_datetime": "2021-06-28T20:36:29.000Z", "max_forks_repo_path": "R/waic.r", "max_forks_repo_name": "josephmbarnby/StanDDM_JMBarnbyEDIT", "max_forks_repo_head_hexsha": "3ebbb6250c41c232d8cf44a1ae16dcd190ba1b0a", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2021-06-23T01:52:34.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-10T12:36:06.000Z", "avg_line_length": 44.1506849315, "max_line_length": 113, "alphanum_fraction": 0.6568414521, "num_tokens": 956, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8519527906914787, "lm_q2_score": 0.6297746004557471, "lm_q1q2_score": 0.5365382283648847}} {"text": "#' Function that propagates measurement uncertainty through model results\n#' \n#' Function to propagate combined errors on \\code{x} (= \\code{Dsam}) and\n#' \\code{y} (= \\code{Osam}) on the modeled X (= \\code{D}) and Y \n#' (= \\code{d18Oc}) values by means of direct projection of y–uncertainty\n#' on \\code{x} and then combine the errors on both in the \\code{x} domain\n#' \n#' Note: projection y_err on x_err leads to large X errors on shallow\n#' slopes due to numerical calculation of fist derivative.\n#' @param x Vector of \\code{x} values of input data\n#' @param x_err Vector of uncertainties on \\code{x} values\n#' @param y Vector of \\code{y} values of input data\n#' @param y_err Vector of uncertainties on \\code{y} values\n#' @param X Vector of modeled \\code{X} values on which the uncertainty is\n#' to be projected\n#' @param Y Matrix of modeled x and \\code{Y} values\n#' @param MC Number of Monte Carlo simulations to apply for error propagation\n#' Default = 1000\n#' @return A vector listing the standard deviations of propagated errors \n#' propagated on all \\code{X} values.\n#' @examples\n#' # Create dummy data for input\n#' x <- seq(1, 40, 1)\n#' x_err <- rep(0.1, 40)\n#' y <- sin((2 * pi * (seq(1, 40, 1) - 8 + 30 / 4)) / 30)\n#' y_err <- rep(0.1, 40)\n#' X <- seq(1.5, 39.5, 1)\n#' Y <- cbind(seq(1, 39, 1), 0.9 * sin((2 * pi * (seq(1, 39, 1) - 9 +\n#' 25 / 4)) / 25))\n#' # Run function\n#' result <- mc_err_proj(x, x_err, y, y_err, X, Y, 1000)\n#' @export\nmc_err_proj <- function(x,\n x_err,\n y,\n y_err,\n X,\n Y,\n MC = 1000){ # Function to propagate combined errors on x and y on the modeled X and Y values by means of local projection of y uncertainty on x and subsequent combination of uncertainties in X domain\n \n dYdX <- diff(Y[, 2]) / diff(X) # Create first derivative of Y by X\n dYdX[which(abs(dYdX) <= 10 ^ (floor(log(mean(abs(dYdX)), 10)) - 1))] <- sign(dYdX[which(abs(dYdX) <= 10 ^ (floor(log(mean(abs(dYdX)), 10)) - 1))]) * 10 ^ (floor(log(mean(abs(dYdX)), 10)) - 1) # Remove small absolute values (more than one order of magnitude smaller than the mean), preserving their sign\n dYdX <- append(dYdX, dYdX[length(dYdX)]) # repeat last value to increase the length of the vector to match X/Y (366 days)\n localslope <- dYdX[apply(abs(outer(x, X, FUN = \"-\")), 1, which.min)] # Find the local slope belonging to each sample position\n x_err_proj <- y_err / localslope # Project uncertainty on Y on x domain using slope\n x_err_comb <- sqrt(x_err_proj ^2 + x_err ^2) # Combine uncertainties on x and y in the X domain\n return(x_err_comb)\n}", "meta": {"hexsha": "23d27a8b296b79f610e38c470b8bf6fe5e87a9d2", "size": 2579, "ext": "r", "lang": "R", "max_stars_repo_path": "R/mc_err_proj.r", "max_stars_repo_name": "nhoeche/ShellChron.jl", "max_stars_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/mc_err_proj.r", "max_issues_repo_name": "nhoeche/ShellChron.jl", "max_issues_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/mc_err_proj.r", "max_forks_repo_name": "nhoeche/ShellChron.jl", "max_forks_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 53.7291666667, "max_line_length": 306, "alphanum_fraction": 0.6645986817, "num_tokens": 808, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8670357735451834, "lm_q2_score": 0.6187804337438502, "lm_q1q2_score": 0.5365047720257232}} {"text": "##################################################################\r\n#\r\n#\r\n# >>> Geodesy In R <<<\r\n#\r\n# This file contains a series of geodetic functions to aid in performing\r\n# spatial analysis across large distances (100s of km). Note: R Trig.\r\n# functions take arguments in radians. 1 degree = pi/180 radians\r\n#\r\n# original source by Darren Gillis and others\r\n#\r\n#\tVarious Chart Datum Parameters:\r\n#\r\n#\tName Major Axis, a (km) Flattening (f)\r\n#\tWGS84 6378.13700\t 1/298.257223563\r\n#\tGRS80/NAD83 6378.13700\t 1/298.257222101\r\n#\tWGS66 6378.145 1/298.25\r\n#\tGRS67/IAU68 6378.16000 1/298.2472\r\n#\tWGS72 6378.135 1/298.26\r\n#\tKrasovsky 6378.245 1/298.3\r\n#\tClarke66/NAD27 6378.2064 1/294.9786982138\r\n#\r\n#\tCoordinate Systems and Map Projections, D. H. Maling (Pergamon 1992)\r\n#\tSummarized by Ed Williams (http://williams.best.vwh.net/index.html)\r\n#\r\n# Earth Radius as geometric mean of major & minor radia (WGS84)\r\n# Nautical Miles per degree as 57.29\r\n#\r\n#\r\n#\tNote: Radius of minor axis = a - a*f\r\n#\r\n# test data\r\n# point = data.frame(lat=-30, lon=150)\r\n# locations = data.frame(lat=c(-31, -35, -40), lon=c(150, 151, 151))\r\n# vincenty distances = 110.8609, 562.3759, 1113.1415\r\n# great circle = 111.1327 563.5021 1115.0260\r\n##################################################################\r\n\r\n\r\n\r\n\r\n##################################################################\r\n# vincenty method .. more accurate\r\n# the original fortran code and documentation is copied at the bottom \r\n##################################################################\r\n\r\n\r\n\r\n\r\nvincenty = function (loc1, loc2, a, f) {\r\n\r\n ## for history and equations:: see https://github.com/OSGeo/proj.4/wiki/GeodesicCalculations\r\n warning( \"Deprecated. Consider using geosphere::distVincetyEllipsoid \" )\r\n\r\n names(loc1) = c(\"lon\", \"lat\")\r\n names(loc2) = c(\"lon\", \"lat\")\r\n \r\n eps = 0.5e-13\r\n pi = 3.1415926535897932384626433832795\r\n rad = 0.0174532925199432957692369076848861\r\n r = 1.0-f\r\n\r\n# convert decimal degrees to radians\r\n\r\n rloc1 = rad*loc1\r\n rloc2 = rad*loc2\r\n \r\n tu1 = r * sin(rloc1$lat) / cos(rloc1$lat)\r\n tu2 = r * sin(rloc2$lat) / cos(rloc2$lat)\r\n \r\n cu1 = 1. / sqrt(tu1 * tu1 + 1. )\r\n su1 = cu1 * tu1\r\n cu2 = 1. / sqrt(tu2 * tu2 + 1. )\r\n \r\n s = cu1 * cu2\r\n baz = s * tu2\r\n faz = baz * tu1\r\n dlon = (rloc2$lon - rloc1$lon) \r\n x=dlon\r\n \r\n q = c(1:dim(loc2)[1]) # get it started into the loop\r\n\r\n sx = NULL\r\n cx = NULL\r\n tu1 = NULL\r\n tu2 = NULL\r\n sy = NULL\r\n cy = NULL\r\n y = NULL\r\n sa = NULL\r\n c2a = NULL\r\n cz = NULL\r\n e = NULL\r\n c = NULL\r\n\r\n \r\n \t\trepeat {\r\n \t\t\tsx[q] = sin(x[q])\r\n cx[q] = cos(x[q])\r\n tu1[q] = cu2[q] * sx[q]\r\n tu2[q] = baz[q] - su1 * cu2[q] * cx[q]\r\n sy[q] = sqrt(tu1[q] * tu1[q] + tu2[q] * tu2[q])\r\n cy[q] = s[q] * cx[q] + faz[q]\r\n y[q] = atan2(sy[q], cy[q])\r\n sa[q] = s[q] * sx[q] / sy[q]\r\n c2a[q] = - sa[q] * sa[q] + 1.\r\n nz = intersect(which(c2a != 0), q)\r\n cz[q] = faz[q] + faz[q]\r\n cz[nz] = -cz[nz] / c2a[nz] + cy[nz]\r\n e[q] = cz[q] * cz[q] * 2 - 1\r\n c[q] = ((-3. * c2a[q] + 4.) * f + 4.) * c2a[q] * f / 16.\r\n d = x\r\n x[q] = ((e[q] * cy[q] * c[q] + cz[q]) * sy[q] * c[q] + y[q]) * sa[q]\r\n x[q] = (1. - c[q]) * x[q] * f + dlon[q]\r\n q = which(abs(d - x) > eps)\r\n if (length(q) < 1) break\r\n }\r\n\r\n #faz = atan2(tu1, tu2)\r\n #baz = atan2(cu1 * sx, baz * cx - su1 * cu2) + pi\r\n x = sqrt( (1. / r / r - 1.) * c2a + 1. ) + 1.\r\n x = (x - 2.) / x\r\n c = 1. - x\r\n c = (x * x / 4. + 1.) / c\r\n d = (0.375 * x * x - 1.) * x\r\n x = e * cy\r\n s = 1. - e - e\r\n s = ((((sy * sy * 4. - 3.) * s * cz * d / 6. - x ) * d / 4. + cz) * sy * d + y) * c * a * r\r\n \r\n s [!is.finite(s)] = 0\r\n \r\n #faz = faz / rad # these are azimuths .. not needed\r\n #baz = baz / rad\r\n\r\n return(s)\r\n\r\n }\r\n\r\n\r\n##################################################################\r\n#~ C *** SOLUTION OF THE GEODETIC INVERSE PROBLEM AFTER T.VINCENTY\r\n#~ C *** MODIFIED RAINSFORD'S METHOD WITH HELMERT'S ELLIPTICAL TERMS\r\n#~ C *** EFFECTIVE IN ANY AZIMUTH AND AT ANY DISTANCE SHORT OF ANTIPODAL\r\n#~ C *** STANDPOINT/FOREPOINT MUST NOT BE THE GEOGRAPHIC POLE\r\n#~ C\r\n#~ C *** A IS THE SEMI-MAJOR AXIS OF THE REFERENCE ELLIPSOID\r\n#~ C *** F IS THE FLATTENING (NOT RECIPROCAL) OF THE REFERNECE ELLIPSOID\r\n#~ C *** LATITUDES AND LONGITUDES IN RADIANS POSITIVE NORTH AND EAST\r\n#~ C *** FORWARD AZIMUTHS AT BOTH POINTS RETURNED IN RADIANS FROM NORTH\r\n#~ C\r\n#~ C *** PROGRAMMED FOR CDC-6600 BY LCDR L.PFEIFER NGS ROCKVILLE MD 18FEB75\r\n#~ C *** MODIFIED FOR IBM SYSTEM 360 BY JOHN G GERGEN NGS ROCKVILLE MD 7507\r\n#~ C\r\n#~ C *** Modified for R by D.Gillis Zoology University of Manitoba 16JUN03\r\n#~ C Replaced common blocks for constants with DATA statements. Datum\r\n#~ C parameters moved from common block to subroutine arguements.\r\n#~ C\r\n#~ C *** Input Variables: DLAT1,DLON1 - initial fix (P1) in degrees\r\n#~ C (latitude, north +) (longitude east +)\r\n#~ C DLAT2,DLON2 - destination fix (P2) in degrees\r\n#~ C\r\n#~ C Ellipsoid (spheroid model, eg. WGS84)\r\n#~ C A - radius of major axis in distance units\r\n#~ C F - flattening factor\r\n#~ C\r\n#~ C *** Output Variables: FAZ - azimuth of the geodesic (P1 to P2)\r\n#~ C BAZ - azimuth of the geodesic (P2 to P1)\r\n#~ C S - spheroidal distance = length of the geodesic\r\n#~ C in distance units\r\n#~ C\r\n#~ C *** After Vincenty,T. 1975. Direct and inverse solutions of geodesics\r\n#~ C on the ellipsoid with application of nested equations. Survey\r\n#~ C Review 23(176):88-94.\r\n#~ C\r\n#~ C\r\n#~ C These routines were compiled in Windows 98 (DOS Window) using the\r\n#~ C Gnu FORTRAN complier: g77 --share -o geodesy.dll marspat.for\r\n#~ C creating: geodesy.dll\r\n#~ C\r\n#~ C\r\n#~ C\r\n#\r\n# original code (\"geodesy.for\") is as follows:\r\n# note the return of azimuths has been turned off\r\n#\r\n #~ SUBROUTINE INVER1(DLAT1,DLON1,DLAT2,DLON2,A,F,FAZ,BAZ,S)\r\n #~ IMPLICIT DOUBLE PRECISION (A-H,O-Z)\r\n#~ C COMMON/CONST/PI,RAD - original code\r\n#~ C COMMON/ELIPSOID/A,F - original code\r\n #~ DATA EPS/0.5D-13/\r\n #~ DATA PI/3.1415926535897932384626433832795D0/\r\n #~ DATA RAD/0.0174532925199432957692369076848861D0/\r\n #~ R=1.D0-F\r\n#~ C\r\n#~ C Convert decimal degrees to radians\r\n#~ C\r\n #~ GLAT1=DLAT1*RAD\r\n #~ GLON1=DLON1*RAD\r\n #~ GLAT2=DLAT2*RAD\r\n #~ GLON2=DLON2*RAD\r\n#~ C\r\n #~ TU1=R*DSIN(GLAT1)/DCOS(GLAT1)\r\n #~ TU2=R*DSIN(GLAT2)/DCOS(GLAT2)\r\n #~ CU1=1./DSQRT(TU1*TU1+1.)\r\n #~ SU1=CU1*TU1\r\n #~ CU2=1./DSQRT(TU2*TU2+1.)\r\n #~ S=CU1*CU2\r\n #~ BAZ=S*TU2\r\n #~ FAZ=BAZ*TU1\r\n #~ X=GLON2-GLON1\r\n #~ 100 SX=DSIN(X)\r\n #~ CX=DCOS(X)\r\n #~ TU1=CU2*SX\r\n #~ TU2=BAZ-SU1*CU2*CX\r\n #~ SY=DSQRT(TU1*TU1+TU2*TU2)\r\n #~ CY=S*CX+FAZ\r\n #~ Y=DATAN2(SY,CY)\r\n #~ SA=S*SX/SY\r\n #~ C2A=-SA*SA+1.\r\n #~ CZ=FAZ+FAZ\r\n #~ IF(C2A.GT.0.)CZ=-CZ/C2A+CY\r\n #~ E=CZ*CZ*2.-1.\r\n #~ C=((-3.*C2A+4.)*F+4.)*C2A*F/16.\r\n #~ D=X\r\n #~ X=((E*CY*C+CZ)*SY*C+Y)*SA\r\n #~ X=(1.-C)*X*F+GLON2-GLON1\r\n #~ IF(DABS(D-X).GT.EPS) GOTO 100\r\n #~ FAZ=DATAN2(TU1,TU2)\r\n #~ BAZ=DATAN2(CU1*SX,BAZ*CX-SU1*CU2)+PI\r\n #~ X=DSQRT((1./R/R-1.)*C2A+1.)+1.\r\n #~ X=(X-2.)/X\r\n #~ C=1.-X\r\n #~ C=(X*X/4.+1.)/C\r\n #~ D=(0.375*X*X-1.)*X\r\n #~ X=E*CY\r\n #~ S=1.-E-E\r\n #~ S=((((SY*SY*4.-3.)*S*CZ*D/6.-X)*D/4.+CZ)*SY*D+Y)*C*A*R\r\n #~ FAZ=FAZ/RAD\r\n #~ BAZ=BAZ/RAD\r\n #~ END\r\n#\r\n# example method of access to shared library from R\r\n#\r\n#~ dyn.load(\"~/src/grids/geodesy.lib\")\r\n\t#~ inver1<- function (Lat1, Lon1, Lat2, Lon2) {\r\n\t #~ a<-6378137.00 # WGS84 major axis\r\n\t #~ f<-1/298.257223563 # WGS84 flattening parameter\r\n #~ .Fortran(\"inver1\",as.double(Lat1),as.double(Lon1),\r\n #~ as.double(Lat2),as.double(Lon2),\r\n #~ as.double(a),as.double(f),\r\n #~ as.double(0),as.double(0),\r\n #~ as.double(0)) [[9]]\r\n #~ }\r\n\r\n\r\n #~ GDvincenty <- function (Lat1, Lon1, Lat2, Lon2) {\r\n \t\t#~ n = length(Lat1)\r\n\t #~ ans = inver1(Lat1,Lon1,Lat2,Lon2)\r\n\r\n #~ if (n>1) {\r\n #~ for (i in 2:n) {\r\n \t#~ ans<-c(ans, inver1(Lat1[i],Lon1[i],Lat2[i],Lon2[i]))\r\n #~ }\r\n #~ }\r\n #~ ans\r\n\t#~ }\r\n\r\n#\r\n##################################################################\r\n\r\n", "meta": {"hexsha": "300b87f36bf9e667212f328952a2111437d82dd8", "size": 9113, "ext": "r", "lang": "R", "max_stars_repo_path": "R/vincenty.r", "max_stars_repo_name": "PEDsnowcrab/aegis", "max_stars_repo_head_hexsha": "92d4b045c17773c184df4ff3ed47c226ba4eddbf", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/vincenty.r", "max_issues_repo_name": "PEDsnowcrab/aegis", "max_issues_repo_head_hexsha": "92d4b045c17773c184df4ff3ed47c226ba4eddbf", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/vincenty.r", "max_forks_repo_name": "PEDsnowcrab/aegis", "max_forks_repo_head_hexsha": "92d4b045c17773c184df4ff3ed47c226ba4eddbf", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-04-21T12:58:58.000Z", "max_forks_repo_forks_event_max_datetime": "2020-04-21T12:58:58.000Z", "avg_line_length": 33.6273062731, "max_line_length": 98, "alphanum_fraction": 0.4773400636, "num_tokens": 3242, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8670357735451834, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5365047720257231}} {"text": "\necent_coords <- function(pts){ # pts je vektor sa 4 elementa x, y, e_cent, seed.\n e_cent <- pts[3]/1000\n sigma <- matrix(c(e_cent^2, 0, 0, e_cent^2),2)\n set.seed(pts[4])\n pt_coords <- mvrnorm(1, pts[1:2], sigma)\n return(pt_coords)\n}\n\nni3 <- function(x1, y1, x2, y2){\n ## body\n dEasting <- as.numeric(x2 - x1)\n dNorthing <- as.numeric(y2 - y1)\n atg <- ifelse(dNorthing < 0, atan(dEasting/dNorthing)*180/pi + 180, atan(dEasting/dNorthing)*180/pi)\n ang <- ifelse(atg < 0, atg + 360, atg)\n return(ang)\n}\n\n\nsim_st <- function(survey.net.st, red){\n prj <- st_crs(survey.net.st)\n ro <- 180/pi*3600\n st_data <- survey.net.st %>% st_drop_geometry() %>% .[1, c(\"x_from\", \"y_from\", \"e_cent_from\", \"seed_from\")] %>% as.numeric()\n st_coords_e <- ecent_coords(pts = st_data)\n e_cent_from <- dist(pt1_coords = st_data[1:2], pt2_coords = st_coords_e)*1000\n\n\n target_coords_e <- survey.net.st %>% st_drop_geometry() %>% .[, c(\"x_to\", \"y_to\", \"e_cent_to\", \"seed_to\")] %>% # TODO: seed ne radi.\n apply(., 1, function(x) ecent_coords(pts = as.numeric(x))) %>% t()\n\n e_cent_to <- data.frame((survey.net.st %>% st_drop_geometry() %>% .[, c(\"x_to\", \"y_to\")] - target_coords_e)*1000) %>%\n dplyr::rename(e_x = x_to, e_y = y_to) %>%\n dplyr::mutate(e_cent_to = sqrt(e_x^2+e_y^2)) %>%\n .$e_cent_to\n\n survey.net.st$e_cent_from <- e_cent_from\n survey.net.st$e_cent_to <- e_cent_to\n\n survey.net.st <- survey.net.st %>%\n dplyr::mutate(x_from = st_coords_e[1],\n y_from = st_coords_e[2],\n x_to = target_coords_e[, 1],\n y_to = target_coords_e[, 2],\n d = sqrt((x_from-x_to)^2+(y_from-y_to)^2),\n sd_e_cent_from = sqrt((e_cent_from^2*ro^2)/(2*(d*1000)^2)),\n sd_e_cent_to = sqrt((e_cent_to^2*ro^2)/(2*(d*1000)^2)),\n sd_e_cent_dist_from = sqrt(e_cent_from^2/2),\n sd_e_cent_dist_to = sqrt(e_cent_to^2/2),\n ni1 = ni3(x_from, y_from, x_to, y_to),\n z = 360-Hz0,\n Hz_sim = ifelse(ni1 >= Hz0, z+ni1-360, z+ni1),\n sd_hz_sq = (e_air^2 + e_focus^2 + sd_Hz^2) + sd_e_cent_from^2 + sd_e_cent_to^2,\n sd_dist_sq = sd_e_cent_dist_from^2 + sd_e_cent_dist_to^2 + sd_dist^2)\n\n if(red == TRUE){\n survey.net.st <- survey.net.st %>%\n # dplyr::mutate_at(., .vars = \"Hz_sim\", ~Hz_sim - Hz_sim[1]) %>%\n dplyr::mutate_at(., .vars = c(\"sd_hz_sq\", \"sd_dist_sq\"), ~replace(., is.na(.), 0)) %>%\n dplyr::mutate(Hz_sim = ifelse(Hz_sim < 0, Hz_sim + 360, Hz_sim),\n Hz_sim = Hz_sim - Hz_sim[1],\n Hz_sim = ifelse(Hz_sim < 0, Hz_sim + 360, Hz_sim), # dplyr::case_when(Hz_sim < 0 ~ Hz_sim + 360, Hz_sim >= 0 ~ Hz_sim),\n sd_hz_sq = ifelse(is.na(sd_hz_sq), NA, sd_hz_sq + sd_hz_sq[1]),\n sd_hz_sample = rnorm(dim(survey.net.st)[1], 0, sqrt(sd_hz_sq)),\n Hz_sim = Hz_sim + sd_hz_sample/3600,\n Hz = ifelse(round(Hz_sim, 0) == 0, round(Hz_sim, 0), Hz_sim),\n Hz = pmax(Hz, 0),\n Hz = ifelse(is.na(sd_Hz), NA, Hz), #sa ifelse se moze i ostalo resiti da ne stavlja 0 tamo za sd_hz_sq,\n HzD = floor(Hz), HzM = floor((Hz-HzD)*60), HzS = round(((Hz-HzD)*60-HzM)*60, 1),\n sd_dist_sample = rnorm(dim(survey.net.st)[1], 0, sqrt(sd_dist_sq)),\n dist_sim = d + sd_dist_sample/1000,\n HD = round(dist_sim, 4),\n HD = ifelse(is.na(sd_dist), NA, HD)) %>%\n as.data.frame() %>%\n dplyr::arrange(ID, Hz) %>%\n st_sf(., crs = prj)\n }else{\n survey.net.st <- survey.net.st %>%\n dplyr::mutate(Hz_sim = ifelse(Hz_sim < 0, Hz_sim + 360, Hz_sim),\n sd_hz_sample = rnorm(dim(survey.net.st)[1], 0, sqrt(sd_hz_sq)),\n Hz_sim = Hz_sim + sd_hz_sample/3600,\n Hz = pmax(Hz, 0),\n HzD = floor(Hz), HzM = floor((Hz-HzD)*60), HzS = round(((Hz-HzD)*60-HzM)*60, 1),\n sd_dist_sample = rnorm(dim(survey.net.st)[1], 0, sqrt(sd_dist_sq)),\n dist_sim = d + sd_dist_sample/1000,\n HD = round(dist_sim, 4),\n HD = ifelse(is.na(sd_dist), NA, HD)) %>%\n as.data.frame() %>%\n dplyr::arrange(ID, Hz) %>%\n st_sf(., crs = prj)\n }\n return(survey.net.st)\n}\n\n#survey.net = mreza_s; red = TRUE; Hz0 = NA; seed = NULL; type = list(\"dms\", \"dec\", \"rad\")\n\nsim_snetobs <- function(survey.net, red = TRUE, Hz0 = NA, seed = NULL, type = list(\"dms\", \"dec\", \"rad\")){\n\n survey.net[[2]] <- survey.net[[2]] %>% dplyr::mutate_at(., .vars = c(\"e_cent_from\", \"e_cent_to\", \"e_focus\", \"e_air\"), ~replace(., is.na(.), 0))\n\n if(!is.na(Hz0)){\n if(length(Hz0) == 1) Hz0 <- rep(Hz0, length(unique(survey.net[[2]]$from)))\n if(length(Hz0) != length(unique(survey.net[[2]]$from))) stop(\"Hz0 must be of length either 1 or number of stations\")\n }else{\n Hz0 <- sample(1:359, size = length(unique(survey.net[[2]]$from)))\n }\n\n\n if(!is.null(seed)){\n if(length(seed) == 1) seed <- rep(seed, length(unique(survey.net[[2]]$from)))\n if(length(seed) != length(unique(survey.net[[2]]$from))) stop(\"seed must be of length either 1 or the number of stations\")\n }else{\n seed <- runif(length(unique(survey.net[[2]]$from)),0,100)#\n }\n\n nto <- survey.net[[2]] %>% st_drop_geometry() %>% dplyr::mutate(from = as_factor(from)) %>% group_by(from) %>% summarize(n = n())\n survey.net[[2]] <- survey.net[[2]] %>%\n group_by(from) %>%\n mutate(id = row_number()) %>% ungroup() %>%\n dplyr::mutate(Hz0 = rep(Hz0, nto$n),\n Hz_sim = NA,\n dist_sim = NA,\n seed_from = rep(seed, nto$n),\n seed_to = seed_from + id)\n\n survey.net[[2]] <- survey.net[[2]] %>% split(., f = factor(.$from)) %>% lapply(., function(x) sim_st(survey.net.st = x, red = red)) %>% do.call(rbind,.) %>% dplyr::arrange(ID)\n\n # fixed_points <- survey.net[[1]] %>% dplyr::filter(FIX_2D | FIX_1D) %>% .$Name\n #\n # if(length(fixed_points) > 1){\n # survey.net[[2]][survey.net[[2]]$from %in% fixed_points & survey.net[[2]]$to %in% fixed_points, \"HD\"] <- NA\n # }\n\n return(survey.net)\n}\n\n#hz = mreza_sim$observations$Hz[2]; sd_Hz = mreza_sim$observations$sd_Hz[2]; e_colim = 15; sd_e_colim = 0.5; ng = 1; faces = \"f1f2\"\n\nsim_face_hz <- function(hz, sd_Hz, e_colim, sd_e_colim, ng, faces = list(\"mean\", \"f1f2\"), sd_sd_Hz, seed = NULL, hz_of = TRUE){\n faces = faces[1]\n if(faces == \"f1f2\"){\n set.seed(seed)\n hz_f1 <- hz - (rnorm(1, e_colim, sd_e_colim))/3600 + rnorm(1, 0, sd_Hz)*sqrt(2)/3600\n if(hz_f1 < 0){hz_f1 + 360}\n if(hz_f1 >= 180){\n set.seed(seed)\n hz_f2 <- (hz_f1 - 180) + (rnorm(1, e_colim, sd_e_colim))/3600 + rnorm(1, 0, sd_Hz)*sqrt(2)/3600\n }else{\n set.seed(seed)\n hz_f2 <- (hz_f1 + 180) + (rnorm(1, e_colim, sd_e_colim))/3600 + rnorm(1, 0, sd_Hz)*sqrt(2)/3600\n }\n hz <- data.frame(hz_f1 = ifelse(hz_f1 < 0, hz_f1 + 360, hz_f1) , hz_f2 = ifelse(hz_f2 < 0, hz_f2 + 360, hz_f2))\n }\n return(hz)\n}\n\n#a <- sim_face_hz(hz = mreza_sim$observations$Hz[1], sd_Hz = mreza_sim$observations$sd_Hz[1], e_colim = 15, sd_e_colim = 0.5, ng = 1, faces = \"f1f2\")\n#a\n#(a[2]+180-a[1])*3600\n\n\n\n", "meta": {"hexsha": "0c5664234e1504cb0a1f2ed9ccf5c8fd4da6b392", "size": 7315, "ext": "r", "lang": "R", "max_stars_repo_path": "R/deprecated/simulation_funs.r", "max_stars_repo_name": "pejovic/Surveyor", "max_stars_repo_head_hexsha": "40839e3cea8836b2b8e2681ffed591a6567ab173", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2020-11-14T22:40:36.000Z", "max_stars_repo_stars_event_max_datetime": "2020-11-14T22:40:36.000Z", "max_issues_repo_path": "R/deprecated/simulation_funs.r", "max_issues_repo_name": "pejovic/Surveyor", "max_issues_repo_head_hexsha": "40839e3cea8836b2b8e2681ffed591a6567ab173", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/deprecated/simulation_funs.r", "max_forks_repo_name": "pejovic/Surveyor", "max_forks_repo_head_hexsha": "40839e3cea8836b2b8e2681ffed591a6567ab173", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 45.1543209877, "max_line_length": 177, "alphanum_fraction": 0.557621326, "num_tokens": 2564, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.855851154320682, "lm_q2_score": 0.6261241772283034, "lm_q1q2_score": 0.5358690998289307}} {"text": "require(ggplot2)\nrequire(scales)\nSL = 32\np = 3\n\nf = function(x,k,l,SL) { 1 - (1 - (1 - x/SL)^k)^l }\n\nK = function(l,p,alpha,L) {round(log(1-(1-alpha)^(1/l))/log(1-p/L));}\n\nK(4,2,0.95,10)\n\nalpha=0.95\nggplot(data.frame(x = c(1, 3*p)), aes(x)) + theme_classic()+\n mapply(function( L) {\n stat_function(fun = f, args = list(k = K(l=L,p=p,alpha=alpha,L=SL), l=L, SL=SL), aes(color=paste(L,K(l=L,p=p,alpha=alpha,L=SL),sep=\"/\")))\n },L=c(7:14))+\n geom_vline(xintercept = p,color=\"black\",linetype=2)+\n geom_hline(yintercept = alpha,color=\"black\",linetype=2)+\n scale_color_brewer(palette = \"Spectral\",name=\"l/h\")+ylab(\"probability of a match\")+xlab(\"Hamming distance\")+scale_linetype(name=\"l\")\nggsave(\"clus-95.pdf\",width=5,height = 4)\n\nalpha=0.1\nggplot(data.frame(x = c(1, 3*p)), aes(x)) + theme_classic()+\n mapply(function( L) {\n stat_function(fun = f, args = list(k = K(l=L,p=p,alpha=alpha,L=SL), l=L, SL=SL), aes(color=paste(L,K(l=L,p=p,alpha=alpha,L=SL),sep=\"/\")))\n },L=c(2:8))+\n geom_vline(xintercept = p,color=\"black\",linetype=2)+\n geom_hline(yintercept = alpha,color=\"black\",linetype=2)+\n scale_color_brewer(palette = \"Spectral\",name=\"l/h\")+ylab(\"probability of a match\")+xlab(\"Hamming distance\")+scale_linetype(name=\"l\")\nggsave(\"clus-10.pdf\",width=5,height = 4)\n\nalpha=0.1\nggplot(data.frame(x = c(1, 3*p)), aes(x)) + theme_classic()+\n mapply(function( L) {\n stat_function(fun = function(x,k,l,SL) (150-SL+1)*f(x,k,l,SL), args = list(k = K(l=L,p=p,alpha=alpha,L=SL), l=L, SL=SL), aes(color=paste(L,K(l=L,p=p,alpha=alpha,L=SL),sep=\"/\")))\n },L=c(2:8))+\n geom_vline(xintercept = 4,color=\"black\",linetype=2)+\n geom_hline(yintercept = 3,color=\"black\",linetype=2)+\n scale_color_brewer(palette = \"Dark2\",name=\"l/k\")+ylab(\"probability of a match\")+xlab(\"distance\")+scale_linetype(name=\"l\")\n\n\n\nalpha=0.1\nggplot(data.frame(x = c(1, 16)), aes(x)) + theme_classic()+\n mapply(FUN=function( L, K) {\n stat_function(fun = function(x,k,l,SL) (1+150-SL)*f(x,k,l,SL), \n args = list(k = K, l=L, SL=SL), \n aes(color=as.factor(K),linetype=as.factor(L)))\n },L=rep(c(2),each=5), K=c(14))+\n #geom_vline(xintercept = 3,color=\"black\",linetype=2)+\n #geom_vline(xintercept = SL*0.4,color=\"black\",linetype=2)+\n #geom_hline(yintercept = 90,color=\"black\",linetype=2)+\n scale_y_continuous(lim=c(0.1,100))+\n scale_linetype_manual(values=c(1,2,3),name=expression(h))+\n scale_x_continuous(labels=function(x) paste(x,percent(x/SL),sep=\"\\n\"))+\n scale_color_manual(name=\"l\",values = c(\"black\",\"gray80\"))+\n ylab(\"Expected number of kmer matches per 150bp read\")+\n xlab(\"Hamming distance\")+\n #stat_function(fun = function(x) (1+150-SL)*f(x,35-7,1,35)/2, color=\"blue\",linetype=1)+\n geom_vline(xintercept = 3,color=\"red\",linetype=3)+\n geom_vline(xintercept = 10,color=\"red\",linetype=3)+\n #theme(legend.position = \"none\")+\nggsave(\"exp-def.pdf\",width=5,height = 4)\n", "meta": {"hexsha": "e9e2b0a9b4262105fabc90c9f8b367ce2d959262", "size": 2892, "ext": "r", "lang": "R", "max_stars_repo_path": "old/test-2.r", "max_stars_repo_name": "noraracht/lsh_scripts", "max_stars_repo_head_hexsha": "aa6614449019f11e5f3590fc10ca3dc882addd65", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "old/test-2.r", "max_issues_repo_name": "noraracht/lsh_scripts", "max_issues_repo_head_hexsha": "aa6614449019f11e5f3590fc10ca3dc882addd65", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "old/test-2.r", "max_forks_repo_name": "noraracht/lsh_scripts", "max_forks_repo_head_hexsha": "aa6614449019f11e5f3590fc10ca3dc882addd65", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 45.1875, "max_line_length": 181, "alphanum_fraction": 0.648340249, "num_tokens": 1046, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8652240825770432, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5353837331026475}} {"text": "#' @title\n#' Select ranks with cross validation\n#'\n#' @description\n#' This function selects the optimal ranks \\code{(r, rx, ry)} using a cross\n#' validation procedure. The blockwise coordinate descent algorithm is used to fit\n#' the model with any combinations of \\code{(r, rx, ry)}.\n#'\n#' @usage\n#' NRRR.cv(Y, X, nfold = 10, norder = NULL, Ag0 = NULL, Bg0 = NULL,\n#' jx, jy, p, d, n, maxiter = 300, conv = 1e-4,\n#' method = c('RRR','RRS')[1], lambda=0,\n#' dimred = c(TRUE,TRUE,TRUE), rankfix = NULL, xrankfix = NULL,\n#' yrankfix = NULL, lang=c('R','Rcpp')[1])\n#'\n#' @param Y response matrix of dimension n-by-jy*d.\n#' @param X design matrix of dimension n-by-jx*p.\n#' @param nfold the number of folds used in cross validation. Default is 10.\n#' @param norder a vector of length n that assigns samples to multiple folds for cross validation.\n#' @param Ag0 an initial estimator of matrix U. If \\code{Ag0 = NULL} then generate it\n#' by \\code{\\link{NRRR.ini}}. Default is NULL. If \\code{lang = 'Rcpp'},\n#' then \\code{Ag0} is automatically generated by \\code{\\link{NRRR.ini}}.\n#' @param Bg0 an initial estimator of matrix V, if \\code{Bg0 = NULL} then generate it\n#' by \\code{\\link{NRRR.ini}}. Default is NULL. If \\code{lang = 'Rcpp'},\n#' then \\code{Bg0} is automatically generated by \\code{\\link{NRRR.ini}}.\n#' @param jx number of basis functions to expand the functional predictor.\n#' @param jy number of basis functions to expand the functional response.\n#' @param p number of predictors.\n#' @param d number of responses.\n#' @param n sample size.\n#' @param maxiter the maximum iteration number of the\n#' blockwise coordinate descent algorithm. Default is 300.\n#' @param conv the tolerance level used to control the convergence of the\n#' blockwise coordinate descent algorithm. Default is 1e-4.\n# @param quietly a logical value with two options. FALSE (default): show the\n# rank selection process; TRUE: do not show the process.\n#' @param method 'RRR' (default): no additional ridge penalty; 'RRS': add an\n#' additional ridge penalty.\n#' @param lambda the tuning parameter to control the amount of ridge\n#' penalization. It is only used when \\code{method = 'RRS'}.\n#' Default is 0.\n# @param ic the user-specified information criterion. Four options are available,\n# including BIC, BICP, AIC, GCV.\n#' @param dimred a vector of logical values to decide whether to use cross validation\n#' do rank selection on certain dimensions. TRUE means the rank is selected\n#' by cross validation.\n#' If \\code{dimred[1] = FALSE}, r is provided by \\code{rankfix}\n#' or \\eqn{min(jy*d, rank(X))};\n#' If \\code{dimred[2] = FALSE}, rx equals to \\code{xrankfix} or p; If \\code{dimred[3] = FALSE},\n#' ry equals to \\code{yrankfix} or d. Default is \\code{c(TRUE, TRUE, TRUE)}.\n#' @param rankfix a user-provided value of r when \\code{dimred[1] = FALSE}. Default is NULL\n#' which leads to \\eqn{r = min(jy*d, rank(X))}.\n#' @param xrankfix a user-provided value of rx when \\code{dimred[2] = FALSE}. Default is NULL\n#' which leads to \\code{rx = p}.\n#' @param yrankfix a user-provided value of ry when \\code{dimred[3] = FALSE}. Default is NULL\n#' which leads to \\code{ry = d}.\n#' @param lang 'R' (default): the R version function is used; 'Rcpp': the Rcpp\n#' version function is used.\n#'\n#'\n#' @return The function returns a list:\n#' \\item{Ag}{the estimated U.}\n#' \\item{Bg}{the estimated V.}\n#' \\item{Al}{the estimated A.}\n#' \\item{Bl}{the estimated B.}\n#' \\item{C}{the estimated coefficient matrix C.}\n#' \\item{df}{the estimated degrees of freedom of the selected model.}\n#' \\item{sse}{the sum of squared errors of the selected model.}\n#' \\item{ic}{a vector containing values of BIC, BICP, AIC, GCV of the selected model.}\n#' \\item{rx_path}{a matrix displays the path of selecting rx with cross validation.}\n#' \\item{ry_path}{a matrix displays the path of selecting ry with cross validation.}\n#' \\item{r_path}{a matrix displays the path of selecting r with cross validation.}\n# \\item{r_fitseq}{a sequence of possible r values.}\n# \\item{iter}{the number of iterations needed to converge in the selected model.}\n#' \\item{rank}{the estimated r.}\n#' \\item{rx}{the estimated rx.}\n#' \\item{ry}{the estimated ry.}\n#'\n#' @details\n#' A three-dimensional grid search procedure of the rank\n#' values is performed, and the best model is chosen as the one with the\n#' smallest prediction error. Instead of a nested rank selection method, we apply a\n#' one-at-a-time selection approach. We first set \\eqn{rx = p, ry = d}, and\n#' select the best local rank \\eqn{\\hat r} among the models with\n#' \\eqn{1 \\le r \\le min(rank(X), jy*d)}. We then fix the local rank at\n#' \\eqn{\\hat r} and repeat a similar procedure to determine \\eqn{\\hat rx}\n#' and \\eqn{\\hat ry}, one at a time. Finally, with fixed \\eqn{\\hat rx} and \\eqn{\\hat ry},\n#' we refine the estimation of r.\n#'\n#' @references Liu, X., Ma, S., & Chen, K. (2020).\n#' Multivariate Functional Regression via Nested Reduced-Rank Regularization.\n#' arXiv: Methodology.\n#'\n# @importFrom rrpack cv.rrr\n#' @export\n#' @examples\n#' library(NRRR)\n#' set.seed(1)\n#' # Simulation setting 1 in NRRR paper\n#' simDat <- NRRR.sim(n = 100, ns = 60, nt = 60, r = 5, rx = 3, ry = 3,\n#' jx = 8, jy = 8, p = 10, d = 10, s2n = 1,\n#' rho_X = 0.5, rho_E = 0, Sigma = \"CorrAR\")\n#' # using R function\n#' fit_R <- with(simDat, NRRR.cv(Yest, Xest, nfold = 10, norder = NULL,\n#' Ag0 = NULL, Bg0 = NULL, jx = 8, jy = 8, p = 10, d = 10,\n#' n = 100, maxiter = 300, conv = 1e-4,\n#' method = c(\"RRR\", \"RRS\")[1], lambda = 0,\n#' dimred = c(TRUE, TRUE, TRUE), rankfix = NULL,\n#' xrankfix = NULL, yrankfix = NULL, lang=c('R','Rcpp')[1]))\n#' # using Rcpp function\n#' fit_Rcpp <- with(simDat, NRRR.cv(Yest, Xest, nfold = 10, norder = NULL,\n#' Ag0 = NULL, Bg0 = NULL, jx = 8, jy = 8, p = 10,\n#' d = 10, n = 100, maxiter = 300, conv = 1e-4,\n#' method = c(\"RRR\", \"RRS\")[1], lambda = 0,\n#' dimred = c(TRUE, TRUE, TRUE), rankfix = NULL,\n#' xrankfix = NULL, yrankfix = NULL, lang=c('R','Rcpp')[2]))\nNRRR.cv <- function(Y,X,nfold=10,norder=NULL,Ag0=NULL,Bg0=NULL,jx,jy,p,d,n,\n maxiter=300,conv=1e-4,#quietly=FALSE,\n method=c('RRR','RRS')[1],lambda=0,\n dimred = c(TRUE,TRUE,TRUE),\n rankfix=NULL,xrankfix=NULL,\n yrankfix=NULL, lang=c('R','Rcpp')[1]\n){\n #require(rrpack)\n if (method == \"RRS\" & lambda == 0) stop(\"A positive tuning parameter should be provided when 'RRS' is used.\")\n if (!(lang %in% c('R','Rcpp'))) stop(\"Please choose from 'R' and 'Rcpp'\")\n\n xr <- sum(svd(X)$d>1e-2)\n if (is.null(norder))\n norder <- sample(seq_len(n),n)\n\n # initialize r\n if(dimred[1]){\n fitRRR <- cv.rrr(Y,X,nfold=10,norder = norder)\n rest <- fitRRR$rank\n # If zero fit\n if(rest==0){\n fitRRR <- RRR(Y,X,nrank=1)\n rest <- fitRRR$rank\n }\n # if(!quietly) {\n # cat(\"Initial r = \",rest, \"\\n\",sep=\"\")\n # }\n }else{\n rest <- ifelse(is.null(rankfix),min(ncol(Y),ncol(X),xr),rankfix)\n }\n rfit <- rest\n\n if (lang == 'R') {\n # Select rx by cross validation\n rx_path <- matrix(ncol = nfold, nrow = p, NA)\n if (p == 1) {\n rxest <- p\n } else {\n if(dimred[2]){\n ndel <- round(n/nfold)\n for (f in seq_len(nfold)){\n if (f != nfold) {\n iddel <- norder[(1 + ndel * (f - 1)):(ndel * f)]\n }\n else {\n iddel <- norder[(1 + ndel * (f - 1)):n]\n }\n ndel <- length(iddel)\n nf <- n - ndel\n idkeep <- (seq_len(n))[-iddel]\n Xf <- X[-iddel, ]\n Xfdel <- X[iddel, ]\n Yf <- Y[-iddel, ]\n Yfdel <- Y[iddel, ]\n for (i in seq_len(p)){\n rxfit <- i\n ryfit <- d\n fit1 <- NRRR.est(Yf,Xf,NULL,NULL,rini=rfit,rfit,rxfit,ryfit,jx,jy,p,\n d,n=nf,maxiter=maxiter,conv=conv,#quietly=TRUE,\n method=method,lambda=lambda)\n rx_path[i,f] <- sum((Yfdel-Xfdel%*%fit1$C)^2)\n }\n }\n index <- order(colSums(rx_path))\n crerr <- rowSums(rx_path[, index])/length(index) * nfold\n rxest <- which.min(crerr)\n } else {\n rxest <- ifelse(is.null(xrankfix),p,xrankfix)\n }\n }\n\n # if(!quietly) {\n # cat(\"Selected rx = \",rxest, \"\\n\",sep=\"\")\n # }\n\n # Select ry by cross validation\n ry_path <- matrix(ncol = nfold, nrow = d, NA)\n if (d == 1) {\n ryest <- d\n } else {\n if(dimred[3]){\n ndel <- round(n/nfold)\n for (f in seq_len(nfold)){\n if (f != nfold) {\n iddel <- norder[(1 + ndel * (f - 1)):(ndel * f)]\n }\n else {\n iddel <- norder[(1 + ndel * (f - 1)):n]\n }\n ndel <- length(iddel)\n nf <- n - ndel\n idkeep <- (seq_len(n))[-iddel]\n Xf <- X[-iddel, ]\n Xfdel <- X[iddel, ]\n Yf <- Y[-iddel, ]\n Yfdel <- Y[iddel, ]\n for (i in seq_len(d)){\n rxfit <- rxest\n ryfit <- i\n fit1 <- NRRR.est(Yf,Xf,NULL,NULL,rini=rfit,rfit,rxfit,ryfit,jx,jy,p,\n d,n=nf,maxiter=maxiter,conv=conv,#quietly=TRUE,\n method=method,lambda=lambda)\n ry_path[i,f] <- sum((Yfdel-Xfdel%*%fit1$C)^2)\n }\n }\n index <- order(colSums(ry_path))\n crerr <- rowSums(ry_path[, index])/length(index) * nfold\n ryest <- which.min(crerr)\n } else {\n ryest <- ifelse(is.null(yrankfix),d,yrankfix)\n }\n }\n\n # if(!quietly) {\n # cat(\"Selected ry = \",ryest, \"\\n\",sep=\"\")\n # }\n\n\n # refine rank selection by cross validation\n rfitseq <- max(1,rest-5):min(rest+5,min(n,p*jx,d*jy))\n r_path <- matrix(ncol = nfold, nrow = length(rfitseq), NA)\n if(dimred[1]){\n rxfit <- rxest\n ryfit <- ryest\n\n ndel <- round(n/nfold)\n for (f in seq_len(nfold)){\n if (f != nfold) {\n iddel <- norder[(1 + ndel * (f - 1)):(ndel * f)]\n }\n else {\n iddel <- norder[(1 + ndel * (f - 1)):n]\n }\n ndel <- length(iddel)\n nf <- n - ndel\n idkeep <- (seq_len(n))[-iddel]\n Xf <- X[-iddel, ]\n Xfdel <- X[iddel, ]\n Yf <- Y[-iddel, ]\n Yfdel <- Y[iddel, ]\n for (i in 1:length(rfitseq)){\n rfit <- rfitseq[i]\n fit1 <- NRRR.est(Yf,Xf,NULL,NULL,rini=rfit,rfit,rxfit,ryfit,jx,jy,p,\n d,n=nf,maxiter=maxiter,conv=conv,#quietly=TRUE,\n method=method,lambda=lambda)\n r_path[i,f] <- sum((Yfdel-Xfdel%*%fit1$C)^2)\n }\n }\n index <- order(colSums(r_path))\n crerr <- rowSums(r_path[, index])/length(index) * nfold\n rest <- rfitseq[which.min(crerr)]\n fit <- NRRR.est(Y,X,NULL,NULL,rini=rest,rest,rxfit,ryfit,jx,jy,p,\n d,n,maxiter=maxiter,conv=conv,#quietly=TRUE,\n method=method,lambda=lambda)\n\n\n } else {\n rxfit <- rxest\n ryfit <- ryest\n fit <- NRRR.est(Y,X,NULL,NULL,rini=rest,rest,rxfit,ryfit,jx,jy,\n p,d,n,maxiter=maxiter,conv=conv,#quietly=TRUE,\n method=method,lambda=lambda)\n\n }\n if (fit$iter == maxiter) stop(\"The algorithm reaches the maximum iteration.\")\n\n\n return(list(Ag=fit$Ag,Bg=fit$Bg,Al=fit$Al,Bl=fit$Bl,C=fit$C,df=fit$df,\n sse=fit$sse,ic=fit$ic,#obj=fit$obj,\n rx_path=rx_path,ry_path=ry_path,\n r_path=r_path,#rfitseq=rfitseq,\n #iter=fit$iter,\n rank=rest,rx=rxest,ry=ryest))\n } else {\n method <- ifelse(method == 'RRR', 1, 2)\n\n dimred1 <- ifelse(dimred[1],1,0)\n dimred2 <- ifelse(dimred[2],1,0)\n dimred3 <- ifelse(dimred[3],1,0)\n\n xrankfix <- ifelse(is.null(xrankfix),0,xrankfix)\n yrankfix <- ifelse(is.null(yrankfix),0,yrankfix)\n\n norder <- norder - 1\n\n fit <- nrrr_cv_my(Y, X, norder, nfold, xr, rfit, xrankfix, yrankfix,\n jx, jy, p, d, n, maxiter, method,\n dimred1, dimred2, dimred3, conv, lambda)\n if( sum(fit$rxErrmat)>0 | sum(fit$ryErrmat)>0 | sum(fit$rErrmat)>0 ) stop('Error occurs or the algorithm reaches the maximum iteration')\n\n\n return(list(Ag=fit$Ag,Bg=fit$Bg,Al=fit$Al,Bl=fit$Bl,C=fit$C,df=fit$df,\n sse=fit$sse,ic=fit$ic,#obj=fit$obj,\n rx_path=fit$rx_path,ry_path=fit$ry_path,\n r_path=fit$r_path,\n rank=fit$rank,rx=fit$rx,ry=fit$ry\n #,rxErrmat=fit$rxErrmat,ryErrmat=fit$ryErrmat,rErrmat=fit$rErrmat\n ))\n }\n}\n", "meta": {"hexsha": "1e488c7a99a4fd78e7dbcaf1555d56e90f211565", "size": 13029, "ext": "r", "lang": "R", "max_stars_repo_path": "R/NestRRR.cv.select.r", "max_stars_repo_name": "xliu-stat/NRRR", "max_stars_repo_head_hexsha": "e51d9df7500b287f8c4daa8839f4cc1782525cef", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/NestRRR.cv.select.r", "max_issues_repo_name": "xliu-stat/NRRR", "max_issues_repo_head_hexsha": "e51d9df7500b287f8c4daa8839f4cc1782525cef", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/NestRRR.cv.select.r", "max_forks_repo_name": "xliu-stat/NRRR", "max_forks_repo_head_hexsha": "e51d9df7500b287f8c4daa8839f4cc1782525cef", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 41.1009463722, "max_line_length": 139, "alphanum_fraction": 0.5612863612, "num_tokens": 4075, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929002541068, "lm_q2_score": 0.685949467848392, "lm_q1q2_score": 0.5350357148548285}} {"text": "C <- matrix(c(-2,4,6,2,4,1),2,3)\r\nD <- matrix(c(3,9,7,2,5,1,4,-2,8),3,3)\r\nE <- matrix(c(4,8,1,-1,6,-6,0,4,7),3,3)\r\n# C+D cannot be computed because of different dimensions.\r\n\r\nx=D+E\r\ny=D-E\r\nz=C%*%D\r\np=t(E)\r\nx\r\ny\r\nz\r\np", "meta": {"hexsha": "21fe66c357db87f086aea25ae13e6ca3ef807d2d", "size": 217, "ext": "r", "lang": "R", "max_stars_repo_path": "An_Introduction_To_Statistical_Methods_And_Data_Analysis_by_R_Lyman_Ott_And_Michael_Longnecker/CH12/EX12.26/Ex12_26.r", "max_stars_repo_name": "prashantsinalkar/R_TBC_Uploads", "max_stars_repo_head_hexsha": "b3f3a8ecd454359a2e992161844f2fb599f8238a", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "An_Introduction_To_Statistical_Methods_And_Data_Analysis_by_R_Lyman_Ott_And_Michael_Longnecker/CH12/EX12.26/Ex12_26.r", "max_issues_repo_name": "prashantsinalkar/R_TBC_Uploads", "max_issues_repo_head_hexsha": "b3f3a8ecd454359a2e992161844f2fb599f8238a", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "An_Introduction_To_Statistical_Methods_And_Data_Analysis_by_R_Lyman_Ott_And_Michael_Longnecker/CH12/EX12.26/Ex12_26.r", "max_forks_repo_name": "prashantsinalkar/R_TBC_Uploads", "max_forks_repo_head_hexsha": "b3f3a8ecd454359a2e992161844f2fb599f8238a", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 3, "max_forks_repo_forks_event_min_datetime": "2020-04-07T16:44:56.000Z", "max_forks_repo_forks_event_max_datetime": "2020-11-13T06:35:28.000Z", "avg_line_length": 16.6923076923, "max_line_length": 58, "alphanum_fraction": 0.534562212, "num_tokens": 113, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7718434768461855, "lm_q2_score": 0.6926419894793246, "lm_q1q2_score": 0.534611201369381}} {"text": "softmax <- function (x) {\n m <- max(x)\n exps <- exp(x - m)\n s <- sum(exps)\n return(exps / s)\n}\nscore <- function(input) {\n if ((input[3]) >= (2.45)) {\n var0 <- -0.21995015\n } else {\n var0 <- 0.4302439\n }\n if ((input[3]) >= (2.45)) {\n var1 <- -0.19691855\n } else {\n var1 <- 0.29493433\n }\n if ((input[3]) >= (2.45)) {\n if ((input[4]) >= (1.75)) {\n var2 <- -0.20051816\n } else {\n var2 <- 0.36912444\n }\n } else {\n var2 <- -0.21512198\n }\n if ((input[3]) >= (2.45)) {\n if ((input[3]) >= (4.8500004)) {\n var3 <- -0.14888482\n } else {\n var3 <- 0.2796613\n }\n } else {\n var3 <- -0.19143805\n }\n if ((input[4]) >= (1.6500001)) {\n var4 <- 0.40298507\n } else {\n if ((input[3]) >= (4.95)) {\n var4 <- 0.21724138\n } else {\n var4 <- -0.21974029\n }\n }\n if ((input[3]) >= (4.75)) {\n if ((input[4]) >= (1.75)) {\n var5 <- 0.28692952\n } else {\n var5 <- 0.06272897\n }\n } else {\n if ((input[4]) >= (1.55)) {\n var5 <- 0.009899145\n } else {\n var5 <- -0.19659369\n }\n }\n return(softmax(c((0.5) + ((var0) + (var1)), (0.5) + ((var2) + (var3)), (0.5) + ((var4) + (var5)))))\n}\n", "meta": {"hexsha": "9779024aad6000020dd1af651ebf41f8a216bb91", "size": 1377, "ext": "r", "lang": "R", "max_stars_repo_path": "generated_code_examples/r/classification/xgboost.r", "max_stars_repo_name": "Symmetry-International/m2cgen", "max_stars_repo_head_hexsha": "3157e0cbd5bd1ee7e044a992223c60224e2b7709", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2161, "max_stars_repo_stars_event_min_datetime": "2019-01-13T02:37:56.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-30T13:24:09.000Z", "max_issues_repo_path": "generated_code_examples/r/classification/xgboost.r", "max_issues_repo_name": "Symmetry-International/m2cgen", "max_issues_repo_head_hexsha": "3157e0cbd5bd1ee7e044a992223c60224e2b7709", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 380, "max_issues_repo_issues_event_min_datetime": "2019-01-17T15:59:29.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-31T20:59:20.000Z", "max_forks_repo_path": "generated_code_examples/r/classification/xgboost.r", "max_forks_repo_name": "Symmetry-International/m2cgen", "max_forks_repo_head_hexsha": "3157e0cbd5bd1ee7e044a992223c60224e2b7709", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 201, "max_forks_repo_forks_event_min_datetime": "2019-02-13T19:06:44.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-12T09:45:46.000Z", "avg_line_length": 22.95, "max_line_length": 103, "alphanum_fraction": 0.3674655047, "num_tokens": 505, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8438950907764119, "lm_q2_score": 0.6334102636778401, "lm_q1q2_score": 0.5345318119651219}} {"text": "#' @title Aggregate keyword-country data and compute DOI\n#'\n#' @aliases\n#' compute_doi\n#' compute_doi.numeric\n#' compute_doi.list\n#'\n#' @description\n#' The function computes degree of internationalization (DOI) for object\n#' keywords. Degree of internationalization is measured based on the\n#' distribution of country search scores.\n#'\n#' @details\n#' The function uses an inverted Gini-coefficient\n#' `dplyr::coalesce(1 - ineq::ineq(series, type = \"Gini\"), 0)`\n#' as measure for the degree of internationalization. The more uniform the\n#' distribution of search scores across all countries, the higher the inverted\n#' Gini-coefficient and the greater the degree of internationalization. In\n#' addition to the Gini-coefficient, the package uses inverted Herfindahl index\n#' `coalesce(1 - sum((series / sum(series))^2), 0)` and inverted Entropy\n#' `dplyr::coalesce(-1 * ineq::ineq(series, parameter = 1, type = \"entropy\"), 0)`\n#' as measures for internationalization.\n#'\n#' @param control Control batch for which the search score is used. Object\n#' of type `numeric`.\n#' @param object Object batch for which the keyword-country data\n#' is aggregated and DOI is computed. Object of type `numeric`.\n#' @param locations List of locations for which the search score is used.\n#' Object of type `character`. Defaults to *\"countries\"*.\n#'\n#' @seealso\n#' * [example_doi()]\n#' * [ineq::ineq()]\n#'\n#' @return\n#' Message that data was aggregated successfully. Data is written to table\n#' *data_doi*.\n#'\n#' @examples\n#' \\dontrun{\n#' compute_doi(\n#' object = 1,\n#' control = 1,\n#' locations = \"countries\"\n#' )\n#' compute_doi(\n#' object = as.list(1:5),\n#' control = 1,\n#' locations = \"countries\"\n#' )\n#' }\n#'\n#' @export\n#' @rdname compute_doi\n#' @importFrom DBI dbWriteTable\n#' @importFrom dplyr collect\n#' @importFrom dplyr filter\n#' @importFrom dplyr mutate\n#' @importFrom dplyr select\n#' @importFrom glue glue\n#' @importFrom purrr map_dbl\n#' @importFrom purrr map_lgl\n#' @importFrom purrr walk\n#' @importFrom rlang .data\n#' @importFrom tidyr nest\n#' @importFrom tidyr pivot_longer\n\ncompute_doi <- function(object, control = 1, locations = \"countries\") UseMethod(\"compute_doi\", object)\n\n#' @rdname compute_doi\n#' @method compute_doi numeric\n#' @export\n\ncompute_doi.numeric <- function(object, control = 1, locations = \"countries\") {\n control <- unlist(control)\n .check_length(control, 1)\n .check_length(locations, 1)\n .check_input(locations, \"character\")\n if (length(object) > 1) {\n compute_doi(control = control, object = as.list(object), locations = locations)\n } else {\n walk(list(control, object), .check_batch)\n if (.test_empty(table = \"data_doi\", batch_c = control, batch_o = object, locations = locations)) {\n data <- filter(.tbl_score, .data$batch_c == control & .data$batch_o == object)\n data <- collect(data)\n tmp_locations <- pull(collect(filter(.tbl_locations, .data$type == locations)), .data$location)\n data <- filter(data, .data$location %in% tmp_locations & .data$synonym == 0)\n\n # compute doi measures\n data <- pivot_longer(data, cols = contains(\"score\"), names_to = \"type\", values_to = \"score\")\n data <- nest(data, data = c(.data$location, .data$score))\n data <- mutate(data, check = map_lgl(.data$data, ~ !all(is.na(.x$score))))\n out1 <- filter(data, .data$check)\n out1 <- mutate(\n out1,\n gini = map_dbl(.data$data, ~ .compute_gini(series = .x$score)),\n hhi = map_dbl(.data$data, ~ .compute_hhi(series = .x$score)),\n entropy = map_dbl(.data$data, ~ .compute_entropy(series = .x$score))\n )\n out2 <- filter(data, !.data$check)\n out2 <- mutate(\n out2,\n gini = NA,\n hhi = NA,\n entropy = NA\n )\n out <- bind_rows(out1, out2)\n out <- select(\n out,\n .data$date,\n .data$keyword,\n .data$type,\n .data$gini,\n .data$hhi,\n .data$entropy\n )\n\n # write data\n out <- mutate(\n out,\n batch_c = control,\n batch_o = object,\n locations = locations\n )\n dbWriteTable(conn = globaltrends_db, name = \"data_doi\", value = out, append = TRUE)\n }\n message(glue(\"Successfully computed DOI | control: {control} | object: {object} [{object}/{total}]\", total = max(.keywords_object$batch)))\n }\n}\n\n#' @rdname compute_doi\n#' @method compute_doi list\n#' @export\n\ncompute_doi.list <- function(object, control = 1, locations = \"countries\") {\n walk(object, compute_doi, control = control, locations = locations)\n}\n\n#' @title Compute gini coefficient\n#'\n#' @rdname hlprs\n#' @keywords internal\n#' @noRd\n#'\n#' @importFrom dplyr coalesce\n\n.compute_gini <- function(series) {\n out <- coalesce(1 - ineq::ineq(series, type = \"Gini\"), 0)\n return(out)\n}\n\n#' @title Compute herfindahl hirschman index\n#'\n#' @rdname hlprs\n#' @keywords internal\n#' @noRd\n\n.compute_hhi <- function(series) {\n out <- coalesce(1 - sum((series / sum(series))^2), 0)\n return(out)\n}\n\n#' @title Compute entropy\n#'\n#' @rdname hlprs\n#' @keywords internal\n#' @noRd\n\n.compute_entropy <- function(series) {\n out <- coalesce(-1 * ineq::ineq(series, parameter = 1, type = \"entropy\"), 0)\n if (out == -Inf) {\n out <- 0\n }\n return(out)\n}\n", "meta": {"hexsha": "ed3166cae2cc52b7087beb9eb9a74f337258cfb0", "size": 5241, "ext": "r", "lang": "R", "max_stars_repo_path": "R/compute_doi.r", "max_stars_repo_name": "ha-pu/doiGT", "max_stars_repo_head_hexsha": "cb0583aea4f14df583d27cfcb2b305f2586255ee", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/compute_doi.r", "max_issues_repo_name": "ha-pu/doiGT", "max_issues_repo_head_hexsha": "cb0583aea4f14df583d27cfcb2b305f2586255ee", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 24, "max_issues_repo_issues_event_min_datetime": "2020-09-10T08:53:52.000Z", "max_issues_repo_issues_event_max_datetime": "2020-09-16T19:06:13.000Z", "max_forks_repo_path": "R/compute_doi.r", "max_forks_repo_name": "ha-pu/doiGT", "max_forks_repo_head_hexsha": "cb0583aea4f14df583d27cfcb2b305f2586255ee", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 29.7784090909, "max_line_length": 142, "alphanum_fraction": 0.6508299943, "num_tokens": 1463, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.8438951025545426, "lm_q2_score": 0.6334102498375401, "lm_q1q2_score": 0.5345318077457494}} {"text": "\n\n\n\n#' Fitting bivariate mixed MGL and MGL-EV copula regression models.\n#'\n#' @description \\code{MGL.reg.mixed} is used to fit bivariate MGL and MGL-EV copula regression models for continuous and semi-continuous variables.\n#' @param obs two-dimensional matrix for observations.\n#' @param U two-dimensional matrix for pseudo copula data with values in \\eqn{[0,1]} for (F(y1), F(y2)).\n#' @param U_ two-dimensional matrix for pseudo copula data for the data (F(y1), F(y2-1)).\n#' @param X design matrix.\n#' @param copula 'MGL', 'MGL180', \"MGL-EV\", \"MGL-EV180\"\n#' @param umin threshold value used in the semi-continuous data.\n#' @param f two-dimensional matrix for the density function of marginal distributions.\n#' @param initpar Initial values for the parameters to be optimized over.\n#' @param hessian Logical. Should a numerically differentiated Hessian matrix be returned?\n#' @param ... additional arguments, see \\code{\\link[stats]{optim}} for more details.\n#' @importFrom stats qbeta\n#' @importFrom stats optim\n#' @return A list containing the following components:\n#' * loglike: the value of the estimated maximum of the loglikelihood function.\n#' * copula: the name of the fitted copula. \"MGL180\" and \"MGL-EV180\" denote the survival MGL and MGL-EV copula respectively.\n#' * estimates: the point at which the maximum value of the loglikelihood is obtained.\n#' * se: the standard errors of the estimators.\n#' * AIC, BIC: the goodness fit of the regression models.\n#' * hessian: the hessian at the estimated maximum of the loglikelihood (if requested).\n#' @details\n#' The estimation method is performed via \\code{\\link[stats]{optim}} function. Y1 and Y2 are both continuous variables.\n#' * Y1: continuous data.\n#' * Y2: semi-continuous data where Y2>umin is continuous and Y2<=umin is discrete.\n#'\n#' copula: \"MGL180\" and \"MGLEV180\" denote the survival MGL and survival MGL-EV copula respectively.\n#' * For \"Gumbel\" regression model, the copula parameter \\deqn{\\delta_i = \\exp(X\\beta) + 1.}\n#' * For \"MGL\", \"MGL180\", \"MGL-EV\", \"MGL-EV180\" regression model, the copula parameter \\deqn{\\delta_i = \\exp(X\\beta),} where \\eqn{\\beta} is the vector of coefficients to be estimated in the copula regression.\n#'\n#' @examples\n#' library(rMGLReg)\n#' u <- cbind(earthqCHI$u1, earthqCHI$u2)\n#' u_ <- cbind(earthqCHI$u1, earthqCHI$u2_)\n#' y <- cbind(earthqCHI$y1, earthqCHI$y2)\n#' f <- cbind(earthqCHI$f1, earthqCHI$f2)\n#' obs <- y\n#' U <- u\n#' U_ <- u_\n#' umin <- 20\n#' library(splines)\n#' X <- ns(earthqCHI$year, knots = quantile(earthqCHI$year, c(0.333, 0.667)), intercept = TRUE)\n#' m.MGL180 <- MGL.reg.mixed(obs = y, U = U, U_ = U_, umin = umin, f = f, X = X,\n#' copula = \"MGL180\",\n#' method = \"Nelder-Mead\",\n#' control = list(maxit = 100000),\n#' initpar = c(0.64, 1.2, 1, -0.2))\n#'\n#' m.MGLEV180 <- MGL.reg.mixed(obs = y, U = U, U_ = U_, umin = umin, f = f, X = X,\n#' copula = \"MGL-EV180\",\n#' method = \"Nelder-Mead\",\n#' control = list(maxit = 100000),\n#' initpar = c(-0.32, 1, 1, 1))\n#' m.MGL180\n#' m.MGLEV180\n#'\n#' @export\n#'\n#'\nMGL.reg.mixed <- function(obs, U, U_, f, X, copula = c(\n \"MGL\", \"MGL180\", \"MGL-EV\",\n \"MGL-EV180\",\n \"Gumbel\", 'MGB2'\n ),\n umin = 0,\n hessian = TRUE, initpar, ...) {\n dcMGL.reg <- function(U, param) {\n dim <- length(U)\n a <- 1 / param[1]\n q <- qbeta(1 - U, shape1 = 0.5, shape2 = a) / (1 - qbeta(1 - U, shape1 = 0.5, shape2 = a))\n logdc <- (dim - 1) * lgamma(a) + lgamma(a + dim / 2) - dim * lgamma(a + 0.5) + (a + 0.5) * sum(log(q + 1)) - (a + dim / 2) * log(sum(q) + 1)\n out <- exp(logdc)\n return(out)\n }\n\n dcMGL180.reg <- function(U, param) {\n dcMGL.reg(1 - U, param = param)\n }\n dcMGLEV180.reg <- function(U, param) {\n u1 <- U[1]\n u2 <- U[2]\n as.numeric(dcMGLEV180.bivar(u1, u2, param = param[1]))\n }\n\n dcMGLEV.reg <- function(U, param) {\n u1 <- U[1]\n u2 <- U[2]\n as.numeric(dcMGLEV.bivar(u1, u2, param = param[1]))\n }\n\n dgumcop.reg <- function(U, param) {\n as.numeric(fCopulae::devCopula(u = U[1], v = U[2], type = \"gumbel\", param = param[1])) # Bivariate Extreme\n }\n\n hcMGL.reg <- function(U, param) {\n hfunc1 <- hcMGL.bivar(u1 = U[1], u2 = U[2], pars = param[1])$hfunc1\n hfunc2 <- hcMGL.bivar(u1 = U[1], u2 = U[2], pars = param[1])$hfunc2\n out <- list(hfunc1 = hfunc1, hfunc2 = hfunc2)\n out\n }\n\n hcMGL180.reg <- function(U, param) {\n hfunc1 <- 1 - hcMGL.bivar(u1 = 1 - U[1], u2 = 1 - U[2], pars = param[1])$hfunc1\n hfunc2 <- 1 - hcMGL.bivar(u1 = 1 - U[1], u2 = 1 - U[2], pars = param[1])$hfunc2\n out <- list(hfunc1 = hfunc1, hfunc2 = hfunc2)\n out\n }\n\n hcMGLEV180.reg <- function(U, param) {\n hfunc1 <- hcMGLEV180.bivar(u1 = U[1], u2 = U[2], param = param[1])$hfunc1\n hfunc2 <- hcMGLEV180.bivar(u1 = U[1], u2 = U[2], param = param[1])$hfunc2\n out <- list(hfunc1 = hfunc1, hfunc2 = hfunc2)\n out\n }\n hcMGLEV.reg <- function(U, param) {\n hfunc1 <- 1 - hcMGLEV180.bivar(u1 = 1 - U[1], u2 = 1 - U[2], param = param[1])$hfunc1\n hfunc2 <- 1 - hcMGLEV180.bivar(u1 = 1 - U[1], u2 = 1 - U[2], param = param[1])$hfunc2\n out <- list(hfunc1 = hfunc1, hfunc2 = hfunc2)\n out\n }\n\n hgumcop.reg <- function(U, param) {\n VineCopula::BiCopHfunc(u1 = U[1], u2 = U[2], family = 4, par = param[1])\n }\n\n if (copula == \"MGL\") {\n dcop <- dcMGL.reg\n hcop <- hcMGL.reg\n } else if (copula == \"MGL180\") {\n dcop <- dcMGL180.reg\n hcop <- hcMGL180.reg\n } else if (copula == \"MGL-EV\") {\n dcop <- dcMGLEV.reg\n hcop <- hcMGLEV.reg\n } else if (copula == \"MGL-EV180\") {\n dcop <- dcMGLEV180.reg\n hcop <- hcMGLEV180.reg\n } else if (copula == \"Gumbel\") {\n dcop <- dgumcop.reg\n hcop <- hgumcop.reg\n }\n # Obs1 <- obs[, 1] # y1 - the first vector of observations\n Obs2 <- obs[, 2] # y2 - the second vector of observations\n f1 <- f[, 1]\n f2 <- f[, 2]\n copLogL <- function(pars, X) {\n ll <- 0\n if (copula == \"Gumbel\") {\n delta <- exp(X %*% pars) + 1\n for (i in seq_len(nrow(X))) {\n if (Obs2[i] <= umin) {\n ll[i] <- f1[i] * (hcop(U[i, ], param = delta[i])$hfunc1 - hcop(U_[i, ], param = delta[i])$hfunc1)\n } else {\n ll[i] <- f1[i] * f2[i] * dcop(U[i, ], param = delta[i])\n }\n }\n } else if(copula == \"MGB2\"){\n p1 <- exp(pars[ncol(X) + 1])\n p2 <- exp(pars[ncol(X) + 2])\n q <- exp(X%*%pars[1:(ncol(X))])\n for (i in seq_len(nrow(X))) {\n if (Obs2[i] <= umin) {\n ll[i] <- f1[i] * (hcMGB2.bivar(u1 = U[i,1], u2 = U[i,2], pars1 = p1, pars2 = p2, pars3 = q[i])$hfunc1 - hcMGB2.bivar(u1 = U[i,1], u2 = U_[i,2], pars1 = p1, pars2 = p2, pars3 = q[i])$hfunc1)\n } else {\n ll[i] <- f1[i] * f2[i] * dcMGB2.bivar(u1 = U[i,1], u2 = U[i,2], pars1 = p1, pars2 = p2, pars3 = q[i])\n }\n }\n } else {\n delta <- exp(X %*% pars)\n for (i in seq_len(nrow(X))) {\n if (Obs2[i] <= umin) {\n ll[i] <- f1[i] * (hcop(U[i, ], param = delta[i])$hfunc1 - hcop(U_[i, ], param = delta[i])$hfunc1)\n } else {\n ll[i] <- f1[i] * f2[i] * dcop(U[i, ], param = delta[i])\n }\n }\n }\n\n res <- - sum((log(ll)))\n return(res)\n }\n\n resopt <- optim(\n par = initpar,\n fn = copLogL,\n X = X,\n hessian = hessian, ...\n )\n resopt\n\n list(\n loglike = -resopt$value,\n copula = list(name = copula),\n estimates = resopt$par,\n se = sqrt(diag(solve(resopt$hessian))),\n hessian = -resopt$hessian,\n AIC = 2 * length(resopt$par) + 2 * resopt$value,\n BIC = log(nrow(U)) * length(resopt$par) + 2 * resopt$value\n )\n}\n", "meta": {"hexsha": "61fd9a7d28ee93750d57136f8322bb72528ae54d", "size": 7841, "ext": "r", "lang": "R", "max_stars_repo_path": "R/MGLReg-mixed.r", "max_stars_repo_name": "lizhengxiao/rMGLReg", "max_stars_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/MGLReg-mixed.r", "max_issues_repo_name": "lizhengxiao/rMGLReg", "max_issues_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2021-08-10T13:04:07.000Z", "max_issues_repo_issues_event_max_datetime": "2021-11-23T11:13:27.000Z", "max_forks_repo_path": "R/MGLReg-mixed.r", "max_forks_repo_name": "lizhengxiao/rMGLReg", "max_forks_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 38.0631067961, "max_line_length": 208, "alphanum_fraction": 0.5593674276, "num_tokens": 2806, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8688267762381844, "lm_q2_score": 0.6150878555160665, "lm_q1q2_score": 0.5344047986112822}} {"text": "\n#' @title Fitting bivariate MGL copula models for mixed data\n#' @description \\code{MGL.mle.mixed} is used to fit bivariate mixed copula regression models via maximum likelihood (ML) method for continuous and semi-continuous variables.\n#' @param U two-dimenstional matrix for pseudo copula data with values in \\eqn{[0,1]} for (F(y1), F(y2)).\n#' @param copula copula 'MGL', 'MGL180', \"MGL-EV\", \"MGL-EV180\", \"MGB2\", \"Normal\" , \"t\".\n#' @param hessian Logical. Should a numerically differentiated Hessian matrix be returned?\n#' @param initpar Initial values for the parameters to be optimized over.\n#' @param U_ two-dimensional matrix for pseudo copula data for the data (F(y1), F(y2-1)).\n#' @param obs two-dimensional matrix for loss observations (y1, y2).\n#' @param f values of the density function for marginal distribution.\n#' @param umin threshold value used in the semi-continuous data.\n#' @param ... additional arguments, see \\code{\\link[stats]{nlm}} for more details.\n#' @importFrom stats nlm\n#' @md\n#' @return A list containing the following components:\n#' * loglike: the value of the estimated maximum of the loglikelihood function.\n#' * copula: the name of the fitted copula. \"MGL180\" and \"MGL-EV180\" denote the survival MGL and MGL-EV copula respectively.\n#' * estimates: the point at which the maximum value of the loglikelihood is obtained.\n#' * se: the standard errors of the estimators.\n#' * AIC, BIC: the goodness fit of the regression models.\n#' * hessian: the hessian at the estimated maximum of the loglikelihood (if requested).\n#'\n#' @details\n#' The estimation method is performed via \\code{\\link[stats]{nlm}} function.\n#' Y1: continuous variable\n#' Y2: semi-continuous variable when Y2>umin, it is continuous and Y2<=umin is discrete.\n#'\n#' For a portfolio of \\eqn{n} observations \\eqn{(y_{i1},y_{i2}; \\; i=1,\\ldots,n)}, the joint density function of \\eqn{(Y_1,Y_2)} can be written as\n#' \\deqn{\n#' \tf_{Y_{1},Y_2}(y_{i1},y_{i2})=\\begin{cases}\n#' \tf_{Y_1}(y_{i1})[\n#' \th_{2|1}(F_{Y_{1}}(y_{i1}),F_{Y_{2}}(y_{i2})) - h_{2|1}(F_{Y_{1}}(y_{i1}),F_{Y_{2}}(y_{i2}-1))\n#' \t], & y_{i2}\\le umin,\\\\\n#' \tf_{Y_1}(y_{i1})f_{Y_2}(y_{i2})c(F_{Y_{1}}(y_{i1}), F_{Y_{2}}(y_{i2})), & y_{i2} > umin,\n#' \t\\end{cases}\n#' }\n#' where the density \\eqn{f_{Y_j}(\\cdot)} and cdf \\eqn{F_{Y_j}(\\cdot)} of the marginal distributions (\\eqn{i=1,2}) are specified respectively. Here\n#' \\eqn{h_{2|1}(u_1, u_2)=\\partial C(u_1,u_2)/\\partial u_1} is the \\eqn{h}-function of bivariate copula.\n#'\n#'\n#' copula:\n#' * \"MGB2\" is multivariate GB2.\n#' * \"Normal\" and \"t\" denote the Gaussian copula and Student-t copula respectively.\n#' * \"MGL\" and \"MGL-EV\" denote the MGL and MGL-EV copula respectively.\n#' * \"MGL180\" and \"MGL-EV180\" denote the survival MGL and survival MGL-EV copula respectively.\n#' * \"Gumbel\" is Gumbel copula.#'\n#'\n#' @examples\n#' library(rMGLReg)\n#' # load the Chinese earthquake data set\n#' u <- cbind(earthqCHI$u1, earthqCHI$u2) # cdf of marginal distribution\n#' u_ <- cbind(earthqCHI$u1, earthqCHI$u2_) # cdf of marginal distribution for Y1 and Y2 - 1\n#' y <- cbind(earthqCHI$y1, earthqCHI$y2) # observations\n#' f <- cbind(earthqCHI$f1, earthqCHI$f2) # pdf of marginal distribution\n#' obs <- y\n#' U <- u\n#' U_ <- u_\n#' umin <- 20\n#' m.MGLMGA180 <- MGL.mle.mixed(obs = y, U = U, U_ = U_,\n#' umin = umin, f = f,\n#' copula = \"MGL180\",\n#' method = \"L-BFGS-B\", initpar = c(2))\n#' m.MGLMGA180\n#' @export\n#'\nMGL.mle.mixed <- function(obs, U, U_, f, copula = c(\n \"MGL\", \"MGL180\", \"MGL-EV\",\n \"MGL-EV180\",\n \"Gumbel\",\n \"Normal\", \"MGB2\", \"t\"\n ), umin,\n hessian = TRUE,\n initpar, ...) {\n dnormcop <- function(U, param) {\n as.numeric(fCopulae::dellipticalCopula(U, rho = param[1], type = \"norm\"))\n } # normal copula\n\n dtcop <- function(U, param) {\n as.numeric(fCopulae::dellipticalCopula(U,\n rho = param[1], type = \"t\",\n param = param[2]\n ))\n } # t copula\n\n dgumcop <- function(U, param) {\n as.numeric(fCopulae::devCopula(U, type = \"gumbel\", param = param[1]))\n } # Bivariate Extreme\n\n\n dMGL <- function(U, param) {\n as.numeric(dcMGL.bivar(u1 = U[, 1], u2 = U[, 2], pars = param[1]))\n }\n\n dMGL180 <- function(U, param) {\n as.numeric(dcMGL180.bivar(u1 = U[, 1], u2 = U[, 2], pars = param[1]))\n }\n\n dMGB2 <- function(U, param) {\n as.numeric(dcMGB2.bivar(u1 = U[, 1], u2 = U[, 2], pars1 = param[1], pars2 = param[2], pars3 = param[3]))\n }\n\n dMGLEV <- function(U, param) {\n as.numeric(dcMGLEV.bivar(u1 = U[, 1], u2 = U[, 2], param = param[1])) # Bivariate Extreme\n }\n\n dMGLEV180 <- function(U, param) {\n as.numeric(dcMGLEV180.bivar(u1 = U[, 1], u2 = U[, 2], param = param[1])) # Bivariate Extreme\n }\n\n hnormcop <- function(U, param) {\n VineCopula::BiCopHfunc(U[, 1], U[, 2], family = 1, par = param[1])\n }\n htcop <- function(U, param) {\n VineCopula::BiCopHfunc(U[, 1], U[, 2], family = 2, par = param[1], par2 = param[2])\n }\n hgumcop <- function(U, param) {\n VineCopula::BiCopHfunc(U[, 1], U[, 2], family = 4, par = param[1])\n }\n hMGL <- function(U, param) {\n hcMGL.bivar(u1 = U[, 1], u2 = U[, 2], pars = param[1])\n }\n hMGL180 <- function(U, param) {\n hfunc1 <- 1 - hcMGL.bivar(u1 = 1 - U[, 1], u2 = 1 - U[, 2], pars = param[1])$hfunc1\n hfunc2 <- 1 - hcMGL.bivar(u1 = 1 - U[, 1], u2 = 1 - U[, 2], pars = param[1])$hfunc2\n out <- list(hfunc1 = hfunc1, hfunc2 = hfunc2)\n out\n }\n hMGLEV180 <- function(U, param) {\n hfunc1 <- hcMGLEV180.bivar(u1 = U[, 1], u2 = U[, 2], param = param[1])$hfunc1\n hfunc2 <- hcMGLEV180.bivar(u1 = U[, 1], u2 = U[, 2], param = param[1])$hfunc2\n out <- list(hfunc1 = hfunc1, hfunc2 = hfunc2)\n out\n }\n hMGLEV <- function(U, param) {\n hfunc1 <- 1 - hcMGLEV180.bivar(u1 = 1 - U[, 1], u2 = 1 - U[, 2], param = param[1])$hfunc1\n hfunc2 <- 1 - hcMGLEV180.bivar(u1 = 1 - U[, 1], u2 = 1 - U[, 2], param = param[1])$hfunc2\n out <- list(hfunc1 = hfunc1, hfunc2 = hfunc2)\n out\n }\n hMGB2 <- function(U, param) {\n hcMGB2.bivar(u1 = U[, 1], u2 = U[, 2], pars1 = param[1], pars2 = param[2], pars3 = param[3])\n }\n\n argnorm <- list(length = 1, lower = 0, upper = 1, name = \"Gaussian\")\n argt <- list(length = 2, lower = c(0, 0), upper = c(1, 100), name = \"Student\")\n arggum <- list(length = 1, lower = 1, upper = 50, name = \"Gumbel\")\n argMGB2 <- list(length = 3, lower = c(0, 0, 0), upper = c(50, 50, 50), name = \"MGB2\")\n argMG180 <- list(length = 1, lower = 0, upper = 10, name = \"MGL180\")\n argMGLEV180 <- list(length = 1, lower = 0, upper = 10, name = \"MGL-EV180\")\n argMG <- list(length = 1, lower = 0, upper = 10, name = \"MGL\")\n argMGLEV <- list(length = 1, lower = 0, upper = 10, name = \"MGL-EV\")\n if (copula == \"MGL\") {\n dcop <- dMGL\n arg.cop <- argMG\n hcop <- hMGL\n } else if (copula == \"MGL180\") {\n dcop <- dMGL180\n arg.cop <- argMG180\n hcop <- hMGL180\n } else if (copula == \"MGL-EV\") {\n dcop <- dMGLEV\n arg.cop <- argMGLEV\n hcop <- hMGLEV\n } else if (copula == \"MGL-EV180\") {\n dcop <- dMGLEV180\n arg.cop <- argMGLEV180\n hcop <- hMGLEV180\n } else if (copula == \"Gumbel\") {\n dcop <- dgumcop\n arg.cop <- arggum\n hcop <- hgumcop\n } else if (copula == \"Normal\") {\n dcop <- dnormcop\n arg.cop <- argnorm\n hcop <- hnormcop\n } else if (copula == \"t\") {\n dcop <- dtcop\n arg.cop <- argt\n hcop <- htcop\n } else if (copula == \"MGB2\") {\n dcop <- dMGB2\n arg.cop <- argMGB2\n hcop <- hMGB2\n }\n\n\n # Obs1 <- obs[, 1] # y1\n Obs2 <- obs[, 2] # y2\n f1 <- f[, 1]\n f2 <- f[, 2]\n copLogL <- function(x) {\n if (all(arg.cop$lower < x && arg.cop$upper > x)) {\n index1 <- which(Obs2 <= umin)\n index2 <- which(Obs2 > umin)\n m1 <- f1[index1] * (hcop(U[index1, ], param = x)$hfunc1 - hcop(U_[index1, ], param = x)$hfunc1)\n logL1 <- (log(m1))\n m2 <- f1[index2] * f2[index2] * dcop(U[index2, ], param = x)\n logL2 <- (log(m2))\n ll <- c(logL1, logL2)\n res <- -sum((ll)) # define the loglikelihood\n } else {\n res <- 100000000000000000\n }\n return(res)\n }\n\n resopt <- nlm(\n f = copLogL,\n p = initpar,\n hessian = hessian\n )\n\n\n # list(\n # loglike = -resopt$minimum,\n # copula = list(name = arg.cop$name),\n # estimates = resopt$estimate,\n # se = sqrt(diag(solve(resopt$hessian))),\n # hessian = -resopt$hessian,\n # AIC = 2 * length(resopt$estimate) + 2 * resopt$minimum,\n # BIC = log(nrow(U)) * length(resopt$estimate) + 2 * resopt$minimum\n # )\n # resopt\n if (hessian == TRUE){\n out <- list(\n loglike = -resopt$minimum,\n copula = list(name = arg.cop$name),\n estimates = resopt$estimate,\n se = sqrt(diag(solve(resopt$hessian))),\n hessian = -resopt$hessian,\n AIC = 2 * length(resopt$estimate) + 2 * resopt$minimum,\n BIC = log(nrow(U)) * length(resopt$estimate) + 2 * resopt$minimum\n )\n } else {\n out <- list(\n loglike = -resopt$minimum,\n copula = list(name = arg.cop$name),\n estimates = resopt$estimate,\n AIC = 2 * length(resopt$estimate) + 2 * resopt$minimum,\n BIC = log(nrow(U)) * length(resopt$estimate) + 2 * resopt$minimum\n )\n }\n out\n}\n", "meta": {"hexsha": "dcf7d888f495b2487e35414d2771ca227a00900c", "size": 9262, "ext": "r", "lang": "R", "max_stars_repo_path": "R/MGL-mle-mixed.r", "max_stars_repo_name": "lizhengxiao/rMGLReg", "max_stars_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/MGL-mle-mixed.r", "max_issues_repo_name": "lizhengxiao/rMGLReg", "max_issues_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2021-08-10T13:04:07.000Z", "max_issues_repo_issues_event_max_datetime": "2021-11-23T11:13:27.000Z", "max_forks_repo_path": "R/MGL-mle-mixed.r", "max_forks_repo_name": "lizhengxiao/rMGLReg", "max_forks_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 37.8040816327, "max_line_length": 173, "alphanum_fraction": 0.5875620816, "num_tokens": 3420, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8289388083214155, "lm_q2_score": 0.6442251201477016, "lm_q1q2_score": 0.5340232033859565}} {"text": "# function that returns basic streak statistics given a sequence\nstreak_stats = function(results_sequence, streak_length){\n n_trials = length(results_sequence)\n n_success = sum(results_sequence)\n \n run_lengths = rle(results_sequence)$lengths\n if_run_success = rle(results_sequence)$values\n n_runs = length(run_lengths)\n \n # runs of successes\n longest_run_success = max(run_lengths*if_run_success)\n n_runs_success = sum(if_run_success)\n \n # successes after streaks of Successes\n run_lengths_adj = run_lengths - streak_length + 1\n run_values_adj_S =\n if_run_success *\n (!((if_run_success == 1) &\n (run_lengths < streak_length)))\n n_after_Sstreak = sum(run_lengths_adj * run_values_adj_S) -\n run_values_adj_S[length(run_values_adj_S)]\n n_success_after_Sstreak = sum((run_lengths - streak_length) * run_values_adj_S)\n \n # runs of failures\n longest_run_failure = max(run_lengths*(1-if_run_success))\n n_runs_failure = sum(1-if_run_success)\n \n # successes after streaks of Failures\n run_values_adj_F =\n (1-if_run_success) *\n (!((if_run_success == 0) &\n (run_lengths < streak_length)))\n n_after_Fstreak = sum(run_lengths_adj * run_values_adj_F) -\n run_values_adj_F[length(run_values_adj_F)]\n n_success_after_Fstreak = n_after_Fstreak -\n sum((run_lengths - streak_length) * run_values_adj_F)\n \n # Proportion of hits on trials after streaks\n phat_after_Sstreak = n_success_after_Sstreak / n_after_Sstreak\n \n # Difference in proportion of hits (trials after streaks - all other trials)\n phat_after_no_Sstreak = (n_success - n_success_after_Sstreak)/\n (n_trials - n_after_Sstreak)\n phat_Sstreak_vs_others = phat_after_Sstreak - phat_after_no_Sstreak\n \n # Difference in proportion of hits (trials after hit streaks - trials after miss streaks)\n phat_after_Fstreak = n_success_after_Fstreak / n_after_Fstreak\n phat_Sstreak_vs_Fstreak = phat_after_Sstreak - phat_after_Fstreak\n \n # Hit streak frequency - proportion of trials that occur after hit streaks\n Sstreak_frequency = n_after_Sstreak / (n_trials - streak_length)\n \n # collect the streak stats for the results sequence\n return(data.frame(phat_after_Sstreak,\n phat_Sstreak_vs_others,\n phat_Sstreak_vs_Fstreak,\n Sstreak_frequency,\n longest_run_success,\n n_runs\n # n_runs_success,\n # phat_after_Fstreak,\n # n_runs_failure,\n # longest_run_failure,\n ))\n}", "meta": {"hexsha": "d5e9cf817de3f45c2bbc80f7b6abf25b9774a768", "size": 2626, "ext": "r", "lang": "R", "max_stars_repo_path": "Hothand/streak_stats.r", "max_stars_repo_name": "townsenddw/shiny1", "max_stars_repo_head_hexsha": "62e2f54120aa4a3b75fc16da09999cec3c361947", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Hothand/streak_stats.r", "max_issues_repo_name": "townsenddw/shiny1", "max_issues_repo_head_hexsha": "62e2f54120aa4a3b75fc16da09999cec3c361947", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Hothand/streak_stats.r", "max_forks_repo_name": "townsenddw/shiny1", "max_forks_repo_head_hexsha": "62e2f54120aa4a3b75fc16da09999cec3c361947", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2019-11-06T12:59:23.000Z", "max_forks_repo_forks_event_max_datetime": "2020-06-21T12:48:05.000Z", "avg_line_length": 40.4, "max_line_length": 91, "alphanum_fraction": 0.6949733435, "num_tokens": 645, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7981867777396212, "lm_q2_score": 0.6688802669716107, "lm_q1q2_score": 0.5338913849876875}} {"text": "### fimd7.R\n### R code from Van Buuren, S. (2012). \n###\t\tFlexible Imputation of Missing Data. \n###\t\tCRC/Chapman & Hall, Boca Raton, FL.\n### (c) 2012 Stef van Buuren, www.multiple-imputation.com\n### Version 1, 22mar2012\n### Version 2, 4nov2015 tested with mice 2.23\n### Tested with Mac OS X 10.7.3, R2.14-2, mice 2.12\n\nif (packageVersion(\"mice\")<'2.12') stop(\"This code requires mice 2.12.\")\n\n### Section 7.1 Too many columns\n\nlibrary(\"mice\")\nlibrary(\"lattice\")\nlibrary(\"foreign\")\nlibrary(\"survival\")\n\n### Section 7.1.3 Data exploration\n\ndata <- leiden85 ## Note: the leiden85 data is not yet avialable in V2.12\nif (!is.data.frame(data)) warning(\"The code for section 7.1/7.2 requires access to the LEIDEN85 data.\")\n\nini <- mice(data, maxit=0) # recommended \ntable(ini$nmis)\n\ntable(data$beroep1, useNA=\"always\")\n\nv1 <- names(ini$nmis[ini$nmis==0])\noutlist1 <- v1[c(1,3:5,7:10,16:47,51:60,62,64:65,69:72)]\nlength(outlist1)\n\n### Section 7.1.4 Outflux\n\nfx <- fluxplot(data, main=NULL, cex=0.9)\n\noutlist2 <- row.names(fx)[fx$outflux<0.5]\nlength(outlist2)\n\ndata2 <- data[,!names(data) %in% outlist2]\nfx2 <- flux(data2)\noutlist3 <- row.names(fx2)[fx2$outflux<0.5]\n\n\n### Section 7.1.5 Logged events\n\nini$log[1:3,]\n\noutlist4 <- as.character(ini$log[,\"out\"])\n\n\n### Section 7.1.6 Quick predictor selection for wide data\n\noutlist <- unique(c(outlist1, outlist2, outlist4))\nlength(outlist)\n\ndata2 <- data[,!names(data) %in% outlist]\n\ninlist <- c(\"sex\",\"lftanam\",\"rrsyst\",\"rrdiast\")\npred <- quickpred(data2, minpuc=0.5, inc=inlist)\n\ntable(rowSums(pred))\n\nrowSums(pred[c(\"rrsyst\",\"rrdiast\"),])\n\nnames(data2)[pred[\"rrsyst\",]==1]\n\nvname <- \"rrsyst\"\ny <- cbind(data2[vname], r=!is.na(data2[,vname]))\nvdata <- data2[,pred[vname,]==1]\nround(cor(y=y,x=vdata,use=\"pair\"),2)\n\n\n### Section 7.1.7 Generating the imputations \n\n### Smart imputation using quickpred - - TIME COMSUMING (30 MINUTES)\nimp.qp <- mice(data2, pred=pred, ridge = 0.0001, seed=29725)\n\n### Blind imputation - MUCH MORE TIME CONSUMING (10 HOURS)\n### Not recommended\nimp <- mice(data, ridge=0.01, seed=32417, maxit=2)\n\nvnames <- c(\"rrsyst\",\"rrdiast\")\ncd1 <- complete(imp)[,vnames]\ncd2 <- complete(imp.qp)[,vnames]\ntyp <- factor(rep(c(\"blind imputation\",\"quickpred\"),each=nrow(cd1)))\nmis <- ici(data2[,vnames])\nmis <- is.na(imp$data$rrsyst)|is.na(imp$data$rrdiast)\ncd <- data.frame(typ=typ,mis=mis,rbind(cd1, cd2))\ntp72 <- xyplot(jitter(rrdiast,10) ~ jitter(rrsyst,10) | typ, data = cd, groups = mis, \n xlab = \"Systolic BP (mmHg)\", ylab = \"Diastolic BP (mmHg)\", \n col=c(mdc(1),mdc(2)), pch=c(1,19),type=c(\"g\",\"p\"), \n strip = strip.custom(bg=\"grey95\"), \n scales = list(alternating=1, tck=c(1,0)))\nprint(tp72)\n\n\n### Section 7.1.8 A further improvements: Survival as predictor variable\n\ndat <- cbind(data2, dead = 1 - data2$dwa)\nhazard <- nelsonaalen(dat, survda, dead)\n\n### calculate correlations between hazard, t and logt (not in \ntmp <- data.frame(hazard, t=data2$survda, logt=log(data2$survda),\n SBP=data2$rrsyst, DBP=data2$rrdiast)\nround(cor(tmp, use=\"pair\"),3)\n\n\n### Section 7.2 Sensitivity analysis\n\n### Figure 7.3\n\nfit <- survfit(Surv(survda/365, 1-dwa) ~ is.na(rrsyst), data = data2) \nplot(fit, lty = 1, lwd=1.5, xlab=\"Years since intake\",\n ylab=\"K-M Survival probability\", las=1, \n col=c(mdc(4),mdc(5)), mark.time=FALSE)\ntext(4,0.7,\"BP measured\")\ntext(2,0.3,\"BP missing\")\n\n\n### Section 7.2.3 Generating imputations under the delta-adjustment\ndelta <- c(0,-5,-10,-15,-20)\npost <- imp.qp$post\n\n### undamped sensitivity analysis\n### TIME CONSUMING (SEVERAL HOURS)\nimp.all.undamped <- vector(\"list\", length(delta))\nfor (i in 1:length(delta)) {\n d <- delta[i]\n cmd <- paste(\"imp[[j]][,i] <- imp[[j]][,i] +\",d)\n post[\"rrsyst\"] <- cmd\n imp <- mice(data2, pred=pred, post=post, maxit=1, seed=i*22, ridge=0.0001)\n imp.all.undamped[[i]] <- imp\n}\n\n\n### damped sensitivity analysis\n### TIME CONSUMING (SEVERAL HOURS)\nimp.all.damped <- vector(\"list\", length(delta))\nfor (i in 1:length(delta)) {\n d <- delta[i]\n cmd <- paste(\"fit <- lm(y ~ as.matrix(x)); \n damp <- sqrt(1 - summary(fit)$r.squared);\n imp[[j]][, i] <- imp[[j]][, i] + damp * \", d)\n post[\"rrsyst\"] <- cmd\n imp <- mice(data2, pred=pred, post=post, maxit=1, seed=i*22, ridge=0.0001)\n imp.all.damped[[i]] <- imp\n}\n\n\n### Section 7.2.4 Complete data analysis\n\ncda <- expression(\n sbpgp <- cut(rrsyst, breaks = c(50, 124, \n 144, 164, 184, 200, 500)),\n agegp <- cut(lftanam, breaks = c(85, 90, \n 95, 110)),\n dead <- 1 - dwa,\n coxph(Surv(survda, dead) \n ~ C(sbpgp, contr.treatment(6, base = 3)) \n + strata(sexe, agegp)))\nimp <- imp.all.damped[[1]]\nfit <- with(imp, cda)\nas.vector(exp(summary(pool(fit))[,1]))\n\n### Table 7.5\n\nfit1 <- with(imp.all.damped[[1]], cda)\nfit2 <- with(imp.all.damped[[2]], cda)\nfit3 <- with(imp.all.damped[[3]], cda)\nfit4 <- with(imp.all.damped[[4]], cda)\nfit5 <- with(imp.all.damped[[5]], cda)\nr1<-as.vector(t(exp(summary(pool(fit1))[,c(1,6:7)])))\nr2<-as.vector(t(exp(summary(pool(fit2))[,c(1,6:7)])))\nr3<-as.vector(t(exp(summary(pool(fit3))[,c(1,6:7)])))\nr4<-as.vector(t(exp(summary(pool(fit4))[,c(1,6:7)])))\nr5<-as.vector(t(exp(summary(pool(fit5))[,c(1,6:7)])))\nround(t(matrix(c(r1,r2,r3,r4,r5),nrow=15)),2)\n\n\n\n### Section 7.3 Correct prevalence estimates from self-reported data\n\nlibrary(\"mice\")\nkrul <- selfreport[selfreport$src==\"krul\",]\nmgg <- selfreport[selfreport$src==\"mgg\",]\n\n### Figure 7.4\n\nxy <- xy.coords(krul$bm, krul$br-krul$bm)\nplot(xy,col=mdc(1),xlab=\"Measured BMI\",ylab=\"Reported - Measured BMI\",\n xlim=c(17,45),ylim=c(-5,5), type=\"n\",lwd=0.7)\npolygon(x=c(30,20,30),y=c(0,10,10),col=\"grey95\",border=NA) \npolygon(x=c(30,40,30),y=c(0,-10,-10),col=\"grey95\",border=NA)\nabline(0,0,lty=2,lwd=0.7)\npoints(xy, col=mdc(1),cex=0.7)\nlines(lowess(xy),lwd=2,col=mdc(4))\ntext(1:4,x=c(40,28,20,32),y=c(4,4,-4,-4),cex=3)\nbox(lwd=1)\n\n### Figure 7.5 Krul data - males\nmales <- krul[krul$sex==\"Male\",]\nfit <- lm(bm~br,data=males)\nplot(x=males$br,y=males$bm,\n xlim=c(26,34),ylim=c(26,34),\n xlab=\"Self-reported BMI\",ylab=\"Measured BMI\",col=mdc(1),\n cex.lab=1.5, cex.axis=1.3)\nabline(h=30,v=30,lty=2)\nabline(coef(fit),col=mdc(4))\nabline(v=(30-coef(fit)[1])/coef(fit)[2],col=mdc(4))\ntext(1:4,x=c(33.8,33.8,26.2,26.2),y=c(33.8,26.2,26.2,33.8),cex=4)\ntext(c(\"a\",\"b\"),x=c(29.1,29.7,29.1,29.7),y=c(34.1,34.1,26.4,26.4),cex=3,adj=c(0.5,1))\n\n\n### Section 7.3.4 Data\n\nmd.pattern(selfreport[,c(\"age\",\"sex\",\"hm\",\"hr\",\"wm\",\"wr\")])\n\n\n### Section 7.3.5 Application\n\nbmi <- function(h,w){return(w/(h/100)^2)}\ninit <- mice(selfreport,maxit=0)\nmeth <- init$meth\nmeth[\"bm\"] <- \"~bmi(hm,wm)\"\nmeth[c(\"prg\", \"edu\", \"etn\")] <- \"\"\npred <- init$pred\npred[,c(\"src\",\"id\",\"pop\",\"prg\",\"edu\",\"etn\",\"web\",\"bm\",\"br\")] <- 0\nimp <- mice(selfreport, pred=pred, meth=meth, seed=66573, maxit=20, m=10)\n\n\n### Figure 7.6 \n\ncd <- complete(imp, 1)\nxy <- xy.coords(cd$bm, cd$br-cd$bm)\nplot(xy,col=mdc(2),xlab=\"Measured BMI\",ylab=\"Reported - Measured BMI\",\n xlim=c(17,45),ylim=c(-5,5), type=\"n\",lwd=0.7)\npolygon(x=c(30,20,30),y=c(0,10,10),col=\"grey95\",border=NA) \npolygon(x=c(30,40,30),y=c(0,-10,-10),col=\"grey95\",border=NA)\nabline(0,0,lty=2,lwd=0.7)\n\nidx <- cd$src==\"krul\"\nxyc <- xy; xyc$x <- xy$x[idx]; xyc$y <- xy$y[idx]\nxys <- xy; xys$x <- xy$x[!idx]; xys$y <- xy$y[!idx]\npoints(xyc,col=mdc(1), cex=0.7)\npoints(xys,col=mdc(2), cex=0.7)\nlines(lowess(xyc),col=mdc(4),lwd=2)\nlines(lowess(xys),col=mdc(5),lwd=2)\ntext(1:4,x=c(40,28,20,32),y=c(4,4,-4,-4),cex=3)\nbox(lwd=1)\n\n\n### Table 7.7\n### Not particularly elegant code, at the moment\n prev <- matrix(NA,nr=15,nc=4)\n prev[1,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\")))))[1,1:2]\n prev[1,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\")))))[1,1:2]\n prev[2,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Male\")))))[1,1:2]\n prev[2,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Male\")))))[1,1:2]\n prev[3,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Female\")))))[1,1:2]\n prev[3,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Female\")))))[1,1:2]\n prev[4,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>=18&age<30)))))[1,1:2]\n prev[4,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>18&age<30)))))[1,1:2]\n prev[5,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>=30&age<40)))))[1,1:2]\n prev[5,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>30&age<40)))))[1,1:2]\n prev[6,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>=40&age<50)))))[1,1:2]\n prev[6,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>40&age<50)))))[1,1:2]\n prev[7,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>=50&age<60)))))[1,1:2]\n prev[7,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>50&age<60)))))[1,1:2]\n prev[8,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>=60&age<80)))))[1,1:2]\n prev[8,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Male\"&age>60&age<80)))))[1,1:2]\n prev[10,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>=18&age<30)))))[1,1:2]\n prev[10,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>18&age<30)))))[1,1:2]\n prev[11,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>=30&age<40)))))[1,1:2]\n prev[11,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>30&age<40)))))[1,1:2]\n prev[12,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>=40&age<50)))))[1,1:2]\n prev[12,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>40&age<50)))))[1,1:2]\n prev[13,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>=50&age<60)))))[1,1:2]\n prev[13,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>50&age<60)))))[1,1:2]\n prev[14,1:2] <- summary(pool(with(imp, lm(br>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>=60&age<80)))))[1,1:2]\n prev[14,3:4] <- summary(pool(with(imp, lm(bm>=30~1,subset=(src==\"mgg\"&sex==\"Female\"&age>60&age<80)))))[1,1:2]\n \n### find group sizes\ntable(mgg$sex,mgg$web)\ntable(mgg$sex,mgg$web,mgg$age>55)\n\n\n### Section 7.4 Enhancing comparability\n\nlibrary(\"mice\")\n\n### Section 7.4.3 Independence: Imputation without a bridge study\n\nfA <- c(242, 43, 15, 0, 6)\nfB <- c(145, 110, 29, 8)\nYA <- rep(ordered(c(0:3,NA)), fA)\nYB <- rep(ordered(c(0:3)), fB)\nY <- rbind(data.frame(YA,YB=ordered(NA)),\n data.frame(YB,YA=ordered(NA)))\n\nmd.pattern(Y)\n\n### simulation function\n### Warning: micemill() overwrites objects 'imp' and 'tau'\n### in the global enviroment\n\nmicemill <- function(n) {\n for (i in 1:n) {\n imp <<- mice.mids(imp) # global assignment \n cors <- with(imp, cor(as.numeric(YA),\n as.numeric(YB),\n method=\"kendall\"))\n tau <<- rbind(tau, getfit(cors, s=TRUE)) # global assignment\n }\n}\n\ntau <- NULL\nimp <- mice(Y, max=0, m=10, seed=32662)\nmicemill(50)\n\n### Figure 7.7\nplotit <- function()\n matplot(x=1:nrow(tau),y=tau,\n ylab=expression(paste(\"Kendall's \",tau)), \n xlab=\"Iteration\", type=\"l\", lwd=1,\n lty=1:10,col=\"black\")\nplotit()\n\n### Section 7.4.4 Fully dependent or independent?\n### Use the 'walking' dataset of the mice package\n\n## split walking: ab is sources A and B (598 records)\nab <- walking[walking$src==\"A\"|walking$src==\"B\",]\n\n## split walking: euridiss is external source (292 records)\nexternal <- walking[walking$src==\"E\",]\n\n### create contingency tables: YA by YB (in ab)\nftable(addmargins(table(ab[,c(\"YA\",\"YB\")],useNA=\"ifany\")))\n\n### Table 7.8 contingency table YA by YB (in euridiss)\nftable(addmargins(table(external[,c(\"YA\",\"YB\")],useNA=\"ifany\")))\n\n### Section 7.4.5 Imputation using a bridge study\nmd.pattern(walking)\n\n### Figure 7.8\ntau <- NULL\nimp <- mice(walking, max=0, m=10, seed=92786)\npred <- imp$pred\npred[,c(\"src\",\"age\",\"sex\")] <- 0\nimp <- mice(walking, max=0, m=10, seed=92786, pred=pred)\nmicemill(20)\nplotit()\n\n\n### impute without and with covariates\n\nthetaAB <- NULL\nprops <- with(imp, mean(YB[src==\"A\"]=='0'))\nthetaAB <<- rbind(thetaAB, getfit(props, s=TRUE)) \n\nmicemill <- function(n) {\n for (i in 1:n) {\n imp <<- mice.mids(imp) # global assignment \n cors <- with(imp, cor(as.numeric(YA[src==\"A\"]),\n as.numeric(YB[src==\"A\"]),\n method=\"kendall\"))\n tau <<- rbind(tau, getfit(cors, s=TRUE)) # global assignment\n means <- with(imp, mean(as.numeric(YA[src==\"A\"]), na.rm=TRUE))\n thetaBA <<- rbind(thetaBA, getfit(means, s=TRUE)-1)\n props <- with(imp, mean(YB[src==\"A\"]=='0'))\n thetaAB <<- rbind(thetaAB, getfit(props, s=TRUE)) \n tabs <- with(imp, ftable(addmargins(\n table(YA[src==\"A\"],YB[src==\"A\"],\n useNA=\"ifany\", dnn=c(\"YA\",\"YB\")))))\n print(getfit(tabs)[[2]])\n }\n}\n\n\ntau <- NULL\nthetaBA <- NULL\nthetaAB <- NULL\nimp <- mice(walking, max=0, m=10, seed=99786)\noldpred <- pred <- imp$pred\npred[,c(\"src\",\"age\",\"sex\")] <- 0\nimp <- mice(walking, max=0, m=10, seed=99786, pred=pred)\nmicemill(20)\npred <- oldpred\npred[,c(\"src\")] <- 0\nimp <- mice(walking, max=0, m=10, pred=pred)\nmicemill(20)\n\n### Figure 7.9\n\nmatplot(x=1:nrow(thetaAB),y=thetaAB,xlab=\"Iteration\",\n ylab = expression(hat(theta)[AB]),\n type=\"l\", lwd=1, lty=1:10,col=\"black\",\n ylim=c(0.4,0.7))\nabline(h=0.497, v=20, lty=2)\ntext(x=5,y=0.488,expression(hat(theta)[BB]),adj=0)\ntext(x=5,y=0.430,\"Without covariates\", adj=0)\ntext(x=25,y=0.430,\"With covariates\", adj=0)\narrows(x0=4.5,y0=0.488,x1=0,y1=0.495,length=0.1, angle=20)\npoints(x=-0.4,y=0.497,pch=20)\n\n## file <- file.path(\"~/Documents/Sync/Impute/mice/V2.12/mice/data/walking.rda\")\n## save(walking, file=file)\n\n## selfreport <- nl\n## file <- file.path(\"~/Documents/Sync/Impute/mice/V2.12/mice/data/selfreport.rda\")\n## save(selfreport, file=file)\n\n", "meta": {"hexsha": "3dc4d76f81b12b21b5cb7f13426dde4dde8cdc84", "size": 14205, "ext": "r", "lang": "R", "max_stars_repo_path": "packrat/lib/x86_64-pc-linux-gnu/3.2.5/mice/doc/fimd7.r", "max_stars_repo_name": "Chicago-R-User-Group/2017-n3-Meetup-RStudio", "max_stars_repo_head_hexsha": "71a3204412c7573af2d233208147780d313430af", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "packrat/lib/x86_64-pc-linux-gnu/3.2.5/mice/doc/fimd7.r", "max_issues_repo_name": "Chicago-R-User-Group/2017-n3-Meetup-RStudio", "max_issues_repo_head_hexsha": "71a3204412c7573af2d233208147780d313430af", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2021-11-12T14:06:52.000Z", "max_issues_repo_issues_event_max_datetime": "2021-12-10T23:26:27.000Z", "max_forks_repo_path": "packrat/lib/x86_64-pc-linux-gnu/3.2.5/mice/doc/fimd7.r", "max_forks_repo_name": "Chicago-R-User-Group/2017-n3-Meetup-RStudio", "max_forks_repo_head_hexsha": "71a3204412c7573af2d233208147780d313430af", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 34.2289156627, "max_line_length": 112, "alphanum_fraction": 0.6133051742, "num_tokens": 5494, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104788995148792, "lm_q2_score": 0.658417500561683, "lm_q1q2_score": 0.5336334912765702}} {"text": "### calculate the sample size needed for achieving a certain power in three ways: TSLS, AR, ARsens\r\n### some parameter values are inferred from the ivmodel\r\n\r\nIVsize=function(ivmodel, power, alpha=0.05, beta=NULL, type=\"TSLS\", deltarange=NULL, delta=NULL){\r\n esti = para(ivmodel)\r\n if(is.null(beta)) \r\n beta = esti$beta\r\n if(is.null(deltarange)) \r\n deltarange = ivmodel$deltarange\r\n\r\n if(type==\"TSLS\"){\r\n return(TSLS.size(power, beta, cor(ivmodel$Zadj, ivmodel$Dadj), esti$sigmau, var(ivmodel$Dadj), alpha))\r\n }\r\n if(type==\"AR\"){\r\n return(AR.size(power, ivmodel$p, ivmodel$L, beta, esti$gamma, var(ivmodel$Zadj), esti$sigmau, esti$sigmav, esti$rho, alpha)) \r\n }\r\n\r\n if(type==\"ARsens\"){\r\n return(ARsens.size(power, ivmodel$p, beta, esti$gamma, var(ivmodel$Zadj), esti$sigmau, esti$sigmav, esti$rho, alpha, deltarange, delta)) \r\n }\r\n print(\"Input error.\")\r\n return(NULL)\t\t \r\n}\r\n", "meta": {"hexsha": "d57bd243a6f36ad122b9ddf4bd9243c1769ac1af", "size": 906, "ext": "r", "lang": "R", "max_stars_repo_path": "R/ivsize.r", "max_stars_repo_name": "qingyuanzhao/ivmodel", "max_stars_repo_head_hexsha": "ad59c7cbd32734b25b12c9aef3deb8ba1559cacb", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 8, "max_stars_repo_stars_event_min_datetime": "2018-09-22T13:38:52.000Z", "max_stars_repo_stars_event_max_datetime": "2021-06-23T02:04:23.000Z", "max_issues_repo_path": "R/ivsize.r", "max_issues_repo_name": "qingyuanzhao/ivmodel", "max_issues_repo_head_hexsha": "ad59c7cbd32734b25b12c9aef3deb8ba1559cacb", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2020-11-09T19:20:30.000Z", "max_issues_repo_issues_event_max_datetime": "2022-02-01T13:44:24.000Z", "max_forks_repo_path": "R/ivsize.r", "max_forks_repo_name": "qingyuanzhao/ivmodel", "max_forks_repo_head_hexsha": "ad59c7cbd32734b25b12c9aef3deb8ba1559cacb", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-06-01T16:33:38.000Z", "max_forks_repo_forks_event_max_datetime": "2020-06-01T16:33:38.000Z", "avg_line_length": 37.75, "max_line_length": 143, "alphanum_fraction": 0.6666666667, "num_tokens": 290, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8519528019683105, "lm_q2_score": 0.626124191181315, "lm_q1q2_score": 0.5334282590570635}} {"text": "#install.packages(\"ggplot2\")\nrequire(\"ggplot2\")\nlibrary(\"markovchain\")\n\nsetwd(\"C:/Code/climData/glaser2019\")\n\ndfHhi <- read.csv(\"csv/hhi_1500_2019.csv\", sep=\",\", na = \"NA\")\n\npic <- dfHhi$hhi\nlen <- length(pic)\n## mirror pic\npic <- append(pic, pic)\nfor(i in 1:len) {\n pic[i+len] <- pic[1+len-i]\n}\n\nfrq <- fft(pic, inverse = FALSE)\n\nfrq1 <- frq\nfilterYears = 0.1 #filter 1m\nstart = round(len/(12.0*filterYears))\nstop = round(2.0*len-start)\nfrq1[start:stop] <- 0.0 \npic1 <- Re(fft(frq1, inverse = TRUE)/length(frq1))\npic1 <- pic1[1:len]\ndfHhi$hhi1 <- pic1\n\nfrq5 <- frq\nfilterYears = 5 #filter 5y\nstart = round(len/(12.0*filterYears))\nstop = round(2.0*len-start)\nfrq5[start:stop] <- 0.0 \npic5 <- Re(fft(frq5, inverse = TRUE)/length(frq5))\npic5 <- pic5[1:len]\ndfHhi$hhi5 <- pic5\n\n\ndfHhi$rhi <- dfHhi$hhi\ndfHhi$rmhi <- dfHhi$hhi\nfor(i in 1:length(dfHhi$rhi)) {\n dfHhi$rhi[i] = 'd4'\n if (dfHhi$hhi1[i] > -3.0) {dfHhi$rhi[i] = 'd3'}\n if (dfHhi$hhi1[i] > -2.0) {dfHhi$rhi[i] = 'd2'}\n if (dfHhi$hhi1[i] > -1.5) {dfHhi$rhi[i] = 'd1'}\n if (dfHhi$hhi1[i] > -1.0) {dfHhi$rhi[i] = 'n0'}\n if (dfHhi$hhi1[i] > 1.0) {dfHhi$rhi[i] = 'w1'}\n if (dfHhi$hhi1[i] > 1.5) {dfHhi$rhi[i] = 'w2'}\n if (dfHhi$hhi1[i] > 2.0) {dfHhi$rhi[i] = 'w3'} \n if (dfHhi$hhi1[i] > 3.0) {dfHhi$rhi[i] = 'w4'}\n dfHhi$rmhi[i] <- paste('m',dfHhi$month[i],':',dfHhi$rhi[i], sep='')\n}\ndfHhi$hstate <- dfHhi$hhi\ndfHhi$hstate[1] <- paste('n0','-',dfHhi$rhi[1], sep='')\ndfHhi$mhstate <- dfHhi$hhi\ndfHhi$mhstate[1] <- paste('m',dfHhi$month[12],':','n0','-','m',dfHhi$month[1],':',dfHhi$rhi[1], sep='')\nfor(i in 2:length(dfHhi$hstate)) {\n dfHhi$hstate[i] <- paste(dfHhi$rhi[i-1],'-',dfHhi$rhi[i], sep='')\n dfHhi$mhstate[i] <- paste('m',dfHhi$month[i-1],':',dfHhi$rhi[i-1],'-','m',dfHhi$month[i],':',dfHhi$rhi[i], sep='')\n}\nmonthStates <- c(\"m1\",\"m2\",\"m3\",\"m4\",\"m5\",\"m6\",\"m7\",\"m8\",\"m9\",\"m10\",\"m11\",\"m12\")\nhumidityStates <- c(\"w4\", \"w3\", \"w2\", \"w1\", \"n0\", \"d1\", \"d2\", \"d3\", \"d4\")\nbyRow <- TRUE\nhumidityMatrix <- matrix(rep( 0.0, len=length(humidityStates)^2), nrow=length(humidityStates))\nfor(row in 1:length(humidityStates)) {\n sumRow <- 0.0\n for(col in 1:length(humidityStates)) {\n trans <- paste(humidityStates[row],'-',humidityStates[col], sep='')\n count <- length(subset(dfHhi, dfHhi$hstate==trans)$hstate)\n humidityMatrix[row,col] <- count\n sumRow <- sumRow + count\n }\n if(sumRow > 0.0) {\n for(col in 1:length(humidityStates)) {\n humidityMatrix[row,col] <- humidityMatrix[row,col] / sumRow\n } \n }\n}\n\nmcHumidity <- new(\"markovchain\", states = humidityStates, byrow = byRow,\n transitionMatrix = humidityMatrix, name = \"Humidity\")\n\ninitialState <- c(0, 0, 0, 0, 1, 0, 0, 0, 0)\nafter1000month <- initialState * (mcHumidity ^ 1000)\ntransitionProbability(mcHumidity, \"w4\", \"w3\")\n\nprint(mcHumidity)\nshow(mcHumidity)\nsteadyStates(mcHumidity) #endless time\n\nsimHumidity <- rmarkovchain(n = 12*10, object = mcHumidity, t0 = \"n0\")\n\nsequence <- c(\"a\", \"b\", \"a\", \"a\", \"a\", \"a\", \"b\", \"a\", \"b\", \"a\", \"b\", \"a\", \"a\", \n \"b\", \"b\", \"b\", \"a\") \nsequenceMatr <- createSequenceMatrix(dfHhi$rhi, sanitize = FALSE)\n\nmcFitMLE <- markovchainFit(data = dfHhi$rhi)\nmcFitBSP <- markovchainFit(data = dfHhi$rhi, method = \"bootstrap\", nboot = 5, name = \"Bootstrap Mc\")\n\nsequenceMatr <- createSequenceMatrix(dfHhi$rmhi, sanitize = FALSE)\nmcFitMLE <- markovchainFit(data = dfHhi$rmhi)\nmcFitBSP <- markovchainFit(data = dfHhi$rmhi, method = \"bootstrap\", nboot = 5, name = \"Bootstrap Mc\")\n\nsteadyStates(mcFitBSP$estimate)\nsimHumidity <- rmarkovchain(n = 12*10, object = mcFitBSP$estimate, t0 = \"m1:n0\")\n\nverifyMarkovProperty(dfHhi$rmhi)\nverifyEmpiricalToTheoretical(data=dfHhi$rhi,object=mcFitBSP$estimate)\nverifyHomogeneity(inputList=list(dfHhi$rmhi[3601:6000], dfHhi$rmhi[1:2400]),verbose=TRUE)\n\n## create monthly\nmcFitBSP <- markovchainFit(data = dfHhi$rmhi, method = \"bootstrap\", nboot = 5, name = \"Bootstrap Mc\")\n\nfrom = \"d2\"\nto = \"d1\"\n\nprops <- monthStates\nmon <- monthStates \nfor(i in 1:length(monthStates)) {\n start = paste(monthStates[i],':',from, sep='')\n stop = paste(monthStates[(i+1)%%length(monthStates)],':',to, sep='') \n props[i] = transitionProbability(mcFitBSP$estimate, start, stop)\n mon[i] <- 0.01*i\n}\n\n\nmp <- ggplot()\nmp + \n theme_classic() +\n geom_point(aes(y=props, x=mon)) \n \n## 100 years\n\nfrom = \"w1\"\nto = \"w2\"\n\n\nresult <- data.frame(years = c(1:9))\nmp <- ggplot(result) + theme_classic() +\n scale_y_continuous(limits=c(0,1))+\n scale_x_continuous(limits=c(1500,2020)) \n\n\nfor (from in humidityStates) {\n for (to in humidityStates) {\n result$prop <- c(1:9) \nfor (i in 1:9) {\n yS = 1450 + 50*i\n yE = 1550 + 50*i\n result$years[i] <- 1500 + 50*i\n seq <- subset(dfHhi, dfHhi$year>yS & dfHhi$yearyS & dfHhi$yearyS & dfHhi$year@@dfo-mpo.gc.ca}\n#' @export\n#\n\nFB2C <- function(F,B) {\n\tC = B*(exp(F)-1)*exp(-F)\n\treturn(C)\n}", "meta": {"hexsha": "b4578dcde5a6b84c83d1d37d83836e99aa357899", "size": 260, "ext": "r", "lang": "R", "max_stars_repo_path": "R/FB2C.r", "max_stars_repo_name": "AtlanticR/bio.utilities", "max_stars_repo_head_hexsha": "aaa52cf86afa4ee9e6f46c4516a48d27cc0bfed9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/FB2C.r", "max_issues_repo_name": "AtlanticR/bio.utilities", "max_issues_repo_head_hexsha": "aaa52cf86afa4ee9e6f46c4516a48d27cc0bfed9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/FB2C.r", "max_forks_repo_name": "AtlanticR/bio.utilities", "max_forks_repo_head_hexsha": "aaa52cf86afa4ee9e6f46c4516a48d27cc0bfed9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 23.6363636364, "max_line_length": 82, "alphanum_fraction": 0.6692307692, "num_tokens": 84, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.8615382236515259, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5331029957179972}} {"text": "#' cond.pr\n#'\n#' Estimate conditional probability for each observation\n#' Default to use local method over knn method,\n#' if both arguement supply or unspecify, only the local method will use.\n#'\n#' @param x vector or matrix whose rows contains predictors\n#' of each observations.\n#' @param y vector of the variable we want to estimate its condtional cdf\n#' @param span fraction of total distance to be the bandwidth, between 0 and 1,\n#' default to be 0.5 if bandwidth not supply.\n#' @param bandwidth the bandwidth of x to evaluate the cdf,\n#' overwrite the span input.\n#' @param rnn fraction of total number of observations to be knn,\n#' between 0 and 1. Set span to be NULL to use knn,\n#' default to be 0.5, if not supply\n#' @param knn the number of nearest neighbor to estimate the cdf,\n#' should be less than the number of observation and greater than 0\n#' (\\code{knn < length(y)}.\n#' setting this value overwrite the rnn\n#'\n#' @return vector of conditional probability \\code{p(Y < y | x)}\n#'\n#' @examples\n#' cond_pr(1:5, 1:5) #[1] 0.33 0.5 0.6 0.75 1\n#' cond_pr(1:5, 5:1, span=0) #[1] 1 1 1 1 1\n#' cond_pr(1:5, 5:1, span=1) #[1] 1.0 0.8 0.6 0.4 0.2\n#' cond_pr(1:5, 5:1, span=NULL) [1] 1. 0.66 0.66 0.66 0.33\n#' cond_pr(1:5, 5:1, span=NULL, rnn=1) [1] 1.0 0.8 0.6 0.4 0.2\n#' cond_pr(1:5, 5:1, span=NULL, rnn=0) [1] 1 1 1 1 1\n#' cond_pr(1:5, 5:1, span=NULL, rnn=0, knn=1) [1] 1. 0.6 0.66 0.66 0.5\n#' cond_pr(1:5, 5:1, span=NULL, rnn=0, knn=3) [1] 1.00 0.75 0.60 0.50 0.25\n#' cond_pr(1:5, 5:1, bandwidth=1) [1] 1 0.6 0.6 0.6 0.5\n#' cond_pr(matrix(1:12, nrow=6), 6:1) [1] 1. 0.75 0.6 0.6 0.5 0.33\n#' cond_pr(matrix(1:12, nrow=6), 6:1, span = 0.6) [1] 1.0.8 0.67 0.5 0.4 0.25\n#' cond_pr(matrix(1:12, nrow=6), 6:1, bandwidth = 5) 1. 0.8 0.6 0.5 0.4 0.25\n#' cond_pr(matrix(1:12, nrow=6), 6:1, span = NULL) [1] 1. 0.75 0.6 0.6 0.5 0.25\n#' cond_pr(matrix(1:12, nrow=6), 6:1, span = NULL, rnn = 0)\n#' cond_pr(matrix(1:12, nrow=6), 6:1, span = NULL, rnn = 1)\n#' cond_pr(matrix(1:12, nrow=6), 6:1, span = NULL, rnn = 1, knn=1)\n\ncond_pr <- function(x, y,\n span = 0.1, bandwidth=NULL,\n rnn = NULL, knn=NULL) {\n #--------------------------------------\n ### Checking---------------------------\n #--------------------------------------\n if (is.vector(x)) {\n if ((s_n <- length(x)) != length(y)) {\n stop(\"input error: length(x)!=length(y)!\")\n }\n is_vec <- T\n } else {\n if ((s_n <- nrow(x)) != length(y)) {\n stop(\"input error: nrow(x)!=length(y)!\")\n }\n is_vec <- F\n }\n\n use_local <- F\n if (!is.null(span) || !is.null(bandwidth)) {\n if ((span < 0 || span > 1) & missing(bandwidth)) {\n stop(\"input error: span out of bound, no bandwidth supply!\")\n }\n use_local <- T\n } else {\n if (missing(knn) || is.null(knn)) {\n if (is.null(rnn)) {\n rnn <- 0.5\n } else if (rnn < 0 || rnn > 1) {\n stop(\"input error: rnn out of bound, no knn supply!\")\n }\n knn <- floor(rnn * s_n)\n }\n\n if (knn >= s_n) {\n print(\"warning: knn > s_n, set knn = s_n.\")\n return(order(y) / s_n)\n }\n }\n\n #---------------------------------------------------\n ### Calculate probability---------------------------\n #---------------------------------------------------\n if (use_local) {\n output <- local_method(s_n, x, y, span, bandwidth, is_vec)\n } else {\n output <- knn_method(s_n, x, y, knn, is_vec)\n }\n\n return(output)\n}\n\n\n\nlocal_method <- function(s_n, x, y, span, bandwidth, is_vec) {\n # preallocate vector for condition probability\n c_prob <- c()\n if (is_vec) {\n s_range <- max(x) - min(x)\n if (missing(bandwidth) || is.null(bandwidth)) {\n bandwidth <- span * s_range\n }\n if (bandwidth >= s_range) {\n print(\"warning: bandwidth > maxwidth, bandwidth = maxwidth.\")\n return(order(y) / s_n)\n }\n # Estimate conditional probability\n for (i in seq_len(s_n)) {\n idx <- which(abs(x - x[i]) <= bandwidth)\n c_prob[i] <- mean(y[idx] <= y[i])\n }\n } else {\n # Preallocate matrix for distance\n m_dist <- matrix(0, nrow = s_n, ncol = s_n)\n # Estimate distance between each x_i\n for (i in seq_len(s_n)) {\n diff <- x[- (1:i), ] - tcrossprod(rep(1, s_n - i), x[i, ])\n m_dist[i, - (1:i)] <- rowSums(diff^2)\n }\n m_dist <- sqrt(m_dist)\n s_range <- max(m_dist) - min(m_dist)\n if (missing(bandwidth) || is.null(bandwidth)) {\n bandwidth <- span * s_range\n }\n if (bandwidth >= s_range) {\n print(\"warning: bandwidth > maxwidth, bandwidth = maxwidth.\")\n return(order(y) / s_n)\n }\n m_dist <- m_dist + t(m_dist)\n for (i in seq_len(s_n)) {\n idx <- which(m_dist[i, ] <= bandwidth)\n c_prob[i] <- mean(y[idx] <= y[i])\n }\n }\n\n return(c_prob)\n}\n\n\nknn_method <- function(s_n, x, y, knn, is_vec) {\n # preallocate vector for condition probability\n c_prob <- c()\n if (is_vec) {\n ord <- order(x)\n # Estimate conditional probability\n for (i in seq_len(s_n)) {\n ord_i <- which(ord == i)\n l_idx <- max(1, ord_i - knn)\n u_idx <- min(s_n, ord_i + knn)\n set <- ord[l_idx:u_idx]\n v_dist <- abs(x[set] - x[i])\n # find the closest k elements\n kth <- sort(v_dist, partial = knn + 1)[knn + 1]\n idx <- which(v_dist <= kth)\n c_prob[i] <- mean(y[set[idx]] <= y[i])\n }\n } else {\n # Estimate distance between each x_i\n for (i in seq_len(s_n)) {\n diff <- abs(x - tcrossprod(rep(1, s_n), x[i, ]))\n v_dist <- rowSums(diff^2)\n\n kth <- sort(v_dist, partial = knn + 1)[knn + 1]\n idx <- which(v_dist <= kth)\n c_prob[i] <- mean(y[idx] <= y[i])\n }\n }\n\n return(c_prob)\n}", "meta": {"hexsha": "b31c6361098b8cc2883d578501e36526e65b945c", "size": 6204, "ext": "r", "lang": "R", "max_stars_repo_path": "R/conditinal_prob.r", "max_stars_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_stars_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/conditinal_prob.r", "max_issues_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_issues_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/conditinal_prob.r", "max_forks_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_forks_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.2807017544, "max_line_length": 80, "alphanum_fraction": 0.499032882, "num_tokens": 1992, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7549149978955811, "lm_q2_score": 0.7057850216484838, "lm_q1q2_score": 0.5328076981324978}} {"text": "##### Chapter 8: Association Rules -------------------\r\n\r\n## Example: Identifying Frequently-Purchased Groceries ----\r\n## Step 2: Exploring and preparing the data ----\r\n\r\n# load the grocery data into a sparse matrix\r\nlibrary(arules)\r\ngroceries <- read.transactions(\"groceries.csv\", sep = \",\")\r\nsummary(groceries)\r\n\r\n# look at the first five transactions\r\ninspect(groceries[1:5])\r\n\r\n# examine the frequency of items\r\nitemFrequency(groceries[, 1:3])\r\n\r\n# plot the frequency of items\r\nitemFrequencyPlot(groceries, support = 0.1)\r\nitemFrequencyPlot(groceries, topN = 20)\r\n\r\n# a visualization of the sparse matrix for the first five transactions\r\nimage(groceries[1:5])\r\n\r\n# visualization of a random sample of 100 transactions\r\nimage(sample(groceries, 100))\r\n\r\n## Step 3: Training a model on the data ----\r\nlibrary(arules)\r\n\r\n# default settings result in zero rules learned\r\napriori(groceries)\r\n\r\n# set better support and confidence levels to learn more rules\r\ngroceryrules <- apriori(groceries, parameter = list(support =\r\n 0.006, confidence = 0.25, minlen = 2))\r\ngroceryrules\r\n\r\n## Step 4: Evaluating model performance ----\r\n# summary of grocery association rules\r\nsummary(groceryrules)\r\n\r\n# look at the first three rules\r\ninspect(groceryrules[1:3])\r\n\r\n## Step 5: Improving model performance ----\r\n\r\n# sorting grocery rules by lift\r\ninspect(sort(groceryrules, by = \"lift\")[1:5])\r\n\r\n# finding subsets of rules containing any berry items\r\nberryrules <- subset(groceryrules, items %in% \"berries\")\r\ninspect(berryrules)\r\n\r\n# writing the rules to a CSV file\r\nwrite(groceryrules, file = \"groceryrules.csv\",\r\n sep = \",\", quote = TRUE, row.names = FALSE)\r\n\r\n# converting the rule set to a data frame\r\ngroceryrules_df <- as(groceryrules, \"data.frame\")\r\nstr(groceryrules_df)\r\n", "meta": {"hexsha": "0dfed40ea175f5c18b97f47f9b524747c45f78f5", "size": 1792, "ext": "r", "lang": "R", "max_stars_repo_path": "Chapter08/MLwR_3rdEd_08.r", "max_stars_repo_name": "takeuchi-kou/Machine-Learning-with-R-Third-Edition", "max_stars_repo_head_hexsha": "c4e492b4bfed3d9240d645154001616a925ef773", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 39, "max_stars_repo_stars_event_min_datetime": "2019-06-13T02:39:49.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-25T06:15:12.000Z", "max_issues_repo_path": "Chapter08/MLwR_3rdEd_08.r", "max_issues_repo_name": "derekwietelman/Machine-Learning-with-R-Third-Edition", "max_issues_repo_head_hexsha": "740ba3086fa0a8f6c669e888e0dac6128acc5c08", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2020-06-28T20:03:23.000Z", "max_issues_repo_issues_event_max_datetime": "2020-12-01T17:08:12.000Z", "max_forks_repo_path": "Chapter08/MLwR_3rdEd_08.r", "max_forks_repo_name": "derekwietelman/Machine-Learning-with-R-Third-Edition", "max_forks_repo_head_hexsha": "740ba3086fa0a8f6c669e888e0dac6128acc5c08", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 76, "max_forks_repo_forks_event_min_datetime": "2019-04-02T08:33:08.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-18T12:38:48.000Z", "avg_line_length": 29.3770491803, "max_line_length": 71, "alphanum_fraction": 0.7020089286, "num_tokens": 439, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7690802264851919, "lm_q2_score": 0.6926419767901476, "lm_q1q2_score": 0.5326972483829178}} {"text": "#' GEE Beta Equations\n#'\n#' Creates GEE Equations for beta coefficients.\n#'\n#' @param betas vector of coefficients\n#' @param rho overdispersion parameter\n#' @param dat matrix of multinomial clusters\n#'\n#' @return estimating equations for betas\ncreate_beta_equations <- function (betas, rho, dat) {\n n <- dim(dat)[[2]] # dimensions of multinom\n m <- n - 1 # order of multinom\n ni <- rowSums(dat) # cluster sizes\n num.clus <- dim(dat)[[1]]\n\n # helper vars\n betas.exp <- exp(betas)\n rho.sq <- rho ^ 2\n denom <- 1 + sum(betas.exp)\n dispi <- (ni * (1 + (ni - 1) * rho.sq))\n p.vec <- as.vector(betas.exp / denom)\n p.mat <- matrix(p.vec, num.clus, m, byrow = TRUE)\n\n # create matrices\n d.mat <- (diag( betas.exp * denom, m ) - betas.exp %*% t( betas.exp )) / (denom ^ 2)\n var.mat <- diag( p.vec, m ) - p.vec %*% t( p.vec )\n r.vec <- dat[, 1:m] - ni * p.mat\n\n # result\n d.mat %*% solve(var.mat) %*% rowSums( t((ni / dispi) * r.vec) )\n}\n\n#' GEE Rho Equations\n#'\n#' Creates GEE Equations for rho (overdispersion parameter).\n#'\n#' @param rho overdispersion parameter\n#' @param betas vector of coefficients\n#' @param dat matrix of multinomial clusters\n#'\n#' @return estimating equations for rho\ncreate_rho_equations <- function (rho, betas, dat) {\n n <- dim(dat)[[2]] # dimensions of multinom\n m <- n - 1 # order of multinom\n ni <- rowSums(dat) # cluster sizes\n num.clus <- dim(dat)[[1]]\n\n # helper vars\n betas.exp <- exp(betas)\n rho.sq <- rho ^ 2\n denom <- 1 + sum(betas.exp)\n dispi <- ni * (1 + (ni - 1) * rho.sq)\n p.vec <- as.vector(betas.exp / denom)\n p.mat <- matrix(p.vec, num.clus, m, byrow = TRUE)\n\n # create matrices\n var.mat <- diag(p.vec, m) - p.vec %*% t(p.vec)\n r.vec <- dat[, 1:m] - ni * p.mat\n res <- tr( r.vec %*% solve(var.mat) %*% t( (1 / dispi) * r.vec) ) - m * num.clus\n\n res\n}\n", "meta": {"hexsha": "d335867874796a9f39e3b6a36ab17ecf42e1e81f", "size": 1823, "ext": "r", "lang": "R", "max_stars_repo_path": "R/equations.r", "max_stars_repo_name": "Express50/omgee", "max_stars_repo_head_hexsha": "5022dd6dc268ecfd6042124b19772717cab75412", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2020-09-27T21:05:42.000Z", "max_stars_repo_stars_event_max_datetime": "2020-09-27T21:05:42.000Z", "max_issues_repo_path": "R/equations.r", "max_issues_repo_name": "Express50/omgee", "max_issues_repo_head_hexsha": "5022dd6dc268ecfd6042124b19772717cab75412", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/equations.r", "max_forks_repo_name": "Express50/omgee", "max_forks_repo_head_hexsha": "5022dd6dc268ecfd6042124b19772717cab75412", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 28.9365079365, "max_line_length": 86, "alphanum_fraction": 0.6039495337, "num_tokens": 628, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8311430478583169, "lm_q2_score": 0.640635854839898, "lm_q1q2_score": 0.5324600369589512}} {"text": "#!/usr/bin/Rscript\n\n# M x N Chess Board.\nM = 8; N = 8; board = matrix(0, nrow = M, ncol = N)\n\n# Get/Set value on a board position.\ngetboard = function (position) { board[position[1], position[2]] }\nsetboard = function (position, x) { board[position[1], position[2]] <<- x }\n\n# (Relative) Hops of a Knight.\nhops = cbind(c(-2, -1), c(-1, -2), c(+1, -2), c(+2, -1),\n c(+2, +1), c(+1, +2), c(-1, +2), c(-2, +1))\n\n# Validate a move.\nvalid = function (move) {\n all(1 <= move & move <= c(M, N)) && (getboard(move) == 0)\n}\n\n# Moves possible from a given position.\nexplore = function (position) {\n moves = position + hops\n cbind(moves[, apply(moves, 2, valid)])\n}\n\n# Possible moves sorted according to their Wornsdorff cost.\ncandidates = function (position) {\n moves = explore(position)\n\n # No candidate moves available.\n if (ncol(moves) == 0) { return(moves) }\n\n wcosts = apply(moves, 2, function (position) { ncol(explore(position)) })\n cbind(moves[, order(wcosts)])\n}\n\n# Recursive function for touring the chess board.\nknightTour = function (position, moveN) {\n\n # Tour Complete.\n if (moveN > (M * N)) {\n print(board)\n quit()\n }\n\n # Available moves.\n moves = candidates(position)\n\n # None possible. Backtrack.\n if (ncol(moves) == 0) { return() }\n\n # Make a move, and continue the tour.\n apply(moves, 2, function (position) {\n setboard(position, moveN)\n knightTour(position, moveN + 1)\n setboard(position, 0)\n })\n}\n\n# User Input: Starting position (in algebraic notation).\nsquare = commandArgs(trailingOnly = TRUE)\n\n# Convert into board co-ordinates.\nrow = M + 1 - as.integer(substr(square, 2, 2))\nascii = function (ch) { as.integer(charToRaw(ch)) }\ncol = 1 + ascii(substr(square, 1, 1)) - ascii('a')\nposition = c(row, col)\n\n# Begin tour.\nsetboard(position, 1); knightTour(position, 2)\n", "meta": {"hexsha": "cfc1bd613ebc879a114fb7d73efc630ea33e182d", "size": 1960, "ext": "r", "lang": "R", "max_stars_repo_path": "Task/Knights-tour/R/knights-tour.r", "max_stars_repo_name": "LaudateCorpus1/RosettaCodeData", "max_stars_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2018-11-09T22:08:38.000Z", "max_stars_repo_stars_event_max_datetime": "2018-11-09T22:08:38.000Z", "max_issues_repo_path": "Task/Knights-tour/R/knights-tour.r", "max_issues_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_issues_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Task/Knights-tour/R/knights-tour.r", "max_forks_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_forks_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2018-11-09T22:08:40.000Z", "max_forks_repo_forks_event_max_datetime": "2018-11-09T22:08:40.000Z", "avg_line_length": 28.0, "max_line_length": 77, "alphanum_fraction": 0.5887755102, "num_tokens": 572, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929104825007, "lm_q2_score": 0.6825737344123242, "lm_q1q2_score": 0.5324026737231782}} {"text": "## Trabajo final MAPP Otoño. Métodos cuantitativos.\n## Integrantes:\n## Guadalupe Álvarez\n## Ángel Carrillo\n## Eduardo Muñiz\n\nSys.setlocale(\"LC_ALL\", \"es_ES.UTF-8\") # Cambiar locale para prevenir problemas con caracteres especiales\noptions(scipen=999) # Prevenir notación científica\nrm(list=ls()) #Limpia todas las variables y objetos creados\n\n##### Paquetería ----\nlibrary(tidyverse)\nlibrary(ggplot2)\nlibrary(ggcorrplot)\nlibrary(ggthemes)\n\nlibrary(cluster)\nlibrary(dplyr)\nlibrary(pacman)\nlibrary(factoextra)\nlibrary(dendextend)\nlibrary(purrr)\n\n#PAQUETERÍA PARA ANOVA Y REGRESIÓN LINEAL\nlibrary(lme4)\nlibrary(lmerTest)\n\n\n##### Base de datos ----\ncomputoeinegi <- read.csv(\"2021 computoeinegi.csv\")\nstr(computoeinegi)\nhead(computoeinegi)\n# View(computoeinegi)\n\n###################################\n####### Indice de cosas por hacer:\n###################################\n\n# FECHA DE ENTREGA: 13 Y 14 DE DICIEMBRE\n\n# Limpiar la base\n# Estadistica descriptiva LALO\n# ANOVA LUPITA\n# Regresión lineal con variables explicativas CARILLO\n# Hito 1: DEBE ESTAR LISTO EL 25-NOV EN LA NOCHE\n# Ahí cordinarnos por mensajería sobre lo siguiente.\n\n# Usar variables explicativas para analisis de componentes principales\n# crear subconjunto de variables\n# utilizar estas variables explicativas para:\n# Conglomerados\n# Nuevo grupos\n# con nuevos grupos analisis jerárquico. \n\n##### Estadística descriptiva ----\n\ngr_participacion <- ggplot( computoeinegi, \n aes(x=distrito , y=participacion) )+ \n geom_point() + \n labs(title=\"Participación electoral\", \n subtitle=\"Proceso Electoral Federal 2021.\", \n x=\"Distrito electoral federal\",\n y=\"Porcentaje de participación\", \n caption=\"Elaboración propia con datos de INE(2021).\") +\n theme_economist()+\n theme(axis.text.x=element_text(angle=50,size=12))\n\nplot(gr_participacion)\n\n########\n ## ANOVA ##\n\n#Creamos una base con la variable categórica complejidad que nos indicará el número de grupos\n# y la variable de interés será el porcentaje de participación\n\nAnova_part <- computoeinegi [, -c(1,2,3,4,5,6,8,9,11,12,13,14,15,16,17)]\n\n# Gráfica simple\nggplot(data = Anova_part, aes(x = 1:nrow(Anova_part), y = participacion)) +\n geom_point() +\n theme_bw() \n\n# Gráfica por complejidad\nggplot(data = Anova_part, aes(x = 1:nrow(Anova_part), \n y = participacion, colour = complejidad)) +\n geom_point() +\n theme_bw() \n\n# Grafica de la gran media\nggplot(data = Anova_part, aes(x = 1:nrow(Anova_part), y = participacion, colour = complejidad)) +\n geom_point() +\n geom_hline(yintercept = mean(Anova_part$participacion), colour = \"purple\", size = 1) +\n theme_bw()\n\n# Graficar la gran media y las medias de los grupos\nggplot(data = Anova_part, aes(x = 1:nrow(Anova_part), y = participacion, colour = complejidad)) +\n geom_point() +\n geom_hline(yintercept = mean(Anova_part$participacion), colour = \"purple\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"\t\nAltamente Concentrado 1\"]), \n colour = \"orangered1\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"\t\nAltamente Concentrado 2\"]), \n colour = \"orange\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"Concentración Media\"]), \n colour = \"olivedrab\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"Concentrado 1\"]), \n colour = \"green\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"Concentrado 2\"]), \n colour = \"turquoise2\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"Disperso 1\"]), \n colour = \"blue\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"Disperso 2\"]), \n colour = \"royalblue\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"Muy Disperso 1\"]), \n colour = \"violet\", size = 1) +\n geom_hline(yintercept = mean(Anova_part$participacion[Anova_part$complejidad==\"Muy Disperso 2\"]), \n colour = \"deeppink\", size = 1) +\n theme_bw()\n\n###\n### ANOVA \n### Hipótesis Nula = las medias son iguales\n### Hipótesis Alternativa = No todas las medias son iguales\n\nanova(lm(participacion ~ complejidad, data = Anova_part))\n\n\n", "meta": {"hexsha": "ee9a127020d9b9744e177faa76af25c75caabf1a", "size": 4596, "ext": "r", "lang": "R", "max_stars_repo_path": "MainCode.r", "max_stars_repo_name": "galvarezr/R_MC_II_itza", "max_stars_repo_head_hexsha": "8dd489313c9c7c11f925f983aa0df7eff3fecb29", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "MainCode.r", "max_issues_repo_name": "galvarezr/R_MC_II_itza", "max_issues_repo_head_hexsha": "8dd489313c9c7c11f925f983aa0df7eff3fecb29", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "MainCode.r", "max_forks_repo_name": "galvarezr/R_MC_II_itza", "max_forks_repo_head_hexsha": "8dd489313c9c7c11f925f983aa0df7eff3fecb29", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.6279069767, "max_line_length": 105, "alphanum_fraction": 0.6714534378, "num_tokens": 1363, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7577943712746406, "lm_q2_score": 0.7025300449389326, "lm_q1q2_score": 0.5323733137060433}} {"text": "model_snowmelt <- function (ps = 0.0,\n M = 0.0){\n #'- Name: SnowMelt -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: Snow Melt\n #' * Author: STICS\n #' * Reference: doi:http://dx.doi.org/10.1016/j.agrformet.2014.05.002\n #' * Institution: INRA\n #' * Abstract: Snow melt\n #'- inputs:\n #' * name: ps\n #' ** description : density of snow cover\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 5000.0\n #' ** unit : kg/m**3\n #' ** uri : \n #' * name: M\n #' ** description : snow in the process of melting\n #' ** inputtype : variable\n #' ** variablecategory : rate\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : \n #' ** max : \n #' ** unit : mmW/d\n #' ** uri : \n #'- outputs:\n #' * name: Snowmelt\n #' ** description : Snow melt\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** min : 0.0\n #' ** max : 500.0\n #' ** unit : m\n #' ** uri : \n Snowmelt <- 0.0\n if (ps > 1e-8)\n {\n Snowmelt <- M / ps\n }\n return (list('Snowmelt' = Snowmelt))\n}", "meta": {"hexsha": "80456b725d53eebbd1913c27f23307821f47ed3e", "size": 1904, "ext": "r", "lang": "R", "max_stars_repo_path": "src/r/STICS_SNOW/Snowmelt.r", "max_stars_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_stars_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/r/STICS_SNOW/Snowmelt.r", "max_issues_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_issues_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/r/STICS_SNOW/Snowmelt.r", "max_forks_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_forks_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 41.3913043478, "max_line_length": 84, "alphanum_fraction": 0.2967436975, "num_tokens": 437, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8128673269042765, "lm_q2_score": 0.6548947290421275, "lm_q1q2_score": 0.5323425278001747}} {"text": "# __ ___ \n# / |/ /___ _______________ \n# / /|_/ / __ `/ ___/ ___/ __ \\ \n# / / / / /_/ / / / /__/ /_/ / \n# /_/ /_/\\__,_/_/ \\___/\\____/ \n# _ ___ __ __ ___ \n# | | / (_)____/ /_/ /_ / (_)___ \n# | | /| / / / ___/ __/ __ \\/ / / __ \\ \n# | |/ |/ / / / / /_/ / / / / / / / / \n# |__/|__/_/_/ \\__/_/ /_/_/_/_/ /_/-------------------------------------------------------------------------\n\n#S2876507\n\n# Running instructions: Press Ctrl+A, then Ctrl+Enter. \"plyr\" package must be installed.\n# If you zoom in on the graph, drag the window a bit to the right, until the plots are roughly of quadratic shape (nicer that way).\n\nlibrary(\"plyr\")\n\n# exp_dict according to Taatgen, van Rijn and Anderson (2004)\nexp_dict <- list(\n tick_increment = 1.1, \n b = 0.015, \n tick_base = 11,\n decay = -0.75, #decay parameter, default 0.5\n amount_subjects = 5, # subjects per interval\n amount_trials = 500, # amount of trials per subject and per interval\n subject_data = data.frame(matrix(ncol = 4, nrow = 0)),\n parm_t = 20\n)\n\n# act-r noise\nnoise <- function(number)\n{\n random <- runif(1, min = 0.0001, max = 0.9999)\n number * log( (1 - random ) / random)\n}\n\n# function which converts time in ms to an internal representation in pulses\nmsecs_to_pulses <- function(msecs)\n{\n pulse_counter = 0 # number of accumulated pulses\n time_counter <- 0 # passed time\n pulse <- exp_dict$tick_base # length of the initial pulse\n \n while (time_counter < msecs) \n {\n pulse <- exp_dict$tick_increment * pulse + noise(exp_dict$b * exp_dict$tick_increment * pulse) # generate next pulse\n pulse_counter <- pulse_counter + 1\n time_counter <- time_counter + pulse\n }\n \n pulse_counter\n}\n\n\n# function which produces time in ms based on an internal representation in pulses\npulse_to_msces <- function(pulses_input)\n{\n\n time_counter <- 0 # passed time\n pulse_counter <- 0 # number of pulses\n pulse <- exp_dict$tick_base # length of the initial pulse\n \n while (pulse_counter < pulses_input)\n {\n pulse <- exp_dict$tick_increment * pulse + noise(exp_dict$b * exp_dict$tick_increment * pulse) # generate next pulse\n pulse_counter <- pulse_counter + 1\n time_counter <- time_counter + pulse\n }\n \n time_counter\n}\n\n\n\nexperiment_setup <- function(current_subject) #I want setup the experiment parameters for each subject, once\n{\n real_time <- 0 #this is the subjects objective time in ms. this will stay with the subject for the duration of the experiment\n DecMem <- matrix(NA, 35, 900)\n #simuli for the trials get determined:\n short_condition_values <- seq(494, 847, length=11) #values of the short interval get determined\n short_condition_sample <- sample(short_condition_values, exp_dict$amount_trials, replace = TRUE)\n medium_condition_values <- seq(671, 1023, length=11) #values of the short interval get determined\n medium_condition_sample <- sample(medium_condition_values, exp_dict$amount_trials, replace = TRUE)\n long_condition_values <- seq(847, 1200, length=11) #values of the short interval get determined\n long_condition_sample <- sample(long_condition_values, exp_dict$amount_trials, replace = TRUE)\n \n condition_order <- sample(c(1, 2, 3)) #randomization of experimental condition order (1=short, 2=medium, 3=long)\n \n list(real_time = real_time, subject_nr = current_subject,\n DecMem = DecMem,\n short_condition_sample = short_condition_sample,\n medium_condition_sample = medium_condition_sample,\n long_condition_sample = long_condition_sample,\n condition_order = condition_order\n )\n}\n\nencounter_add <- function(subject_setup, sample_int_pulse)\n{\n tmp <- sum(!is.na(subject_setup$DecMem[sample_int_pulse, ])) + 1\n subject_setup$DecMem[sample_int_pulse, tmp] <- subject_setup$real_time\n subject_setup$real_time <- subject_setup$real_time + 50\n subject_setup\n}\n\nencounters_get <- function(subject_setup, current_row) \n{\n tmp <- subject_setup$DecMem[current_row,]\n \n if (length(tmp[!is.na(tmp)]) == 0){\n return(NA)\n }else{\n return(tmp[!is.na(tmp)])\n }\n \n}\n\nchunk_activation <- function(encounters, curtime) \n{\n if (curtime < min(encounters)) {\n return(NA)\n } else {\n sum((curtime - encounters[encounters 0)\n {\n encounters <- encounters_get(subject_setup, current_row)\n \n current_activation <- chunk_activation(encounters, subject_setup$real_time)\n \n total_activation <- total_activation + current_activation # ACTIVATION OF THE DECLARATIVE MEMORY \n }\n }\n\n for (current_row in seq(1, nrow(subject_setup$DecMem))) \n {\n amount_presentations <- sum(!is.na(subject_setup$DecMem[current_row, ]))\n \n if (amount_presentations > 0)\n {\n \n encounters <- encounters_get(subject_setup, current_row)\n \n current_activation <- chunk_activation(encounters, subject_setup$real_time)\n \n retrieval_prob <- prob_calc(current_activation, total_activation)\n blended_response <- blended_response + (retrieval_prob * current_row)\n \n }\n }\n \n blended_response\n}\n\ntrial <- function(current_condition, subject_setup)\n{\n if(current_condition == 1){\n current_condition_sample <- subject_setup$short_condition_sample} else if(current_condition == 2){\n current_condition_sample <- subject_setup$medium_condition_sample} else if(current_condition == 3){\n current_condition_sample <- subject_setup$long_condition_sample}\n \n for (.. in seq(1, length(current_condition_sample))) #length = amount_trials (default: 500)\n {\n ts <- current_condition_sample[..] # sample interval (ts) for the current trial\n subject_setup$real_time <- subject_setup$real_time + 1.000 + sample(seq(0.250, 0.850), 1) + 0.100 + ts + 0.50\n \n # presentation of central fixation point (+1000), + variable delay ranging from 0.25–0.85s\n # + \"READY?\" flash for 100 ms + actual time of the sample interval \n # + \"SET!\" flash (lasted 100ms, but tp was recorded alrady after 50ms)\n \n sample_int_pulse <- msecs_to_pulses(ts) # the sample interval gets converted into pulses\n \n subject_setup <- encounter_add(subject_setup, sample_int_pulse)\n \n blended_response <- blending(subject_setup)\n tp <- pulse_to_msces(blended_response)\n\n subject_setup$subject_data <- rbind(subject_setup$subject_data, c(subject_setup$subject_nr, current_condition, ts, tp))\n }\n \n subject_setup\n}\n\ntrain <- function(current_condition, subject_setup)\n{\n train_amount <- 250\n if(current_condition == 1){\n current_condition_sample <- subject_setup$short_condition_sample} else if(current_condition == 2){\n current_condition_sample <- subject_setup$medium_condition_sample} else if(current_condition == 3){\n current_condition_sample <- subject_setup$long_condition_sample}\n \n for (.. in seq(1, train_amount)) \n {\n ts <- current_condition_sample[..] # sample interval (ts) for the current trial\n subject_setup$real_time <- subject_setup$real_time + 1 + sample(seq(0.250, 0.850), 1) + 0.100 + ts + 0.50\n sample_int_pulse <- msecs_to_pulses(ts) # the sample interval gets converted into pulses\n subject_setup <- encounter_add(subject_setup, sample_int_pulse)\n }\n subject_setup\n}\n\n\n\nmain <- function()\n{\n simulated_data <- data.frame(matrix(ncol = 4, nrow = 0))\n \n for (.. in seq(1, exp_dict$amount_subjects)) #loop for subjects\n {\n current_subject <- ..\n subject_setup <- experiment_setup(current_subject)\n \n for (.. in subject_setup$condition_order) # loop containing all conditions\n {\n current_condition <- ..\n subject_setup$DecMem <- matrix(NA, 35, 900) #new DM each day(condition)\n subject_setup <- train(current_condition, subject_setup)\n subject_setup$real_time <- subject_setup$real_time + 1\n subject_setup <- trial(current_condition, subject_setup) #here the trials happen (another loop inside)\n\n }\n \n simulated_data <- rbind(simulated_data, subject_setup$subject_data)\n }\n \n colnames(subject_setup$subject_data) <- c(\"Subject_Nr \", \"Cond\", \"Ts\", \"Tp\")\n subject_data <- data.frame(subject_setup$subject_data)\n subject_data$Response[is.nan(subject_data$Response)] <- subject_data$Prior[is.nan(subject_data$Response)]\n subject_data\n}\n\n\n\nplotting <- function(dataframe)\n{\n\n brown <- \"#8b4513\";\n red <- \"#ff1100\";\n black <- \"#000000\";\n brownT <- \"#8b451322\";\n redT <- \"#ff110022\";\n blackT <- \"#00000022\";\n \n ## ---\n datJS <- dataframe\n par(mfrow=c(1,1))\n \n plotDatJS <- with(datJS,aggregate(list(Tp=Tp),list(Ts=Ts,Cond=Cond),mean))\n agregJS <- ddply(datJS, c(\"Cond\", \"Ts\"), summarise, Ts_mean = mean(Ts), Tp_mean = mean(Tp))\n yrange <- range(plotDatJS$Ts)*c(.95,1.05)\n \n with(plotDatJS[plotDatJS$Cond==3,],plot(Ts,Tp,type=\"b\",col=red,lwd=2,ylim=yrange,xlim=yrange,main=\"J&S All\"))\n with(plotDatJS[plotDatJS$Cond==2,],lines(Ts,Tp,type=\"b\",col=brown,lwd=2,ylim=yrange,xlim=yrange))\n with(plotDatJS[plotDatJS$Cond==1,],lines(Ts,Tp,type=\"b\",col=black,lwd=2,ylim=yrange,xlim=yrange))\n \n lines(c(yrange[1],yrange[2]),c(yrange[1],yrange[2]),col=\"darkgrey\",lty=2)\n \n with(datJS[datJS$Cond==3,],points(jitter(Ts),Tp,col=redT,pch=\".\",cex=3))\n with(datJS[datJS$Cond==2,],points(jitter(Ts),Tp,col=brownT,pch=\".\",cex=3))\n with(datJS[datJS$Cond==1,],points(jitter(Ts),Tp,col=blackT,pch=\".\",cex=3)) \n}\n\nplotting(main())\n", "meta": {"hexsha": "61473d74f27fe23b917a76c81ee27096b86533b2", "size": 10434, "ext": "r", "lang": "R", "max_stars_repo_path": "shadlen_assignment2.r", "max_stars_repo_name": "Seneketh/basic_cogmod", "max_stars_repo_head_hexsha": "1a81e7da92fced011554cfdb6b173f0e2c6a9f02", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "shadlen_assignment2.r", "max_issues_repo_name": "Seneketh/basic_cogmod", "max_issues_repo_head_hexsha": "1a81e7da92fced011554cfdb6b173f0e2c6a9f02", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "shadlen_assignment2.r", "max_forks_repo_name": "Seneketh/basic_cogmod", "max_forks_repo_head_hexsha": "1a81e7da92fced011554cfdb6b173f0e2c6a9f02", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.7394366197, "max_line_length": 131, "alphanum_fraction": 0.6480736055, "num_tokens": 2670, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8499711604559846, "lm_q2_score": 0.6261241842048092, "lm_q1q2_score": 0.5321874994381184}} {"text": "permutation.test.discrete <-\nfunction (x, y=NULL, scores, alternative=\"greater\", trials=1000) \n{\n\t# set up x and y properly\n\tif(length(y)) {\n\t\tn <- length(y)\n\t\tif(length(x) != n) stop(\"x and y have different lengths\")\n\t} else {\n\t\tif(ncol(x) != 2)\n\t\t\tstop(\"x does not have 2 columns and y is missing\")\n\t\ty <- x[, 2]\n\t\tx <- x[, 1]\n\t\tn <- length(y)\n\t}\n\tx <- as.character(x)\n\ty <- as.character(y)\n\n\t# set up alternative\n\tif(length(alternative) != 1 || !is.character(alternative))\n\t\tstop(\"alternative must be a single character string\")\n\taltnum <- pmatch(alternative, c(\"greater\", \"less\"), nomatch=NA)\n\tif(is.na(altnum)) \n\t\tstop(\"alternative must partially match 'greater' or 'less'\")\n\talternative <- c(\"greater\", \"less\")[altnum]\n\n\t# set up scores properly\n\torig.tab <- table(x, y)\n\totd <- dim(orig.tab)\n\todnam <- dimnames(orig.tab)\n\tscnam <- dimnames(scores)\n\tif(!is.matrix(scores) || length(scnam) != 2 || !is.numeric(scores)) \n\t\tstop(\"scores must be a numeric matrix with dimnames\")\n\tscd <- dim(scores)\n\tif(any(scd != otd) && any(rev(scd) != otd)) {\n\t\tstop(paste(\"scores is not the proper size, should be\",\n\t\t\totd[1], \"by\", otd[2]))\t\n\t}\n\tif(any(scd != otd)) {\n\t\tscores <- t(scores)\n\t\tscd <- dim(scores)\n\t\tscnam <- dimnames(scores)\n\t\treverse <- TRUE\n\t} else {\n\t\treverse <- FALSE\n\t}\n\trownum <- match(scnam[[1]], odnam[[1]], nomatch=NA)\n\tif(any(is.na(rownum))) {\n\t\tif(reverse || otd[1] != otd[2])\n\t\t\tstop(\"bad dimnames for scores\")\n\t\tscores <- t(scores)\n\t\tscd <- dim(scores)\n\t\tscnam <- dimnames(scores)\n\t\trownum <- match(scnam[[1]], odnam[[1]], nomatch=NA)\n\t\tif(any(is.na(rownum))) stop(\"bad dimnames for scores\")\n\t}\n\tcolnum <- match(scnam[[2]], odnam[[2]], nomatch=NA)\n\tif(any(is.na(colnum))) stop(\"bad dimnames for scores\")\n\tscores <- scores[rownum, colnum]\n\n\t# the main event\n\transeed <- .Random.seed\n\torig.score <- sum(orig.tab * scores)\n\tperm.scores <- numeric(trials)\n\tfor(i in 1:trials) {\n\t\tperm.scores[i] <- sum(table(x, sample(y)) * scores)\n\t}\n\n\tif(alternative == \"greater\") {\n\t\textreme <- sum(perm.scores >= orig.score)\n\t} else {\n\t\textreme <- sum(perm.scores <= orig.score)\n\t}\n\tans <- list(original.score=orig.score, perm.scores=perm.scores,\n\t\tstats=c(nobs=n, trials=trials, extreme=extreme),\n\t\talternative=alternative, random.seed=ranseed, call=match.call())\n\tclass(ans) <- \"permtstBurSt\"\n\tans\n}\npermutation.test.fun <-\nfunction (x, y=NULL, fun=function(x, y) sum(x * y), alternative=\"greater\", \n\ttrials=1000) \n{\n\t# set up x and y properly\n\tif(length(y)) {\n\t\tn <- length(y)\n\t\tif(length(x) != n) stop(\"x and y have different lengths\")\n\t\tif(!is.numeric(y)) stop(\"y must be numeric\")\n\t} else {\n\t\tif(ncol(x) != 2)\n\t\t\tstop(\"x does not have 2 columns and y is missing\")\n\t\tx <- as.matrix(x)\n\t\tif(!is.numeric(x)) stop(\"x must be numeric\")\n\t\ty <- x[, 2]\n\t\tx <- x[, 1]\n\t\tn <- length(y)\n\t}\n\n\t# set up alternative\n\tif(length(alternative) != 1 || !is.character(alternative))\n\t\tstop(\"alternative must be a single character string\")\n\taltnum <- pmatch(alternative, c(\"greater\", \"less\"), nomatch=NA)\n\tif(is.na(altnum)) \n\t\tstop(\"alternative must partially match 'greater' or 'less'\")\n\talternative <- c(\"greater\", \"less\")[altnum]\n\n\t# the main event\n\transeed <- .Random.seed\n\torig.score <- fun(x, y)\n\tif(length(orig.score) != 1) stop(\"fun must return a single number\")\n\n\tperm.scores <- numeric(trials)\n\tfor(i in 1:trials) {\n\t\tperm.scores[i] <- fun(x, sample(y))\n\t}\n\n\tif(alternative == \"greater\") {\n\t\textreme <- sum(perm.scores >= orig.score)\n\t} else {\n\t\textreme <- sum(perm.scores <= orig.score)\n\t}\n\tans <- list(original.score=orig.score, perm.scores=perm.scores,\n\t\tstats=c(nobs=n, trials=trials, extreme=extreme),\n\t\talternative=alternative, random.seed=ranseed, call=match.call())\n\tclass(ans) <- \"permtstBurSt\"\n\tans\n}\nplot.permtstBurSt <-\nfunction (x, col=c(\"black\", \"red\"), width=10, uniqlim=40, main=\"\", \n\txlab=\"Scores\", col.hist=\"yellow\", ...) \n{\n\torig.score <- x$original.score\n\tulen <- length(unique(x$perm.scores))\n\tif(ulen > uniqlim) {\n\t\thist(x$perm.scores, xlim=range(x$perm.scores, orig.score),\n\t\t\t main=main, xlab=xlab, col=col.hist, ...)\n\t\tbox()\n\t\tabline(v=orig.score, col=col[2])\n\t} else {\n\t\tptab <- table(x$perm.scores)\n\t\tvals <- as.numeric(names(ptab))\n\t\tif(x$alternative == \"greater\") {\n\t\t\textreme <- vals >= orig.score\n\t\t} else {\n\t\t\textreme <- vals <= orig.score\n\t\t}\n\t\tif(all(vals > orig.score) || all(vals < orig.score)) {\n\t\t\txrng <- range(vals, orig.score)\n\t\t} else {\n\t\t\txrng <- range(vals)\n\t\t}\n\t\tplot(vals, as.vector(ptab), type=\"n\", xlab=xlab, ylab=\"Count\", \n\t\t\txlim=xrng, ...)\n\t\tpoints(vals[!extreme], ptab[!extreme], type=\"h\",\n\t\t\tcol=col[1], lwd=width, ...)\n\t\tpoints(vals[extreme], ptab[extreme], type=\"h\",\n\t\t\tcol=col[2], lwd=width, ...)\n\t\tif(nchar(main)) title(main=main)\n\t}\n}\nprint.permtstBurSt <-\nfunction (x, digits=4, ...) \n{\n\tcat(\"Call:\\n\")\n\tprint(x$call)\n\tcat(\"\\nOriginal value:\", x$original.score, \" Number of observations:\",\n\t\tx$stats[\"nobs\"], \"\\n\")\n\tcat(\"Number of random permutations:\", x$stats[\"trials\"], \n\t\t\" Alternative:\", x$alternative, \" p-value:\", \n\t\tround(x$stats[\"extreme\"] / x$stats[\"trials\"], digits), \"\\n\")\n\tinvisible(x)\n}\nsuperscore <-\nstructure(c(1, 0, 0, 1), .Dim = c(2L, 2L), .Dimnames = list(c(\"Up\", \n\"Down\"), c(\"National\", \"American\")))\n", "meta": {"hexsha": "eea3bdb2379859b49bc177c839d87f953d84ef72", "size": 5179, "ext": "r", "lang": "R", "max_stars_repo_path": "Hypotesis_testing/permutationTest.r", "max_stars_repo_name": "isix/R", "max_stars_repo_head_hexsha": "806e2a22e5abd93dc7933d3b9e8c3368562e1eaa", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Hypotesis_testing/permutationTest.r", "max_issues_repo_name": "isix/R", "max_issues_repo_head_hexsha": "806e2a22e5abd93dc7933d3b9e8c3368562e1eaa", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Hypotesis_testing/permutationTest.r", "max_forks_repo_name": "isix/R", "max_forks_repo_head_hexsha": "806e2a22e5abd93dc7933d3b9e8c3368562e1eaa", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 29.5942857143, "max_line_length": 75, "alphanum_fraction": 0.6310098475, "num_tokens": 1682, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7185943925708562, "lm_q2_score": 0.7401743620390163, "lm_q1q2_score": 0.531885146085948}} {"text": "#' f_h1\n#'\n#' Construct the $K_\\tau K_1$ vector $h_1$\n#'\n#' @param n1 matrix n1 output from f_n1\n#' @param g1 $G_1$ diagonal matrix\n#' @param g2 $G_2$ diagonal matrix\n#' @param sigmatau $\\Sigma_tau$ diagonal matrix\n#'\n#'\n#' @return $K_\\tau K_1$ vector $H_1$\n#'\n#' @examples\n#' \\dontrun{\n#' v_taustar <- f_tausearch(y, n1, til_beta, m_sigma, ktau, mtau)\n#' m_ntaustar <- f_ntaustar(v_taustar, ktau, mtau)\n#' m_s <- f_mats(ktau + mtau, k1 + m1, rep(1, 3))\n#' m_g1 <- f_g1(ktau, mtau)\n#' m_g2 <- f_g2(ktau, mtau)\n#' m_sigmatau <- f_sigmatau(ktau + mtau)\n#' m_h1 <- f_h1(n1, m_g1, m_g2, m_sigmatau)\n#' m_h2 <- f_h2(n1, m_ntaustar, m_g1, m_sigmatau)\n#' }\nf_h1 <- function(n1, g1, g2, sigmatau) {\n #form N^{m_tau+2}(\\tau)|_0^1\n nm2 <- rep(0, nrow(sigmatau)); nm2[nrow(sigmatau)] <- 1\n #Calculate $H_1$\n h_1 <- tcrossprod(g1, sigmatau) %*% tcrossprod(g2, sigmatau)\n h_1 <- -h_1 %*% nm2\n h_1 <- kronecker(h_1, t(n1))\n h_1 <- rowSums(h_1)\n return(h_1)\n}\n\n\n#' f_h2\n#'\n#' Construct the $K_\\tauK_1$ vector $H_2$\n#'\n#' @param n1 matrix n1 output from f_n1\n#' @param ntaustar matrix ntau output from f_ntaustar\n#' @param g1 $G_1$ diagonal matrix\n#' @param sigmatau $\\Sigma_tau$ diagonal matrix\n#'\n#'\n#' @return $K_\\tau K_1$ vector $H_2$\n#'\n#' @import mgcv\n#'\n#' @examples\n#' \\dontrun{\n#' v_taustar <- f_tausearch(y, n1, til_beta, m_sigma, ktau, mtau)\n#' m_ntaustar <- f_ntaustar(v_taustar, ktau, mtau)\n#' m_s <- f_mats(ktau + mtau, k1 + m1, rep(1, 3))\n#' m_g1 <- f_g1(ktau, mtau)\n#' m_g2 <- f_g2(ktau, mtau)\n#' m_sigmatau <- f_sigmatau(ktau + mtau)\n#' m_h1 <- f_h1(n1, m_g1, m_g2, m_sigmatau)\n#' m_h2 <- f_h2(n1, m_ntaustar, m_g1, m_sigmatau)\n#' }\nf_h2 <- function(n1, ntaustar, g1, sigmatau) {\n h_2 <- tcrossprod(g1, sigmatau) %*% t(ntaustar)\n#ptm <- proc.time()# Start the clock!\n h_2 <- mgcv::tensor.prod.model.matrix(list(t(h_2), n1))\n h_2 <- colSums(h_2)\n#print(proc.time() - ptm)# Stop the clock\n\n return(h_2)\n}\n\n#' f_lossderiv\n#'\n#' Construct the $K_\\tau K_1$ gradient vector for the loss\n#'\n#' @param til_beta the current estimation fro $\\tilde{\\beta}$\n#' @param v_dim vector contain the dimension of covariates\n#' $(K_\\tau, K_1, K_2, K_3, \\cdots)$\n#'\n#' @param m_sigma $K_\\tau K_1K_2 \\times K_1K_2 \\Sigma$ diagonal block matix.\n#' @param h1 $K_\\tau K_1K_2$ spline vector for the first covariate\n#' @param h2 $K_\\tau K_1K_2$ spline vector for the tau covariate\n#'\n#'\n#' @return the $K_\\tau K_1$ gradient vector for the loss\n#'\n#' @examples\n#' \\dontrun{\n#' v_taustar <- f_tausearch(y, n1, til_beta, m_sigma, ktau, mtau)\n#' m_ntaustar <- f_ntaustar(v_taustar, ktau, mtau)\n#' m_s <- f_mats(ktau + mtau, k1 + m1, rep(1, 3))\n#' m_g1 <- f_g1(ktau, mtau)\n#' m_g2 <- f_g2(ktau, mtau)\n#' m_sigmatau <- f_sigmatau(ktau + mtau)\n#' m_h1 <- f_h1(n1, m_g1, m_g2, m_sigmatau)\n#' m_h2 <- f_h2(n1, m_ntaustar, m_g1, m_sigmatau)\n#' v_lossderiv <- f_lossderiv(til_beta, c(ktau + mtau, k1 + m1), m_sigma, m_h1, m_h2)\n#' }\nf_lossderiv <- function(til_beta, v_dim, m_sigma, h1, h2) {\n m_c <- f_matc(v_dim, til_beta)\n lossderiv <- - m_c %*% crossprod(m_sigma, (h1 + h2))\n return(c(lossderiv))\n}\n\n#' f_grad\n#'\n#' Construct the $K_\\tau K_1K_2$ gradient vector for $\\beta$\n#'\n#' @param beta the current estimation fro $\\beta$\n#' @param loss_deriv the derivative of the loss\n#' @param s $K_\\tau K_1K_2 \\times K_1K_2$ $S$ matrix\n#'\n#'\n#' @return the $K_\\tau K_1K_2$ gradient vector for $\\beta$\n#'\n#' @examples\n#' \\dontrun{\n#' v_taustar <- f_tausearch(y, n1, til_beta, m_sigma, ktau, mtau)\n#' m_ntaustar <- f_ntaustar(v_taustar, ktau, mtau)\n#' m_s <- f_mats(ktau + mtau, k1 + m1, rep(1, 3))\n#' m_g1 <- f_g1(ktau, mtau)\n#' m_g2 <- f_g2(ktau, mtau)\n#' m_sigmatau <- f_sigmatau(ktau + mtau)\n#' m_h1 <- f_h1(n1, m_g1, m_g2, m_sigmatau)\n#' m_h2 <- f_h2(n1, m_ntaustar, m_g1, m_sigmatau)\n#' v_lossderiv <- f_lossderiv(til_beta, c(ktau + mtau, k1 + m1), m_sigma, m_h1, m_h2)\n#' f_grad(beta, v_lossderiv, m_s)\n#' }\nf_grad <- function(beta, loss_deriv, s) {\n loss_deriv + c(s %*% beta)\n}\n\n#' f_hess\n#'\n#' Construct the $K_\\tau K_1K_2 \\times K_\\tau K_1K_2$ Hessian matrix for $\\beta$\n#'\n#' @param loss_deriv the derivative of the loss\n#' @param v_dim vector contain the dimension of covariates\n#' $(K_\\tau, K_1, K_2, K_3, \\cdots)$\n#' @param s $K_\\tau K_1K_2 \\times K_\\tau K_1K_2$ $S$ matrix\n#'\n#'\n#' @return the $K_\\tau K_1K_2$ gradient vector for $\\beta$\n#'\n#' @examples\n#' \\dontrun{\n#' v_taustar <- f_tausearch(y, n1, til_beta, m_sigma, ktau, mtau)\n#' m_ntaustar <- f_ntaustar(v_taustar, ktau, mtau)\n#' m_s <- f_mats(ktau + mtau, k1 + m1, rep(1, 3))\n#' m_g1 <- f_g1(ktau, mtau)\n#' m_g2 <- f_g2(ktau, mtau)\n#' m_sigmatau <- f_sigmatau(ktau + mtau)\n#' n_m2 <- rep(0, ktau + mtau)\n#' n_m2[ktau + mtau] <- 1\n#' m_h1 <- f_h1(n1, m_g1, m_g2, m_sigmatau, n_m2)\n#' m_h2 <- f_h2(n1, m_ntaustar, m_g1, m_sigmatau)\n#' v_lossderiv <- f_lossderiv(til_beta, c(ktau + mtau, k1 + m1), m_sigma, m_h1, m_h2)\n#' f_grad(beta, v_lossderiv, m_s)\n#' m_hess <- f_hess(v_lossderiv, c(ktau + mtau, k1 + m1), m_s)\n#' }\nf_hess <- function(loss_deriv, v_dim, s) {\n v_idx <- seq_len(v_dim[2])\n loss_deriv[v_idx] <- 0\n return(diag(loss_deriv) + s)\n}", "meta": {"hexsha": "d24438575ec42b4854522a39828baf3e639b2289", "size": 5156, "ext": "r", "lang": "R", "max_stars_repo_path": "R/derivative.r", "max_stars_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_stars_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/derivative.r", "max_issues_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_issues_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/derivative.r", "max_forks_repo_name": "ZhuolinSong/-Quantile-sheet-estimator-with-shape-constraints", "max_forks_repo_head_hexsha": "f871f5ac7b1d7a6add799f023b23d10ce899d5d6", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 31.4390243902, "max_line_length": 85, "alphanum_fraction": 0.6404189294, "num_tokens": 2179, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8539127678225574, "lm_q2_score": 0.6224593312018545, "lm_q1q2_score": 0.5315259703635535}} {"text": "# Copyright 2017 Province of British Columbia\n#\n# Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n# http://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and limitations under the License.\n\n\n###############################################################################\n# Fit a Bayesian model for the replicated observations at the same species\n# using data cloning.\n#\n# This function can be used for different distributions \n# by selectively extracting the code with the tag number at the right\n\nmodel {\n # Input data for logConc - log-concentration\n # SpeciesNum - species number to link replicate observations\n # NData - number of data values\n # NSpecies - number of unique species\n #\n # Parameters for distributions vary depending on the distribution\n # lnorm - mu, sigma - intercept and slope (on log scale) - lognormal\n \n # Parameters for observal error\n # sigma_obs - sd of observational error\n #\n \n \n # observed data comes from underlying mean with a lognormal distribution\n # this is true regardless of the distribution of the mu_species whose\n # distributions can change\n # We want separate latent variables for each clone because this sample size\n # is NOT increasing as we increase the number of chemicals tested.\n for(k in 1:K){\n for(i in 1:NData){\n logConc[i,k] ~ dnorm(log_mu_species[SpeciesNum[i],k], tau_obs)\n } \n }\n # Contributions to the likelihood from the observed data#\n# for(i in 1:NData){\n# likc.data[i] <- dnorm(logConc[i,1], mu_species[SpeciesNum[i]], tau_obs)\n# }\n# lik.data <- prod(likc.data[]) # total contribuiton to the likelihood from data\n # priors and derived variables for the observation process\n tau_obs <- 1/(sigma_obs*sigma_obs)\n sigma_obs ~dunif(.01, 5)\n \n #--------------------------------------------------------------------\n # LOGNORMAL distribution for the \"mean\" of each species \n # fit a distribution to the mean of each observation\n for(k in 1:K){\n for(i in 1:NSpecies){\n mu_species[i,k] ~ dlnorm(mu, tau)\n log_mu_species[i,k] <- log(mu_species[i,k])\n# likc.latent[i] <- dlnorm(mu_species[i], mu, tau) # contribution from the likelihoo\n }\n }\n# lik.latent <- prod(likc.latent[]) # total contribution to the likelihood from latent species value\n tau <- 1/(sigma*sigma)\n sigma ~ dunif(.01,5)\n \n mu ~ dnorm(0, .001)\n \n # estimate the average ranking of the species for each clone\n for(k in 1:K){\n rank_mu_species[1:NSpecies,k] <- rank(mu_species[1:NSpecies,k])\n }\n \n # estimate the HC[5]\n hc <- qlnorm(.05, mu, tau) # the HC[5] \n hc.log <- log(hc)\n #------------------------------------------------------------------\n \n # total complete-data likelihood\n# lik <- lik.data * lik.latent\n} # end of model.jags\n\n", "meta": {"hexsha": "a67bdf615fde8335b1dfdc50b7709ce0ef4c89b2", "size": 3215, "ext": "r", "lang": "R", "max_stars_repo_path": "DataCloningTesting/fit.distr.clone-lnorm.r", "max_stars_repo_name": "bcgov/SSD-methods", "max_stars_repo_head_hexsha": "8ec33c5029f8016fe24d66a127ef744b9696d1bd", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2017-04-19T05:30:14.000Z", "max_stars_repo_stars_event_max_datetime": "2017-12-01T05:50:56.000Z", "max_issues_repo_path": "DataCloning/fit.distr.clone.r", "max_issues_repo_name": "bcgov/SSD-methods", "max_issues_repo_head_hexsha": "8ec33c5029f8016fe24d66a127ef744b9696d1bd", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2018-03-01T22:58:58.000Z", "max_issues_repo_issues_event_max_datetime": "2020-10-05T19:13:14.000Z", "max_forks_repo_path": "DataCloningTesting/fit.distr.clone-lnorm.r", "max_forks_repo_name": "bcgov/SSD-methods", "max_forks_repo_head_hexsha": "8ec33c5029f8016fe24d66a127ef744b9696d1bd", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2017-12-01T04:03:46.000Z", "max_forks_repo_forks_event_max_datetime": "2017-12-01T04:03:46.000Z", "avg_line_length": 38.2738095238, "max_line_length": 135, "alphanum_fraction": 0.6435458787, "num_tokens": 784, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8539127529517043, "lm_q2_score": 0.6224593312018546, "lm_q1q2_score": 0.5315259611070524}} {"text": "library(\"animation\")\nlibrary(\"plyr\")\nlibrary(\"MASS\")\n#\n# ################################################################################\n# # 0\n# ################################################################################\n#\n# library(\"ElemStatLearn\")\n# require(\"class\")\n# mixture.example$xnew\n# x <- mixture.example$x\n# g <- mixture.example$y\n# xnew <- mixture.example$xnew\n# mod15 <- knn(x, xnew, g, k=15, prob=TRUE)\n# prob <- attr(mod15, \"prob\")\n# prob <- ifelse(mod15==\"1\", prob, 1-prob)\n# px1 <- mixture.example$px1\n# px2 <- mixture.example$px2\n# prob15 <- matrix(prob, length(px1), length(px2))\n# par(mar=rep(2,4))\n# contour(px1, px2, prob15, levels=0.5, labels=\"\", xlab=\"\", ylab=\"\", main=\n# \"15-nearest neighbour\", axes=FALSE)\n# points(x, col=ifelse(g==1, \"coral\", \"cornflowerblue\"))\n# gd <- expand.grid(x=px1, y=px2)\n# points(gd, pch=\".\", cex=1.2, col=ifelse(prob15>0.5, \"coral\", \"cornflowerblue\"))\n# box()\n\n\n# spirals\nspiral <- function (radius, theta, sequence, col) {\n r <- radius(sequence)\n t <- theta(sequence)\n x <- r * sin(t)\n y <- r * cos(t)\n return(data.frame(x=x, y=y))\n # points(x, y, col=col)\n}\nsigmoid <- function (n) 1/(1 + exp(-n))\n\ncomputeLayer <- function (input, weights) {\n sums <- colSums(input * weights)\n return(sigmoid(sums))\n}\n\n\ncomputeAll <- function (input, weights) {\n numberOfLayers <- length(weights)\n output <- list()\n output[[1]] <- input\n for (i in 1:numberOfLayers) {\n output[[i + 1]] <- computeLayer(output[[i]], weights[[i]])\n }\n return(output)\n}\n\nclassify <- function (input, weights) {\n numberOfLayers <- length(weights)\n output <- input\n for (i in 1:numberOfLayers) {\n output <- computeLayer(output, weights[[i]])\n }\n return(output)\n}\n\nmse <- function (target, output) {\n diff <- target - output\n return(sum(diff * diff)/2)\n}\n\nlearn <- function (weights, input, target, learningRate = 0.5) {\n\n outputs <- computeAll(input, weights)\n numberOfLayers <- length(weights)\n\n # linear programming, tabulate deltas\n deltas <- list()\n\n # tabulate the output layer first\n output <- outputs[[numberOfLayers + 1]]\n deltas[[numberOfLayers]] <- (target - output) * output * (1 - output)\n # tabulate the rest\n if (numberOfLayers > 1) {\n for (layer in (numberOfLayers - 1):1) {\n output <- outputs[[layer + 1]]\n weight <- weights[[layer + 1]]\n\n sums <- rowSums(deltas[[layer + 1]] * weights[[layer + 1]])\n deltas[[layer]] <- sums * output * (1 - output)\n }\n }\n\n for (layer in 1:numberOfLayers) {\n weights[[layer]] <- weights[[layer]] + outputs[[layer]] %o% deltas[[layer]] * learningRate\n }\n\n return(list(\n weights = weights,\n error = mse(target, outputs[[numberOfLayers + 1]])\n ))\n # return(weights)\n}\n\nplotErrors <- function (errors) {\n if (length(errors) > 0) {\n plot(errors, xlab=\"iteration\", ylab=\"error\", type=\"h\")\n title(\"error / iteration\")\n }\n}\n\n\nreconstruct <- function (weights) {\n structure <- c()\n for (i in 1:length(weights)) {\n structure <- c(structure, nrow(weights[[i]]))\n }\n structure <- c(structure, ncol(weights[[i]]))\n return(structure)\n}\n\nidentityWeights <- function (nodes) {\n weights <- list()\n for (i in 1:(length(nodes) - 1)) {\n rows <- nodes[i]\n cols <- nodes[i + 1]\n weights[[i]] <- matrix(0, rows, cols)\n }\n return(weights)\n}\n\ninitializeWeights <- function (nodes) {\n weights <- list()\n for (i in 1:(length(nodes) - 1)) {\n rows <- nodes[i]\n cols <- nodes[i + 1]\n size <- rows * cols\n weights[[i]] <- matrix(sigmoid(rexp(size)), rows, cols)\n }\n return(weights)\n}\n\naccumulateWeights <- function (acc, x) {\n for (i in 1:length(acc)) {\n acc[[i]] <- acc[[i]] + x[[i]]\n }\n return(acc)\n}\n\ntrain <- function (weights, callback, iteration = 100000, until = 0.1, every = 1000) {\n\n errors <- c()\n learningRate <- 1\n\n for (i in 1:iteration) {\n accumulatedError <- 0\n\n # class 0\n size <- length(class0$x)\n for (j in 1:size) {\n data <- class0[j, ]\n result <- learn(weights, c(data$x, data$y, 1), c(1, 0), learningRate)\n weights <- result$weights\n accumulatedError <- accumulatedError + result$error\n }\n # class 1\n size <- length(class1$x)\n for (j in 1:size) {\n data <- class1[j, ]\n result <- learn(weights, c(data$x, data$y, 1), c(0, 1), learningRate)\n weights <- result$weights\n accumulatedError <- accumulatedError + result$error\n }\n errors <- c(errors, accumulatedError)\n\n if (accumulatedError < until) {\n print(i)\n break\n }\n\n if (i %% every == 0) {\n callback(i, weights, errors)\n }\n }\n return(list(\n weights = weights,\n errors = errors\n ))\n}\n\ndrawGrid <- function (weights) {\n grid <- expand.grid(x=seq(-8, 8, by=0.2), y=seq(-8, 8, by=0.2))\n gridResult <- aaply(grid, 1, function (data) {\n result <- classify(c(data$x, data$y, 1), weights)\n return(ifelse(result[1] > result[2], \"cornflowerblue\", \"coral\"))\n })\n points(grid, pch=\".\", cex=3, col=gridResult)\n}\n\n################################################################################\n# 1\n################################################################################\n\n\nplot(NA, NA, xlim=c(-8,8), ylim=c(-8,8),\n xlab=\"x\", ylab=\"y\",\n axes=FALSE, asp=1)\naxis(side=1, at=seq(-8, 8, by=2), cex.axis=0.8)\naxis(side=2, at=seq(-8, 8, by=2), cex.axis=0.8)\ntitle(\"Two Spiral Problem\")\n\n# class0 <- spiral(function (i) (6.5 * (104 - i) / 104),\n# function (i) (pi * i / 8),\n# 40:52)\n#\n# class1 <- spiral(function (i) (6.5 * (104 - i) / -104),\n# function (i) (pi * i / 8),\n# 40:52)\n# class0 <- spiral(function (i) (6.5 * (104 - i) / 104),\n# function (i) (pi * i / 8),\n# 0:10)\n#\n# class1 <- spiral(function (i) (6.5 * (104 - i) / -104),\n# function (i) (pi * i / 8),\n# 0:10)\nclass0 <- data.frame(x=c(-5), y=c(-2))\nclass1 <- data.frame(x=c(1), y=c(-1))\n\n# class0 <- data.frame(x=c(-4, -4, 4, 0), y=c(-4, 4, 4, 0))\n# class1 <- data.frame(x=c(4), y=c(-4))\n\n# class0 <- data.frame(x=c(-4, -4, 4, 0), y=c(-4, 4, -4, 0))\n# class1 <- data.frame(x=c(4, -2, 0, 2, 4), y=c(4, -2, -2, -2, -2))\n# class0 <- data.frame(x=c(-4, 4), y=c(4, -4))\n# class1 <- data.frame(x=c(4, -4), y=c(4, -4))\n\n\npoints(class0, col=\"cornflowerblue\", cex=2)\npoints(class1, col=\"coral\", cex=2)\n\n\n\nlayers <- c(3, 5, 2)\n# weights <- initializeWeights(layers)\n\nweights <- list()\nweights[[1]] <- matrix(\n c(1.716595448658756, 1.1764761912424921, 1.557118196208473, 1.7041037817911824, 1.6406029506187856,\n 0.6146408255673677, 1.2674559523877986, 0.8036832609525906, 0.4969173598004548, 0.7056984961660893,\n 0.978526932134667, 0.3158700142016504, 0.7265454504592824, 0.6660428783020633, 0.6623111268153622),\n 3, 5, byrow = TRUE)\nweights[[2]] <- matrix(\n c(-1.3136394181316513, 1.0616230793282784,\n -0.1289445413211639, 1.0017360756800076,\n -0.9759845197492095, 1.0231283450037518,\n -1.276113700799778, 1.1112948075375635,\n -1.325443401614583, 1.1986825244676422), 5, 2, byrow = TRUE)\n\n# print(computeLayer(c(1, -1, 1), weights[[1]]))\n\nresult <- train(weights, function (i, weights, errors) {\n print(errors[i])\n }, 10000)\nprint(result$weights)\ndrawGrid(result$weights)\nplotErrors(result$errors)\n", "meta": {"hexsha": "19f175c05732535abce3233996748c479cf209f8", "size": 7567, "ext": "r", "lang": "R", "max_stars_repo_path": "dreist/main.r", "max_stars_repo_name": "banacorn/neuron", "max_stars_repo_head_hexsha": "e05367a976cb43c607198b5b93ed4d9222e76160", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "dreist/main.r", "max_issues_repo_name": "banacorn/neuron", "max_issues_repo_head_hexsha": "e05367a976cb43c607198b5b93ed4d9222e76160", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "dreist/main.r", "max_forks_repo_name": "banacorn/neuron", "max_forks_repo_head_hexsha": "e05367a976cb43c607198b5b93ed4d9222e76160", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 28.4473684211, "max_line_length": 104, "alphanum_fraction": 0.5428835734, "num_tokens": 2367, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7745833945721304, "lm_q2_score": 0.6859494485880928, "lm_q1q2_score": 0.531325052392246}} {"text": "covZ <- function(t_seq, Y, C, sigma2)\n{\n #########################################################################\n # covZ: calculate the covariance matrix of paired-y product\n #\n # Input \n # t_seq: defined grid point m-by-1\n # Y: response vector m-by-1 \n # C: covariance matrix of X m-by-m \n # sigma2: variance of random error\n # \n # Output \n # CovZZ: covariance matrix of paired-y product m^2-by-m^2 \n # Z.info: vectorized Z.tri and the corresponding grid points \n # Z.tri: upper triangle matrix of paired-y product m-by-m\n # \n #########################################################################\n temp <- expand.grid(Y,Y)\n t.pair <- expand.grid(t_seq,t_seq)\n d <- length(Y)\n id <- expand.grid(1:d,1:d)\n m <- d\n\n Z <- apply(temp, 1, function(z) prod(z))\n Z.info <- as.matrix(cbind(Z,t.pair,id))\n Z.info <- Z.info[Z.info[,5] <= Z.info[,4],]\n \n# Z.mat <- matrix(Z,nrow=length(Y),ncol=length(Y),byrow=T)\n# Z.tri <- Z.mat\n# Z.tri[lower.tri(Z.tri)] <- NA\n \n Z.tri <- matrix(NA,nrow=length(Y),ncol=length(Y))\n Z.tri[lower.tri(Z.tri, diag=T)] <- Z.info[,1]\n Z.tri <- t(Z.tri)\n\n # Quick-n-dirty implementation (for testing only)\n# CovZZ <- matrix(NA,nrow=length(Z),ncol=length(Z))\n# time.start <- proc.time()\n# for (i in 1:m)\n# {\n# for (j in 1:m)\n# {\n# row.id <- d * (i - 1) + j\n# for (k in 1:m)\n# {\n# for (l in 1:m)\n# {\n# col.id <- d * (k - 1) + l\n# if (row.id <= col.id)\n# {\n# CovZZ[row.id,col.id] = C[i,k] * C[j,l] + C[i,l] * C[j,k] + \n# (i==k) * (j==l) *sigma2^2 + (i==l) * (j==k) *sigma2^2 +\n# (C[i,k] * (j==l) + C[i,l] * (j==k) + C[j,k] * (i==l) \n# + C[j,l] * (i==k)) * sigma2\n# }\n# \n# }\n# }\n# }\n# }\n# print((proc.time()-time.start)[3]) \n# \n# # Avoid loops (even slower, for testing only)\n# time.start <- proc.time()\n# id1 <- as.matrix(expand.grid(1:d,1:d,1:d,1:d))\n# id2 <- as.matrix(expand.grid(1:d^2,1:d^2))\n# id.tab <- as.matrix(cbind(id1,id2))\n# id.tab <- id.tab[id.tab[,5] <= id.tab[,6],]\n# CovZZ2.vec = apply(id.tab,1,function(x)\n# C[x[2],x[4]] * C[x[1],x[3]] + C[x[2],x[3]] * C[x[1],x[4]] + \n# (x[2]==x[4]) * (x[1]==x[3]) *sigma2^2 + \n# (x[2]==x[3]) * (x[1]==x[4]) *sigma2^2 +\n# (C[x[2],x[4]] * (x[1]==x[3]) + C[x[2],x[3]] * (x[1]==x[4]) + \n# C[x[1],x[4]] * (x[2]==x[3]) + C[x[1],x[3]] * (x[2]==x[4])) * sigma2)\n# \n# CovZZ2 <- matrix(NA,nrow=length(Z),ncol=length(Z))\n# CovZZ2[upper.tri(CovZZ2, diag=T)] <- CovZZ2.vec\n# print((proc.time()-time.start)[3])\n \n # Alternative vectorizing\n #time.start <- proc.time()\n id1 <- as.matrix(expand.grid(1:d,1:d,1:d,1:d))\n id2 <- as.matrix(expand.grid(1:d^2,1:d^2))\n id.tab <- as.matrix(cbind(id1,id2))\n id.tab <- id.tab[id.tab[,5] <= id.tab[,6],] # symmetric\n # kronecker delta terms\n # \\delta_{jk},\\delta_{j'k'},\\delta_{jk'},\\delta_{j'k}\n k.del <- cbind(id.tab[,1]==id.tab[,3],id.tab[,2]==id.tab[,4],\n id.tab[,1]==id.tab[,4],id.tab[,2]==id.tab[,3])\n # Cov operator terms\n # C_{jk},C_{j'k'},C_{jk'},C_{j'k}\n c.term <- cbind(C[cbind(id.tab[,1],id.tab[,3])],\n C[cbind(id.tab[,2],id.tab[,4])],\n C[cbind(id.tab[,1],id.tab[,4])],\n C[cbind(id.tab[,2],id.tab[,3])])\n # Form a matrix with pre-calculated terms\n id.tab <- cbind(id.tab,k.del,c.term)\n # CovZZ = C_{jk}C_{j'k'} + C_{jk'}C_{j'k} \n # + (\\delta_{jk}\\delta_{j'k'} + \\delta_{jk'}\\delta_{j'k})\\sigma_2^2 + \n # (C_{jk}\\delta_{j'k'} + C_{jk'}\\delta_{j'k} + C_{j'k}\\delta_{jk'} +\n # C_{j'k'}\\delta_{jk})\\sigma_2\n CovZZ3.vec = id.tab[,12] * id.tab[,11] + id.tab[,14] * id.tab[,13] + \n (id.tab[,8] * id.tab[,7] + id.tab[,10] * id.tab[,9]) *sigma2^2 +\n (id.tab[,12] * id.tab[,7] + id.tab[,14] * id.tab[,9] + \n id.tab[,13] * id.tab[,10] + id.tab[,11] * id.tab[,8]) * sigma2\n # CovZZ.vec = vech(CovZZ)\n CovZZ3 <- matrix(NA,nrow=length(Z),ncol=length(Z))\n CovZZ3[upper.tri(CovZZ3, diag=T)] <- CovZZ3.vec\n #print((proc.time()-time.start)[3])\n \n return (list(CovZZ = CovZZ3,CovZZ.vech = CovZZ3.vec,\n Z.info = Z.info,Z.tri = Z.tri))\n\n}\n", "meta": {"hexsha": "ff1ca602a3298508d3753583acbde35be3b4f4bc", "size": 4375, "ext": "r", "lang": "R", "max_stars_repo_path": "R/covZ.r", "max_stars_repo_name": "ZhuolinSong/Ftesting", "max_stars_repo_head_hexsha": "d0ecb36b44d377ab012a6325220f52110aeaf8d3", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/covZ.r", "max_issues_repo_name": "ZhuolinSong/Ftesting", "max_issues_repo_head_hexsha": "d0ecb36b44d377ab012a6325220f52110aeaf8d3", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/covZ.r", "max_forks_repo_name": "ZhuolinSong/Ftesting", "max_forks_repo_head_hexsha": "d0ecb36b44d377ab012a6325220f52110aeaf8d3", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 38.0434782609, "max_line_length": 77, "alphanum_fraction": 0.4676571429, "num_tokens": 1617, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8289388125473629, "lm_q2_score": 0.6406358479787609, "lm_q1q2_score": 0.531047919098787}} {"text": "\"\r\nThis document provide functions for spatial heterogenetiy analysis\r\n@author: Haoyang Mi\r\n\"\r\n\r\nlibrary(spatstat)\r\nlibrary(dplyr)\r\nlibrary(flexclust)\r\nlibrary(plyr)\r\nlibrary(rjson)\r\nlibrary(spatstat)\r\nlibrary(pracma)\r\n\r\n\r\n\r\nfac2num <- function(x){\r\n \r\n x <- as.numeric(as.character(x))\r\n return(x)\r\n}\r\n\r\nShannonE <- function(types, coreData){\r\n\r\n #types <- ctype_tensor\r\n type.count <- length(types)\r\n\r\n Total <- nrow(coreDat) # total number of cells\r\n\r\n # all int/ext data\r\n ctype_stat_all <- data.frame(matrix(nrow = 0, ncol = 0))\r\n\r\n for (cseq in seq_len(type.count)) { # outer loop, calcualte interior stats\r\n #cseq <- 1\r\n ctype_int <- 0\r\n ctype_ext <- 0\r\n p <- 0 # ratio\r\n\r\n # current cell type\r\n ctype <- types[cseq]\r\n\r\n # coordinates data for the current core, current cell type\r\n\r\n ctype.Dat <- coreDat[coreDat$ctype_no == ctype, c('Xcoord', 'Ycoord')]\r\n\r\n p <- nrow(ctype.Dat)/Total\r\n\r\n # interior score\r\n ctype_int <- mean(as.matrix(dist(ctype.Dat[c('Xcoord', 'Ycoord')])))\r\n\r\n\r\n # other cell types\r\n ctype_other <- types[-cseq]\r\n\r\n ctype_stat <- cbind(p, ctype_int)\r\n\r\n for (cseq_other in ctype_other) {\r\n\r\n ctype_other.Dat <- coreDat[coreDat$ctype_no == cseq_other, c('Xcoord', 'Ycoord')]\r\n\r\n # exterior score\r\n\r\n ctype_ext <- ctype_ext + mean(dist2(ctype.Dat, ctype_other.Dat))\r\n\r\n }\r\n # number of computations = No. cell types - 1\r\n ctype_stat <- cbind(ctype_stat, ctype_ext / (type.count - 1))\r\n\r\n colnames(ctype_stat) <- c('p', 'int', 'ext')\r\n\r\n ctype_stat_all <- rbind(ctype_stat_all ,cbind(ctype_stat, ctype))\r\n\r\n }\r\n\r\n ShannonH <- 0\r\n # combine row data\r\n\r\n\r\n for (dat in seq_len(type.count)) {\r\n p <- as.numeric(as.character(ctype_stat_all[dat, 1]))\r\n d_int <- as.numeric(as.character(ctype_stat_all[dat, 2]))\r\n\r\n d_ext <- as.numeric(as.character(ctype_stat_all[dat, 3]))\r\n d_final <- d_int / d_ext\r\n\r\n if(isTRUE(d_int*d_ext == 0)){\r\n d_final <- 0\r\n }\r\n if(isTRUE(p != 0)){\r\n ShannonH <- -d_final*p*log2(p) + ShannonH\r\n }\r\n }\r\n\r\n return(ShannonH)\r\n}\r\n\r\n\r\n\r\n# convet graph to covariance matrix\r\n#@ input: g -> graph \r\n# null_matrix -> an empty matrix\r\n\r\ngraph2covMatrix <- function(Community_edges, node_type){\r\n \r\n \r\n from_types <- node_type[match(Community_edges$from, node_type$X1),]\r\n to_types <- node_type[match(Community_edges$to, node_type$X1),]\r\n \r\n nt_types <- data.frame(cbind(as.character(from_types$X2), as.character(to_types$X2)))\r\n \r\n # create covariance matrix\r\n covar_Matrix <- matrix(nrow = 17, ncol = 17)\r\n \r\n colnames(covar_Matrix) <- ctype_names\r\n rownames(covar_Matrix) <- ctype_names\r\n \r\n for(row in seq_len(17)){\r\n \r\n # from\r\n row_name <- ctype_names[row]\r\n for (col in seq_len(17)) {\r\n \r\n col_name <- ctype_names[col]\r\n covar_Matrix[row, col] <- nrow(nt_types[nt_types$X1 == row_name & nt_types$X2 == col_name,])+\r\n nrow(nt_types[nt_types$X1 == col_name & nt_types$X2 == row_name,])\r\n \r\n }\r\n }\r\n return(covar_Matrix)\r\n}\r\n\r\nGNN <- function(Community_edges, NodeFeature){\r\n\r\n \r\n \r\n # re-index\r\n NodeFeature$reindex <- seq_len(nrow(NodeFeature)) - 1\r\n edges <- list()\r\n features <- list()\r\n \r\n \r\n ## merge\r\n from_oldNode <- data.frame(Community_edges$from)\r\n colnames(from_oldNode) <- 'id'\r\n to_oldNode <- data.frame(Community_edges$to)\r\n colnames(to_oldNode) <- 'id'\r\n \r\n newNode <- data.frame(cbind(NodeFeature$members, NodeFeature$reindex))\r\n colnames(newNode) <- c('id', 'new_id')\r\n \r\n from_new <- vector()\r\n for(row in seq_len(nrow(from_oldNode))){\r\n \r\n old <- from_oldNode$id[row]\r\n new <- newNode[newNode$id == old, 2]\r\n \r\n from_new <- c(from_new, new)\r\n }\r\n \r\n to_new <- vector()\r\n for(row in seq_len(nrow(to_oldNode))){\r\n \r\n old <- to_oldNode$id[row]\r\n new <- newNode[newNode$id == old, 2]\r\n \r\n to_new <- c(to_new, new)\r\n }\r\n \r\n newEdge <- data.frame(cbind(from_new, to_new))\r\n # edges to list\r\n \r\n for(lid in seq_len(nrow(newEdge))) {\r\n edges[[lid]] <- as.numeric(newEdge[lid,])\r\n }\r\n \r\n F <- NodeFeature[,c(1,68, 69)]\r\n F$ctype_int <- as.character(F$ctype_int)\r\n \r\n # features to list\r\n \r\n for(fid in seq_len(nrow(NodeFeature))) {\r\n features[[fid]] <- F[fid,]$ctype_int \r\n names(features)[fid] <- F[fid, 3]\r\n }\r\n \r\n # combine features and edges to one list\r\n list_final <- list(edges = edges, features = features)\r\n return(list_final)\r\n}\r\n\r\ncor.network <- function(gct_file, allCore){\r\n \r\n #allCore <- NR_core\r\n R_intercor <- data.frame(matrix(nrow = 0, ncol = 0))\r\n \r\n for (core in allCore) {\r\n \r\n #core <- 2\r\n # core - dendrogram\r\n Core_dendro <- gct_file[gct_file$id == core,]\r\n Core_dendro <- Core_dendro[order(Core_dendro$Community_id, decreasing = FALSE),]\r\n \r\n GNN_vec <- read.csv(paste('./Data/Features/', core, '/nci.csv', sep = ''), row.names = 1)\r\n intercor <- outer(1:nrow(GNN_vec),1:nrow(GNN_vec), FUN = Vectorize( function(i,j) cor.test(t(GNN_vec[i,]), t(GNN_vec[j,]), method = 'spearman')$p.value) )\r\n \r\n colnames(intercor) <- Core_dendro$dendrogram_cut\r\n rownames(intercor) <- Core_dendro$dendrogram_cut\r\n diag(intercor) <- -10\r\n \r\n # melt matrix\r\n melt_intercor <- melt(intercor)\r\n melt_intercor <- melt_intercor[!(melt_intercor$value == -10),]\r\n R_intercor <- rbind(R_intercor, melt_intercor)\r\n \r\n }\r\n \r\n Cor_network <- data.frame(matrix(nrow = 0, ncol = 0))\r\n for(row in seq_len(8)){\r\n #row <- 1\r\n #col <- 1\r\n for(col in row:8){\r\n \r\n dat <- R_intercor[R_intercor$Var1 == row & R_intercor$Var2 == col,]\r\n \r\n if(nrow(dat) >= 5){\r\n strength <- nrow(dat[dat$value < 0.05,])/nrow(dat)\r\n \r\n } else {\r\n strength <- 0\r\n }\r\n #hist(dat$value)\r\n \r\n #value <- p.adjust(dat$value, method = 'fdr')\r\n \r\n # normalized communication strength\r\n #strength <- length(value[value < 0.05])/length(value)\r\n Cor_network <- rbind(Cor_network, cbind(row, col, strength))\r\n \r\n }\r\n }\r\n Cor_network <- Cor_network[complete.cases(Cor_network$strength),]\r\n \r\n \r\n Cor_network <- Cor_network[Cor_network$strength > 0,]\r\n \r\n #test <- graph_from_data_frame(Cor_network[,1:2])\r\n #plot(test) \r\n \r\n return(Cor_network)\r\n}\r\n\r\n\r\n#######################\r\n# Kcross function #####\r\n#######################\r\n\r\n\r\n\r\nbivarAnalysis.Kcross <- function(type1, type2){\r\n \r\n #type1 <- typeA\r\n #type2 <- typeB\r\n \r\n colnames(type1) <- c('x', 'y')\r\n \r\n colnames(type2) <- c('x', 'y')\r\n \r\n if(nrow(type1)*nrow(type2) != 0){\r\n \r\n # read pts dat\r\n\r\n type1$attr <- 'ctypeA'\r\n \r\n type2$attr <- 'ctypeB'\r\n \r\n # create multitype df\r\n pts_OI <- rbind(type1, type2)\r\n \r\n # define the type\r\n species <- factor(pts_OI$attr)\r\n \r\n # create multitype ppp\r\n\r\n \r\n # check if empty \r\n ppp1 <- ppp(type1$x, type1$y, owin(c(0,800), c(0, 800)))\r\n ppp2 <- ppp(type2$x, type2$y, owin(c(0,800), c(0, 800)))\r\n \r\n # prevent NA \r\n \r\n if(is.empty(ppp1) == 'FALSE' & is.empty(ppp2) == 'FALSE'){\r\n \r\n \r\n \r\n \r\n multitype_ppp <- ppp(pts_OI$x, pts_OI$y, marks = species, owin(c(0, 800), c(0, 800)))\r\n K.cross <- data.frame(Kcross(multitype_ppp, i = 'ctypeA', j = 'ctypeB', r = seq(0,20,0.1), correction = 'Ripley'))\r\n \r\n #plot(Gihc)\r\n # relocat DF\r\n \r\n K.cross <- K.cross[complete.cases(K.cross),]\r\n \r\n \r\n \r\n # calculate the area (positive - negative ) \r\n K.cross$rs <- K.cross$iso - K.cross$theo\r\n \r\n A.to.B.diff.area <- trapz(K.cross$r, K.cross$iso) \r\n \r\n \r\n # j to i\r\n \r\n K.cross <- data.frame(Kcross(multitype_ppp, i = 'ctypeB', j = 'ctypeA', r = seq(0,20,0.1), correction = 'Ripley'))\r\n \r\n K.cross <- K.cross[complete.cases(K.cross),]\r\n \r\n \r\n \r\n K.cross$iso <- K.cross$iso - K.cross$theo\r\n K.cross <- K.cross[complete.cases(K.cross),]\r\n B.to.A.diff.area <- trapz(K.cross$r, K.cross$iso) \r\n \r\n } else {\r\n \r\n A.to.B.diff.area <- NA\r\n B.to.A.diff.area <- NA\r\n \r\n }\r\n } else {\r\n A.to.B.diff.area <- NA\r\n B.to.A.diff.area <- NA\r\n }\r\n return(list(A.to.B.diff.area, B.to.A.diff.area))\r\n}\r\n\r\n\r\n\r\n#----------------------------------------------------#\r\n# generate poisson distribution ppp from a given ppp #\r\n#----------------------------------------------------#\r\n\r\n\r\npoisp <- function(pos_Target, nsim){\r\n \r\n #pos_Target <- tPos\r\n \r\n n <- nrow(pos_Target)\r\n \r\n simpp <- runifpoint(n = n, win = owin(c(0, 800), c(0, 800)), nsim = nsim)\r\n \r\n # return the result \r\n return(simpp)\r\n}\r\n\r\n\r\n#-----------------------------#\r\n# Find number of interactions #\r\n#-----------------------------#\r\n\r\nnnIntrxn <- function(sDF, simDF){ # sDF: the coordiantes for the source point pattern\r\n \r\n #sDF <- sPos\r\n if(nrow(sDF) == 0 | nrow(simDF) == 0){\r\n simInt <- 0\r\n } else{\r\n dist <- nn2(simDF, query = sDF, k = nrow(simDF), treetype = 'kd', searchtype = 'radius', radius = 20)\r\n \r\n nn.idx <- data.frame(dist$nn.idx)\r\n \r\n nn.idx$source <- seq_len(nrow(sDF))\r\n \r\n nn.merge <- melt(nn.idx, id.vars = 'source')\r\n \r\n # clear non-valid rows\r\n nn.merge <- nn.merge[nn.merge$value != 0, -2]\r\n \r\n \r\n # remove itself\r\n \r\n \r\n nn.merge$diff <- nn.merge$source - nn.merge$value\r\n nn.merge <- nn.merge[nn.merge$diff != 0, -3]\r\n \r\n simInt <- nrow(nn.merge)\r\n }\r\n\r\n \r\n # replace by real cell type\r\n return(simInt)\r\n}\r\n\r\n\r\n\r\n#-----------------------------#\r\n# cor.mtest #\r\n#-----------------------------#\r\n\r\n\r\ncor.mtest <- function(mat, ...) {\r\n mat <- as.matrix(mat)\r\n n <- ncol(mat)\r\n p.mat<- matrix(NA, n, n)\r\n diag(p.mat) <- 0\r\n for (i in 1:(n - 1)) {\r\n for (j in (i + 1):n) {\r\n tmp <- cor.test(mat[, i], mat[, j], ...)\r\n p.mat[i, j] <- p.mat[j, i] <- tmp$p.value\r\n }\r\n }\r\n colnames(p.mat) <- rownames(p.mat) <- colnames(mat)\r\n}", "meta": {"hexsha": "ff771561f75242f8bdd72c2e5c1b8ee1e0d25795", "size": 10157, "ext": "r", "lang": "R", "max_stars_repo_path": "Functions.r", "max_stars_repo_name": "Shawnmhy/HCC-IMC-processing-pipeline", "max_stars_repo_head_hexsha": "fd3e347eabf194f63c2725d6d3bd9db9a6d52078", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Functions.r", "max_issues_repo_name": "Shawnmhy/HCC-IMC-processing-pipeline", "max_issues_repo_head_hexsha": "fd3e347eabf194f63c2725d6d3bd9db9a6d52078", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Functions.r", "max_forks_repo_name": "Shawnmhy/HCC-IMC-processing-pipeline", "max_forks_repo_head_hexsha": "fd3e347eabf194f63c2725d6d3bd9db9a6d52078", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-08-05T01:15:59.000Z", "max_forks_repo_forks_event_max_datetime": "2021-08-05T01:15:59.000Z", "avg_line_length": 24.7128953771, "max_line_length": 159, "alphanum_fraction": 0.5437629221, "num_tokens": 2952, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7853085808877581, "lm_q2_score": 0.6757646010190476, "lm_q1q2_score": 0.5306837398404503}} {"text": "# fit_robust_linear_model.r\n#\n# Copyright (c) 2020 VIB (Belgium) & Babraham Institute (United Kingdom)\n#\n# Software written by Carlos P. Roca, as research funded by the European Union.\n#\n# This software may be modified and distributed under the terms of the MIT\n# license. See the LICENSE file for details.\n\n\n# Returns a matrix by rows, with the intercept and p-value, and coefficient and\n# p-value, of a robust linear model fitted to the input data.\n#\n# Reverts to a standard linear model in case of no convergence.\n\nfit.robust.linear.model <- function( x.data, y.data, x.name, y.name, asp )\n{\n xy.data <- data.frame( x = x.data, y = y.data )\n\n xy.model <- rlm( y ~ x, xy.data, maxit = asp$rlm.iter.max )\n\n if ( xy.model$converged )\n {\n xy.coef <- xy.model$coefficients\n xy.t <- summary( xy.model )$coefficients[ , 3 ]\n xy.df <- summary( xy.model )$df[ 2 ]\n xy.pval <- 2*( pt( abs( xy.t ), xy.df, lower.tail = FALSE ) )\n }\n else\n {\n cat( sprintf( \"WARNING: rlm of %s ~ %s did not converge - then using ols\\n\",\n y.name, x.name ), file = stderr() )\n\n xy.model <- lm( y ~ x, xy.data )\n\n xy.coef <- xy.model$coefficients\n xy.pval <- summary( xy.model )$coefficients[ , 4 ]\n }\n\n res <- cbind( xy.coef, xy.pval )\n dimnames( res ) <- NULL\n res\n}\n\n", "meta": {"hexsha": "019a9fd604172941707d338b9648294a578b5ab1", "size": 1341, "ext": "r", "lang": "R", "max_stars_repo_path": "R/fit_robust_linear_model.r", "max_stars_repo_name": "DillonHammill/autospill", "max_stars_repo_head_hexsha": "2cd601ea9481688497f0800e7c873bbf71ba5f1e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 10, "max_stars_repo_stars_event_min_datetime": "2020-08-07T21:48:31.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-12T03:00:59.000Z", "max_issues_repo_path": "R/fit_robust_linear_model.r", "max_issues_repo_name": "DillonHammill/autospill", "max_issues_repo_head_hexsha": "2cd601ea9481688497f0800e7c873bbf71ba5f1e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 4, "max_issues_repo_issues_event_min_datetime": "2020-09-10T08:08:01.000Z", "max_issues_repo_issues_event_max_datetime": "2021-06-29T23:41:00.000Z", "max_forks_repo_path": "R/fit_robust_linear_model.r", "max_forks_repo_name": "DillonHammill/autospill", "max_forks_repo_head_hexsha": "2cd601ea9481688497f0800e7c873bbf71ba5f1e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 9, "max_forks_repo_forks_event_min_datetime": "2020-09-05T14:15:12.000Z", "max_forks_repo_forks_event_max_datetime": "2021-10-12T14:36:42.000Z", "avg_line_length": 29.8, "max_line_length": 84, "alphanum_fraction": 0.6122296793, "num_tokens": 386, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7690802370707283, "lm_q2_score": 0.6893056295505784, "lm_q1q2_score": 0.5301313369889464}} {"text": "# ST3068/ST2054 -- tutorial -- 21 October 2016\n# Summer Exam 2009 in the course ST2054\n# Question 2 \n\nsample_dist_1 = function() { \n n = sample(x = c(0,1), 1, replace = FALSE, prob = c(0.2, 0.8));\n return(n)\n}\n\nsample_dist_2 = function() { \n n = sample(x = c(0,1), 1, replace = FALSE, prob = c(0.3, 0.7));\n return(n)\n}\n\nsample_dist_3 = function() { \n n = sample(x = c(0,1), 1, replace = FALSE, prob = c(0.4, 0.6));\n return(n)\n}\n\nx_count = 0;\nrun_count = 10000;\n\nfor (i in 1:run_count) {\n \n player_1 = sample_dist_1();\n player_2 = sample_dist_2();\n player_3 = sample_dist_3();\n\n if ((player_1 == 1) || (player_2 == 1) || (player_3 == 1)) {\n if (player_3 == 1){x_count = x_count + 1}\n }\n}\nprint (x_count/run_count)\n", "meta": {"hexsha": "e77e0b612ca94f6ad5301fd384c3cda9340e699d", "size": 729, "ext": "r", "lang": "R", "max_stars_repo_path": "Simulation_SummerExam2009_Q2.r", "max_stars_repo_name": "mashenkaulm/ST2054-ST3068", "max_stars_repo_head_hexsha": "9e8fb604aca623ed908c9c8679bd4c2dd018dfc9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Simulation_SummerExam2009_Q2.r", "max_issues_repo_name": "mashenkaulm/ST2054-ST3068", "max_issues_repo_head_hexsha": "9e8fb604aca623ed908c9c8679bd4c2dd018dfc9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Simulation_SummerExam2009_Q2.r", "max_forks_repo_name": "mashenkaulm/ST2054-ST3068", "max_forks_repo_head_hexsha": "9e8fb604aca623ed908c9c8679bd4c2dd018dfc9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 21.4411764706, "max_line_length": 65, "alphanum_fraction": 0.5939643347, "num_tokens": 287, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7690802370707281, "lm_q2_score": 0.6893056231680122, "lm_q1q2_score": 0.5301313320802409}} {"text": "nextState <- function(x, y, wx, wy, inp) {\n cmd <- substring(inp, 1, 1)\n val <- strtoi(substring(inp, 2, nchar(inp)))\n\n result <- switch(cmd,\n \"N\" = move_w(x, y, wx, wy, \"N\", val),\n \"S\" = move_w(x, y, wx, wy, \"S\", val),\n \"E\" = move_w(x, y, wx, wy, \"E\", val),\n \"W\" = move_w(x, y, wx, wy, \"W\", val),\n \"F\" = move_f(x, y, wx, wy, val),\n \"R\" = rotate_v(x, y, wx, wy, -val),\n \"L\" = rotate_v(x, y, wx, wy, val))\n\n result\n}\n\nmove_w <- function(x, y, wx, wy, moveDir, val) {\n switch(moveDir,\n \"N\" = c(x, y, wx, wy + val),\n \"S\" = c(x, y, wx, wy - val),\n \"E\" = c(x, y, wx + val, wy),\n \"W\" = c(x, y, wx - val, wy))\n}\n\nmove_f <- function(x, y, wx, wy, count) {\n c(x + wx * count, y + wy * count, wx, wy)\n}\n\nrotate_v <- function(x, y, wx, wy, d) {\n r <- d * (pi / 180)\n\n c(x, y, wx * cos(r) - wy * sin(r), wx * sin(r) + wy * cos(r))\n}\n\nfileName <- \"input.txt\"\n\nx <- 0\ny <- 0\n\nwx <- 10\nwy <- 1\n\ninp <- file(fileName, \"r\")\n\nfor (line in readLines(inp)) {\n s <- nextState(x, y, wx, wy, line)\n\n x <- round(as.double(s[1]))\n y <- round(as.double(s[2]))\n wx <- round(as.double(s[3]))\n wy <- round(as.double(s[4]))\n}\n\nclose(inp)\n\nprint(abs(x) + abs(y))\n", "meta": {"hexsha": "b531a03e96eca679fb35c290ac079df0c30b1b21", "size": 1248, "ext": "r", "lang": "R", "max_stars_repo_path": "12/second.r", "max_stars_repo_name": "madetara/advent2020", "max_stars_repo_head_hexsha": "39492ef746baa8e49de880cb2604b5b67a5792ac", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "12/second.r", "max_issues_repo_name": "madetara/advent2020", "max_issues_repo_head_hexsha": "39492ef746baa8e49de880cb2604b5b67a5792ac", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "12/second.r", "max_forks_repo_name": "madetara/advent2020", "max_forks_repo_head_hexsha": "39492ef746baa8e49de880cb2604b5b67a5792ac", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 21.8947368421, "max_line_length": 65, "alphanum_fraction": 0.4607371795, "num_tokens": 471, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8221891392358015, "lm_q2_score": 0.6442250928250375, "lm_q1q2_score": 0.5296748745439219}} {"text": "\n# Simulate VB tag-recapture, otolith and length-frequency growth data\n# =====================================================================\n\n# Specify VB parameters common to all datasets(noting that Linf ~ Norm(mu.Linf, sigma.Linf) )\nk <- 0.2\na0 <- 0\nmu.Linf <- 100\nsigma.Linf <- 5\n\n# a)tag-recapture data:\n# ----------------------\nset.seed(1)\nn <- 500\n\n# generate log-normal release ages (relative to a0)\npar1.A <- 0.5\npar2.A <- 0.3\nA <- rlnorm(n,par1.A,par2.A)\n\n# generate times at liberty assuming gamma distribution\ndt <- rgamma(n,3,.5)\n\n# generate release lengths and recapture lengths assuming random Linf and measurement error\nLinf <- rnorm(n,mu.Linf,sigma.Linf)\nsig.tag <- 5\nL1 <- Linf*(1-exp(-k*A)) + rnorm(n,0,sig.tag)\nL2 <- Linf*(1-exp(-k*(A+dt))) + rnorm(n,0,sig.tag)\n\n# Note that the code allows for a different measurement error for scientist-measured recapture lengths\n# versus fisherman-measured recapture lengths, so need to specify whether L2 was measured by\n# a scientist (=0) or a fisherman (=1). Here I'll specify all as scientist (0) since I generated L2\n# assuming the same measurement error for all observations.\nL2measurer <- rep(0,n)\n\ntagdat <- cbind(L1,L2,rep(0,n),dt,L2measurer)\nplot(c(A,A+dt),c(L1,L2),xlab=\"age\",ylab=\"length\")\npoints(A,L1,col=2)\n\n\n# b) otolith data:\n# ----------------------\nset.seed(1)\nn.oto <- 200\n\n# generate random uniform ages\nage <- runif(n.oto,.5,20)\n# generate corresponding lengths with measurement error\nLinf <- rnorm(n.oto,mu.Linf,sigma.Linf)\nsig.oto <- 5\nlen <- Linf*(1-exp(-k*(age-a0))) + rnorm(n.oto,0,sig.oto)\n\notodat <- cbind(age,len)\npoints(age,len,pch=2,col=3)\n\n\n# c) length-frequency data:\n# -------------------------\nset.seed(1)\n\n# Note: the LF component assumes that a model decomposition has already been applied to the\n# raw data, such that the dataset input to the growth analysis is the ages and lengths corresponding to\n# the modes in the LF data, along with the standard error estimates for each modal length estimate\n\n# generate ages and lengths corresponding to modes in the LF data, assuming we have monthly LF data for\n# fish of ages 1-5, and we have this data for 5 years\nmode.age <- rep(seq(1,5,1/12),times=5)\nn.lf <- length(mode.age)\n# recall we are not including a random Linf parameter for the LF data\nsig.lf <- 5\nmode.len <- mu.Linf*(1-exp(-k*(mode.age-a0))) + rnorm(n.lf,0,sig.lf)\n\n# Specify standard error estimates corresponding to each of the modal length estimates\nse.mode <- rlnorm(n.lf,.01,.5)\n\nlfdat <- cbind(mode.age,mode.len,se.mode)\npoints(mode.age,mode.len,pch=3,col=4)\n\n\n\n# Source the R code with the likelihood and a range of growth functions:\n# -----------------------------------------------------------------------\nsource(\"/Users/stephenscherrer/Google Drive/Weng Lab/Data/Bottomfish/Okamoto's 1990s Mark Recapture Data/src/Laslett Functions/joint_lkhd.r\")\nsource(\"/Users/stephenscherrer/Google Drive/Weng Lab/Data/Bottomfish/Okamoto's 1990s Mark Recapture Data/src/Laslett Functions/growth_functions.r\")\n\n\n\n# Fit a VB model WITHOUT seasonal growth:\n# -----------------------------------------\ngrowth.ssnl.f<- growthvb.f\nnpf <- 1 #number of parameters passed to growth.ssnl.f (in this case k)\nnpA <- 2 #number of parameters in distribution of release ages for tag model\n\n# specify starting parameters, as well as upper and lower bounds\n# mu.L, sig.L, k, mu.A, sig.A, sig.sci, sig.f, a0, sig.oto, sig.lf\np0 <- c( 70, 5.0, .10, 1.0, .10, 5.0, 0, 0, 10.0, 10.0)\nlb <- c( 50, 1.0, .05, .1, .05, 1.0, 0, -2, 0.0, 0.0)\nub <- c( 100, 15.0, .40, 1.5, .50, 10.0, 0, 1, 15.0, 15.0)\nfit.vb <- nlminb(p0,joint.logl.f,lower=lb,upper=ub,npf=npf,npA=npA,tagdat=tagdat,otodat=otodat,lfdat=lfdatz)\nfit.vb\n\n\n# Fit a VB model *WITH* seasonal growth:\n# -----------------------------------------\ngrowth.ssnl.f<- growthvb.ssnl.f\nnpf <- 3 #number of parameters passed to growth.ssnl.f (in this case k, u, w)\nnpA <- 2 #number of parameters in distribution of release ages for tag model\n\n# specify starting parameters, as well as upper and lower bounds\n# mu.L, sig.L, k, u, w, mu.A, sig.A, sig.sci, sig.f, a0, sig.oto, sig.lf\np0 <- c( 110, 5.0, .10, .1, 0, 1.0, .10, 5.0, 0, 0, 10.0, 10.0)\nlb <- c( 70, 1.0, .05, 0, -.5, .1, .05, 1.0, 0, -2, 0.0, 0.0)\nub <- c( 130, 15.0, .40, 1, .5, 1.5, .50, 10.0, 0, 1, 15.0, 15.0)\nfit.ssnl.vb <- nlminb(p0,joint.logl.f,lower=lb,upper=ub,npf=npf,npA=npA,tagdat=tagdat,otodat=otodat,lfdat=lfdat)\nfit.ssnl.vb\n\n\n\n\n", "meta": {"hexsha": "2fbad14f45eaf8c395addb3ed4b2db13e2343d88", "size": 4578, "ext": "r", "lang": "R", "max_stars_repo_path": "Analysis/src/Laslett Functions/Example_for_joint_model.r", "max_stars_repo_name": "stevescherrer/Ch4-Opakapaka-Growth", "max_stars_repo_head_hexsha": "b8c0d3881be889d8197bb0f9251f011ed98850f1", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Analysis/src/Laslett Functions/Example_for_joint_model.r", "max_issues_repo_name": "stevescherrer/Ch4-Opakapaka-Growth", "max_issues_repo_head_hexsha": "b8c0d3881be889d8197bb0f9251f011ed98850f1", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Analysis/src/Laslett Functions/Example_for_joint_model.r", "max_forks_repo_name": "stevescherrer/Ch4-Opakapaka-Growth", "max_forks_repo_head_hexsha": "b8c0d3881be889d8197bb0f9251f011ed98850f1", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 38.15, "max_line_length": 147, "alphanum_fraction": 0.629969419, "num_tokens": 1545, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.84594244507642, "lm_q2_score": 0.6261241842048092, "lm_q1q2_score": 0.529665023307695}} {"text": "#' @title calcHMPwr\n#'\n#' @description Calculate instantaneous main engine power (kW) using the Holtrop & Mennen method.\n#'\n#' @param totalInstalledPwr Total installed main engine power (vector of numericals, kW) (maximum\n#' continuous rated power)\n#' @param shipSpeed Ship actual speed (vector of numericals, m/s) (see \n#' \\code{\\link{calcSpeedUnitConversion}})\n#' @param actualDraft Actual draft (vector of numericals, m)\n#' @param maxDraft Maximum summer load line draft (vector of numericals, m)\n#' @param shipType Ship type (vector of strings, see \\code{\\link{calcShipType}}). \n#' Must align with \\code{tankerBulkCarrierShipTypes}, \\code{tugShipTypes}, \n#' \\code{roroPaxShipTypes}, \\code{gCargoShipTypes},\n#' \\code{containerShipTypes} groupings\n#' @param lwl Waterline length (vector of numericals, m) (see \\code{\\link{calclwl}})\n#' @param breadth Moulded breadth (vector of numericals, m)\n#' @param maxDisplacement Maximum ship displacement (vector of numericals, m^3)\n#' @param Cb Maximum block coefficient (vector of numericals, dimensionless) (see \n#' \\code{\\link{calcCb}})\n#' @param nProp Number of propellers (vector of numericals, see \\code{\\link{calcPropNum}})\n#' @param serviceMargin A service margin to account for weather and sea effects:\n#' \\itemize{\\item At-sea operations = 15 (Default) \\item Coastal operations = 10} Can \n#' supply either a vector of numericals, a single number, or rely on the default\n#' @param shaftEff Shaft efficiency (dimensionless). Default = 0.98.\n#' Ratio of power delivered to the propeller and the brake power delivered by\n#' the engine. Can supply either a vector of numericals, a single number, or rely on the default.\n#' @param relRotationEff Relative rotational efficiency (dimensionless).\n#' Default = 1. Accounts for effect of rotational flow of water around propeller.\n#' Can supply either a vector of numericals, a single number, or rely on the default\n#' @param seawaterTemp Sea water temperature. Default = 15 (degrees Celsius). Can \n#' supply either a vector of numericals, a single number, or rely on the default\n#' @param seawaterDensity Sea water density. Default = 1.025 (g/cm^3). Can \n#' supply either a vector of numericals, a single number, or rely on the default\n#' @param pwrUpperBoundPercent Percent of total installed power at which\n#' required power is capped. Default = 1, which indicates required power cannot\n#' exceed \\code{totalInstalledPwr}. Can supply either a vector of numericals, a \n#' single number, or rely on the default\n#' @param pwrLowerBoundPercent Percent of total installed power to act as lower\n#' bound for required power. Default = 0.02, which indicates required power\n#' cannot go below 2\\% of \\code{totalInstalledPwr}. Can supply either a vector\n#' of numericals, a single number, or rely on the default\n#' @param CmEquationType Type of equation to estimate the midship section\n#' coefficient (see \\code{\\link{calcCm}}): \\itemize{\n#' \\item\"kristensen (Default)\"\n#' \\item\"benford\"\n#' \\item\"schneekluth\"}\n#' This argument is not vectorized, as it takes only a single string\n#' @param tankerBulkCarrierShipTypes Ship types specified in input\n#' \\code{shipTypes} to be modeled as tankers and bulk carriers\n#' @param tugShipTypes Ship types specified in input \\code{shipTypes} to be\n#' modeled as tugs (vector of strings)\n#' @param roroPaxShipTypes Ship types specified in input \\code{shipTypes} to be\n#' modeled as RORO and passenger ships (vector of strings)\n#' @param gCargoShipTypes Ship types specified in input \\code{shipTypes} to be\n#' modeled as general cargo (vector of strings)\n#' @param containerShipTypes Ship types specified in input \\code{shipTypes} to\n#' be modeled as container ships (vector of strings)\n#' @param Cstern Afterbody form coefficient: \\itemize{\\item V-shaped Hull = -10\n#' \\item U-Shaped Hull = 10\n#' \\item Normal Hull = 0 (default)}\n#' Can supply either a vector of numericals, a single number, or rely on the default\n#' @param forwardDraft Forward draft (deviation from actual draft indicates trim) \n#' (vector of numericals, m)\n#' @param aftDraft Aft draft (deviation from actual draft indicates trim)\n#' (vector of numericals, m)\n#' @param appendagesList List of appendages on ship (vector of strings) \\itemize{\n#' \\item\"rudder behind skeg\"\n#' \\item\"rudder behind stern\"\n#' \\item\"twin-screw balance rudders\"\n#' \\item\"shaft brackets\"\n#' \\item\"skeg\"\n#' \\item\"strut bossings\"\n#' \\item\"hull bossings\"\n#' \\item\"shafts\"\n#' \\item\"stabilizer fins\"\n#' \\item\"dome\"\n#' \\item\"bilge keels\"}\n#' @param wettedAppSAList List of wetted surface areas corresponding to list of\n#' appendages (vector of numericals, m^2)\n#'\n#' @details\n#' Primary method from Holtrop & Mennen (1982). Updated equations for high speed\n#' operations (Froude number > 0.55) from Holtrop & Mennen (1984). Estimation of\n#' some inputs use methodology from Rakke (2016).\n#'\n#' This method this requires ship types to be grouped. Use the\n#' \\code{tankerBulkCarrierShipTypes}, \\code{tugShipTypes}, \\code{roroPaxShipTypes},\n#' \\code{gCargoShipTypes}, \\code{containerShipTypes} grouping parameters to\n#' provide these ship type groupings. Any ship types not included in these groupings\n#' will be considered as miscellaneous vessels.\n#'\n#' Ship speed and actual draft are typically obtained from sources such as AIS\n#' messages or ship records.\n#'\n#' @return power (vector of numericals, kW)\n#'\n#' @references\n#'Holtrop, J. and Mennen, G. G. J. 1982. \"An approximate power prediction\n#'method.\" International Shipbuilding Progress 29.\n#'\n#'Holtrop, J. and Mennen, G. G. J. 1984. \"A Statistical Re-Analysis of Resistance\n#'and Propulsion Data'.\n#'\n#'\\href{http://hdl.handle.net/11250/2410741}{Rakke, S. G. 2016. \"Ship Emissions\n#'Calculation from AIS.\" NTNU.}\n#'\n#'@seealso \\itemize{\n#'\\item \\code{\\link{calcSpeedUnitConversion}}\n#'\\item \\code{\\link{calclwl}}\n#'\\item \\code{\\link{calcCb}}\n#'\\item \\code{\\link{calcPropNum}}\n#'\\item \\code{\\link{calcCm}}\n#'\\item \\code{vignette(\"OverviewOfPowerModels\", package=\"ShipPowerModel\")}\n#'\\item \\code{vignette(\"HoltropMennen.Example\", package=\"ShipPowerModel\")}\n#'}\n#'\n#' @family Holtrop-Mennen Calculations\n#'\n#' @examples\n#' calcHMPwr(\n#' totalInstalledPwr=rep(9363,2),,\n#' shipSpeed=seq(0,1,1),\n#' actualDraft=rep(12.48,2),\n#' maxDraft=rep(13.57,2),\n#' shipType=rep(\"bulk.carrier\",2),\n#' lwl=rep(218.75,2),\n#' breadth=rep(32,2),\n#' maxDisplacement=rep(80097,2),\n#' Cb=rep(0.8099003,2),\n#' nProp=rep(1,2),\n#' serviceMargin=15,\n#' shaftEff=0.98,\n#' relRotationEff=1,\n#' seawaterTemp=15,\n#' seawaterDensity=1.025,\n#' Cstern=0,\n#' CmEquationType=\"kristensen\",\n#' forwardDraft=rep(13.57,2),\n#' aftDraft=rep(13.57,2),\n#' appendagesList=c(\"\"),\n#' wettedAppSAList=NA\n#' )\n#'\n#' @export\n\n\n calcHMPwr<- function(\n totalInstalledPwr,\n shipSpeed,\n actualDraft,\n maxDraft,\n shipType,\n lwl,\n breadth,\n maxDisplacement,\n Cb,\n nProp,\n serviceMargin=15,\n shaftEff=0.98,\n relRotationEff=1,\n seawaterTemp=15,\n seawaterDensity=1.025,\n pwrUpperBoundPercent=1,\n pwrLowerBoundPercent=0.02,\n CmEquationType=\"kristensen\",\n tankerBulkCarrierShipTypes=c(\"tanker\",\"chemical.tanker\",\"liquified.gas.tanker\",\"oil.tanker\",\"other.tanker\",\"bulk.carrier\"),\n tugShipTypes=c(\"service.tug\",\"tug\"),\n roroPaxShipTypes=c(\"passenger\",\"ferry.pax\",\"ferry.ro.pax\",\"cruise\",\"cruise.ed\",\"yacht\",\"ro.ro\"),\n gCargoShipTypes=c(\"general.cargo\"),\n containerShipTypes=c(\"container.ship\"),\n Cstern=0,\n forwardDraft=NULL,\n aftDraft=NULL,\n appendagesList=c(\"\"),\n wettedAppSAList=NA\n ){\n#===========================================\nif(is.null(forwardDraft)){forwardDraft<-maxDraft}\nif(is.null(aftDraft)){aftDraft<-maxDraft}\n\n#Inputs\n Cbw<-calcCbw(Cb, actualDraft, maxDraft)\n Cm<-calcCm(shipType,Cbw,maxDraft,actualDraft,CmEquationType,tankerBulkCarrierShipTypes,tugShipTypes,roroPaxShipTypes)\n Cp<-calcCp(Cm, Cbw,shipType, bounds=\"holtrop mennen\", roroPaxContainerShipTypes=union(roroPaxShipTypes,containerShipTypes),gCargoShipTypes,tankerBulkCarrierShipTypes)\n Cwp<-calcCwp(Cbw, CwpEquationType=\"kristensen\")\n froudeNum<-calcFroudeNum(shipSpeed, lwl)\n At<-calcAt(Cm, breadth, maxDraft)\n hb<-calchb(maxDraft)\n Abt<-calcAbt(Cm, breadth, maxDraft)\n propDiam<-calcPropDia(shipType, maxDraft,tankerBulkCarrierGCargoShipTypes=union(tankerBulkCarrierShipTypes,gCargoShipTypes),containerShipTypes)\n lcb<-calclcb(lwl)\n#===========================================\nCf<-calcCf(shipSpeed,\n lwl,\n seawaterTemp,\n seawaterDensity)\n#=============================================\nRapp<-calcHMAppendageRes(shipSpeed,\n Cf,\n appendagesList,\n wettedAppSAList,\n seawaterDensity)\n#=============================================\nRw<-calcHMWaveMakingRes(lwl,\n breadth,\n Cp,\n Cwp,\n Cm,\n maxDisplacement,\n maxDraft,\n froudeNum,\n At,\n hb,\n Abt,\n seawaterDensity,\n forwardDraft,\n lcb)\n#========================================\n\nFormFactor<-calcHMFormFactor(maxDraft,\n lwl,\n breadth,\n maxDisplacement,\n Cp,\n Cstern,\n lcb)\n#==========================================\n\nwettedSA<-calcHMWettedSA(lwl,\n actualDraft,\n breadth,\n Cm,\n Cbw,\n Cwp,\n Abt)\n#==============================================\nRtr<-calcHMImmersedTransomRes(shipSpeed,\n breadth,\n Cwp,\n maxDraft,\n At,\n seawaterDensity)\n#==============================================\nRb<-calcHMBulbousBowRes(shipSpeed,\n maxDraft,\n forwardDraft,\n Abt,\n hb,\n seawaterDensity)\n#=============================================================\nCa<-calcHMCa(maxDraft,\n lwl,\n Cbw,\n breadth,\n forwardDraft,\n Abt,\n hb)\n#=============================================================\nw<-calcHMWakeFraction(breadth,\n wettedSA,\n maxDraft,\n nProp,\n lwl,\n propDiam,\n FormFactor,\n Cf,\n Ca,\n Cbw,\n Cp,\n Cm,\n aftDraft,\n Cstern,\n lcb)\n\nt<-calcHMThrustFactor(breadth,\n lwl,\n maxDraft,\n maxDisplacement,\n nProp,\n Cp,\n propDiam,\n Cbw,\n Cstern,\n lcb,\n seawaterDensity)\n#=============================================================\n\nR<- calcHMTotalRes(Rapp,Rw,Rb,Rtr,seawaterDensity,wettedSA,shipSpeed,\n Cf,FormFactor,Ca,serviceMargin)\n\n#=============================================================\nno<- calcOpenWaterEff(R,t,nProp,w,propDiam,shipSpeed)\n\n#=============================================================\npower<-calcResistanceShipPwr(R,shipSpeed, (1-t)/(1-w), no,totalInstalledPwr, shaftEff,relRotationEff, pwrUpperBoundPercent, pwrLowerBoundPercent)\n\n\n return(power)\n}\n", "meta": {"hexsha": "c217e66a336a8c56f37d7cf5195a9bed93997ff5", "size": 12025, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcHMPwr.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcHMPwr.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcHMPwr.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 40.0833333333, "max_line_length": 170, "alphanum_fraction": 0.5925987526, "num_tokens": 2978, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8558511432905481, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5295839416654954}} {"text": "subroutine dirseg(dirsgs,ndir,nadj,madj,npd,x,y,ntot,rw,eps,nerror)\n\n# Output the endpoints of the segments of boundaries of Dirichlet\n# tiles. (Do it economically; each such segment once and only once.)\n# Called by master.\n\nimplicit double precision(a-h,o-z)\nlogical collin, adjace, intfnd, bptab, bptcd, goferit, rwu\ndimension nadj(-3:ntot,0:madj), x(-3:ntot), y(-3:ntot)\ndimension dirsgs(10,ndir), rw(4)\n\nnerror = -1\n\n# Add in some dummy corner points, outside the actual window.\n# Far enough out so that no resulting tile boundaries intersect the\n# window.\n\n# Note that these dummy corners are needed by the routine `dirout'\n# but will screw things up for `delseg' and `delout'. Therefore\n# this routine (`dirseg') must be called ***before*** dirout, and\n# ***after*** delseg and delout.\n\n# Dig out the corners of the rectangular window.\nxmin = rw(1)\nxmax = rw(2)\nymin = rw(3)\nymax = rw(4)\n\na = xmax-xmin\nb = ymax-ymin\nc = sqrt(a*a+b*b)\n\nnpd = ntot-4\nnstt = npd+1\ni = nstt\nx(i) = xmin-c\ny(i) = ymin-c\ni = i+1\nx(i) = xmax+c\ny(i) = ymin-c\ni = i+1\nx(i) = xmax+c\ny(i) = ymax+c\ni = i+1\nx(i) = xmin-c\ny(i) = ymax+c\n\ndo j = nstt,ntot {\n\tcall addpt(j,nadj,madj,x,y,ntot,eps,nerror)\n\tif(nerror > 0) return\n}\n\n# Put the segments into the array dirsgs.\n\n# For each distinct pair of (genuine) data points, find out if they are\n# adjacent. If so, find the circumcentres of the triangles lying on each\n# side of the segment joining them.\nkseg = 0\ndo i = 2,npd {\n do j = 1,i-1 {\n call adjchk(i,j,adjace,nadj,madj,ntot,nerror)\n\t\tif(nerror > 0) return\n if(adjace) {\n call pred(k,i,j,nadj,madj,ntot,nerror)\n\t\t\tif(nerror > 0) return\n call circen(i,k,j,a,b,x,y,ntot,eps,collin,nerror)\n\t\t\tif(nerror > 0) return\n if(collin) {\n\t\t\t\tnerror = 12\n\t\t\t\treturn\n\t\t\t}\n call succ(l,i,j,nadj,madj,ntot,nerror)\n\t\t\tif(nerror > 0) return\n call circen(i,j,l,c,d,x,y,ntot,eps,collin,nerror)\n\t\t\tif(nerror > 0) return\n if(collin) {\n\t\t\t\tnerror = 12\n\t\t\t\treturn\n\t\t\t}\n # If a circumcentre is outside the rectangular window\n # of interest, draw a line joining it to the other\n # circumcentre. Find the intersection of this line with\n # the boundary of the window; for (a,b) and call the point\n # of intersection (ai,bi). For (c,d), call it (ci,di).\n # Note: rwu = \"right way up\".\n xi = x(i)\n xj = x(j)\n yi = y(i)\n yj = y(j)\n if(yi!=yj) {\n slope = (xi - xj)/(yj - yi)\n rwu = .true.\n } else {\n slope = 0.d0\n rwu = .false.\n }\n call dldins(a,b,slope,rwu,ai,bi,rw,intfnd,bptab,nedgeab)\n\t\t\tif(!intfnd) {\n\t\t\t\tnerror = 16\n\t\t\t\treturn\n\t\t\t}\n call dldins(c,d,slope,rwu,ci,di,rw,intfnd,bptcd,nedgecd)\n\t\t\tif(!intfnd) {\n\t\t\t\tnerror = 16\n\t\t\t\treturn\n\t\t\t}\n\t\t\tgoferit = .false.\n\t\t\tif(bptab & bptcd) {\n\t\t\t\txm = 0.5*(ai+ci)\n\t\t\t\tym = 0.5*(bi+di)\n\t\t\t\tif(xmin ndir) {\n\t\t\t\t\tnerror = 15\n\t\t\t\t\treturn\n\t\t\t\t}\n\t\t\t\tdirsgs(1,kseg) = ai\n\t\t\t\tdirsgs(2,kseg) = bi\n\t\t\t\tdirsgs(3,kseg) = ci\n\t\t\t\tdirsgs(4,kseg) = di\n\t\t\t\tdirsgs(5,kseg) = i\n\t\t\t\tdirsgs(6,kseg) = j\n\t\t\t\tif(bptab) dirsgs(7,kseg) = 1.d0\n\t\t\t\telse dirsgs(7,kseg) = 0.d0\n\t\t\t\tif(bptcd) dirsgs(8,kseg) = 1.d0\n\t\t\t\telse dirsgs(8,kseg) = 0.d0\n if(bptab) dirsgs(9,kseg) = -nedgeab\n else dirsgs(9,kseg) = k\n if(bptcd) dirsgs(10,kseg) = -nedgecd\n else dirsgs(10,kseg) = l\n\t\t\t}\n }\n }\n}\nndir = kseg\n\nreturn\nend\n", "meta": {"hexsha": "f8bbfd9b1ccb98be4035dd011da038b4f68eae01", "size": 4113, "ext": "r", "lang": "R", "max_stars_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/dirseg.r", "max_stars_repo_name": "hyeongmokoo/SAAR_beta1", "max_stars_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2018-08-23T15:35:47.000Z", "max_stars_repo_stars_event_max_datetime": "2020-11-24T12:20:59.000Z", "max_issues_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/dirseg.r", "max_issues_repo_name": "hyeongmokoo/SAAR_beta1", "max_issues_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/dirseg.r", "max_forks_repo_name": "hyeongmokoo/SAAR_beta1", "max_forks_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2019-06-21T00:51:33.000Z", "max_forks_repo_forks_event_max_datetime": "2019-06-21T00:51:33.000Z", "avg_line_length": 28.5625, "max_line_length": 82, "alphanum_fraction": 0.5134938001, "num_tokens": 1331, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8558511396138365, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5295839393904181}} {"text": "#!/usr/bin/env Rscript\n#\n# Fisher Exact Test p-values from 2x2 contingency table\n#\nargs <- commandArgs(TRUE)\ninfile = paste(\"\", args[1], sep=\"\")\noutfile = paste(\"\", args[2], sep=\"\")\nif(nchar(infile) >= 1 && !infile == \"NA\" && nchar(outfile) >= 1 && !outfile == \"NA\") {\n x = read.table(infile, sep=\"\\t\", header=F)\n dx = data.matrix(x)\n n = dim(dx)[1]\n if(n >= 1) {\n p = rep(1, n)\n for(i in 1:n) {d=matrix(dx[i,],nrow=2); p[i]=fisher.test(d, alternative = \"greater\",conf.int=F)$p}\n fdr = p.adjust(p)\n dxp = data.frame(dx,p,fdr)\n names(dxp) = c(\"a\",\"b\",\"c\",\"d\",\"p-value\",\"fdr\")\n if(nchar(outfile) >= 1) {\n write.table(dxp, outfile, quote=F, sep=\"\\t\", row.names=F, col.names=T)\n } else {\n print(dxp)\n }\n }\n} else {\n warning(\"Usage: fisher-pvals.r [input table (a,b,c,d)] [output table (a,b,c,d,p-value,fdr)]\")\n}\n", "meta": {"hexsha": "27b524ada2d6faf583aa17dd1203465e627aabfd", "size": 904, "ext": "r", "lang": "R", "max_stars_repo_path": "fisher-pvals.r", "max_stars_repo_name": "mccrowjp/scripts", "max_stars_repo_head_hexsha": "110bdc01806757fa50efb5c0c0deb8455dcfa2cf", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 6, "max_stars_repo_stars_event_min_datetime": "2018-02-27T02:40:46.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-28T02:50:51.000Z", "max_issues_repo_path": "fisher-pvals.r", "max_issues_repo_name": "ChenSh1ne/utilities", "max_issues_repo_head_hexsha": "ebc63d68c1b7d819a89d59122c18e9419fa313fb", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "fisher-pvals.r", "max_forks_repo_name": "ChenSh1ne/utilities", "max_forks_repo_head_hexsha": "ebc63d68c1b7d819a89d59122c18e9419fa313fb", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 5, "max_forks_repo_forks_event_min_datetime": "2018-02-27T02:48:16.000Z", "max_forks_repo_forks_event_max_datetime": "2021-04-06T17:14:50.000Z", "avg_line_length": 33.4814814815, "max_line_length": 106, "alphanum_fraction": 0.5254424779, "num_tokens": 291, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7606506526772883, "lm_q2_score": 0.6959583313396339, "lm_q1q2_score": 0.529381158969689}} {"text": "\n # ---------------------------------------------------------------------\n # predict weights of individuals that do not have weights recorded from groundfish surveys\n # (for the moment, obtain weight-cw relationship from groundfish surveys and apply to cw from\n # snowcrab surveys\n # ------------------------------------------------------------------\n\n predictweights = function( Y, parameterisation=\"bio\") {\n\n # sex codes\n male = 0\n female = 1\n sex.unknown = 2\n\n # maturity codes\n immature = 0\n mature = 1\n mat.unknown = 2\n\n\n # fill in missing weights from historical data as they used regression to do this ..\n # a slight variation is that the model chosen in 2007 includes shell condition as well\n\n if (parameterisation==\"bio\") {\n # develop regression relationships for each of: using cw, sex, maturity, shell, weighted by sa\n\n x = Y\n\n x$log.mass = log(x$mass)\n x$log.cw = log(x$cw)\n x$log.chela = log(x$chela)\n x$log.abdom =log(x$abdomen)\n x0 = x\n ii = which( x$sex%in% c(male, female) )\n x$sex = factor(x$sex, levels=c(male,female) )\n\n mod = glm( log.mass ~ log.cw + log.cw:sex, weights=sa, data=x, na.action='na.exclude')\n x$predictedmass = predict( mod, x )\n x$rstand = abs( x$log.mass - x$predictedmass )\n x$rstand = x$rstand / sqrt(var(x$rstand, na.rm=T))\n\n jj = which(x$rstand > 3 )\n x$log.mass [ jj ] = x$predictedmass [ jj ]\n kk = which( !is.finite( x$log.mass ) )\n x$log.mass [ kk ] = x$predictedmass [ kk ]\n\n # still a few NA's: fill in with a simpler model dependent only upon cw ~ 500 cases\n mod = glm( log.mass ~ log.chela, weights=sa, data=x, na.action='na.exclude')\n x$predictedmass = predict( mod, x )\n kk = which( !is.finite( x$log.mass ) )\n x$log.mass [ kk ] = x$predictedmass [ kk ]\n\n # still a few NA's: fill in with a simpler model dependent only upon cw ~ 500 cases\n mod = glm( log.mass ~ log.abdom, weights=sa, data=x, na.action='na.exclude' )\n x$predictedmass = predict( mod, x )\n kk = which( !is.finite( x$log.mass ) )\n x$log.mass [ kk ] = x$predictedmass [ kk ]\n\n Y$mass = exp(x$log.mass)\n\n return(Y)\n\n# summary(mod);Anova(mod) from 2007 data\n\n#Call:\n#lm(formula = mass ~ cw + cw:sex + shell, subset = ii, weights = sa,\n# na.action = \"na.exclude\")\n\n#Residuals:\n# Min 1Q Median 3Q Max\n#-0.174287 -0.003357 -0.000181 0.003389 0.141542\n\n#Coefficients:\n# Estimate Std. Error t value Pr(>|t|)\n#(Intercept) -7.464833 0.024413 -305.78 < 2e-16 ***\n#cw 2.890774 0.003796 761.55 < 2e-16 ***\n#shell2 0.053130 0.017581 3.02 0.0025 **\n#shell3 0.125508 0.017627 7.12 1.1e-12 ***\n#shell4 0.154578 0.018308 8.44 < 2e-16 ***\n#shell5 0.164819 0.039400 4.18 2.9e-05 ***\n#cw:sex2 -0.023230 0.000649 -35.81 < 2e-16 ***\n#---\n\n\n#Residual standard error: 0.00838 on 13236 degrees of freedom\n# (36127 observations deleted due to missingness)\n#Multiple R-Squared: 0.985,\tAdjusted R-squared: 0.985\n#F-statistic: 1.46e+05 on 6 and 13236 DF, p-value: <2e-16\n\n#Anova Table (Type II tests)\n\n#Response: mass\n# Sum Sq Df F value Pr(>F)\n#cw 49.76 1 708761.9 <2e-16 ***\n#shell 0.05 4 186.7 <2e-16 ***\n#cw:sex 0.09 1 1282.6 <2e-16 ***\n#Residuals 0.93 13236\n#---\n#Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1\n\n\n#> summary(mod);Anova(mod)\n\n#Call:\n#lm(formula = mass ~ cw + cw:sex, data = newdata, weights = sa, na.action = \"na.exclude\")\n\n#Residuals:\n# Min 1Q Median 3Q Max\n#-0.0840392 -0.0021296 -0.0000281 0.0019451 0.1309121\n\n#Coefficients:\n# Estimate Std. Error t value Pr(>|t|)\n#(Intercept) -7.7028530 0.0014793 -5207 <2e-16 ***\n#cw 2.9630510 0.0003513 8435 <2e-16 ***\n#cw:sex2 -0.0150139 0.0000779 -193 <2e-16 ***\n#---\n#Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1\n\n#Residual standard error: 0.00379 on 205565 degrees of freedom\n# (1109 observations deleted due to missingness)\n#Multiple R-Squared: 0.997,\tAdjusted R-squared: 0.997\n#F-statistic: 3.96e+07 on 2 and 205565 DF, p-value: <2e-16\n\n#Anova Table (Type II tests)\n\n#Response: mass\n# Sum Sq Df F value Pr(>F)\n#cw 1135.4 1 79181818 <2e-16 ***\n#cw:sex 0.5 1 37130 <2e-16 ***\n#Residuals 2.9 205565\n#---\n#Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1\n\n }\n\n if (parameterisation == \"moncton\") {\n m.soft = filter.class(x, \"m.soft\")\n m.hard = filter.class(x, \"m.hard\")\n x$mass[m.hard] = x$cw[m.hard]^ 3.098 * 0.0002665\n x$mass[m.soft] = x$cw[m.soft]^ 3.524 * 0.00002995\n\n # use same equations as for male ... must fix this\n f.soft = filter.class(x, \"f.soft\")\n f.hard = filter.class(x, \"f.hard\")\n x$mass[f.hard] = x$cw[f.hard]^ 3.098 * 0.0002665\n x$mass[f.soft] = x$cw[f.soft]^ 3.524 * 0.00002995\n }\n\n return(x)\n }\n\n\n\n\n", "meta": {"hexsha": "3ba243ce68f85334d3386d14a24f33ae9b805f66", "size": 5095, "ext": "r", "lang": "R", "max_stars_repo_path": "R/predictweights.r", "max_stars_repo_name": "jae0/snowcrab", "max_stars_repo_head_hexsha": "b168df368b739175004275c47f5bdf6907d066d9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/predictweights.r", "max_issues_repo_name": "jae0/snowcrab", "max_issues_repo_head_hexsha": "b168df368b739175004275c47f5bdf6907d066d9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/predictweights.r", "max_forks_repo_name": "jae0/snowcrab", "max_forks_repo_head_hexsha": "b168df368b739175004275c47f5bdf6907d066d9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-04-21T12:57:48.000Z", "max_forks_repo_forks_event_max_datetime": "2020-04-21T12:57:48.000Z", "avg_line_length": 32.8709677419, "max_line_length": 100, "alphanum_fraction": 0.5576054956, "num_tokens": 1817, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8757869851639066, "lm_q2_score": 0.6039318337259583, "lm_q1q2_score": 0.5289156399033668}} {"text": "\n#' Generate prior distributions for k using stream geometry\n#' \n#' @details Generates means and standard deviations for half-normal priors for k600. This \n#' function uses the 7 equations from Raymond et al 2012. Which equation to use is a user\n#' choice, though there are some recommendations from Raymond to guide the choice (see below,\n#' and see Raymond et al 2012).\n#' \n#' Rayment et al provide standard deviations for the parameters, but do NOT provide information\n#' on correlations among parameters; therefore this function treats them as independent (which\n#' will generally produce larger prior standard deviations). The standard deviation is chosen\n#' by simulating across independent parameter combinations.\n#' \n#' Slope and velocity are required for all equations. Discharge and depth are required for\n#' equations 6 & 7 (discharge) and 1, 2, and 7 (depth); for equations 3-5 they can safely be\n#' left set to NA.\n#' \n#' Guidelines for k:\n#' Best for small streams:\n#' includes depth, more preditive, but less general, not in accordance with theory\n#' \t\teq 1, 2\n#'\n#' More general (includes only slope & velocity), but maybe less precise\n#' \t\teq 3-5\n#' \n#' Equation 6 also includes discharge, predictability is lower, an alt to 3-5\n#' Equation 7 includes depth and discharge; theoretically problematic, but\n#' \t\tprediction might be better for small streams\n#' \n#' @param slope Stream slope (unitless; meters of height/meters of length)\n#' @param velocity In meters/second\n#' @param discharge In m^3/sec\n#' @param depth In meters\n#' @param eqn Which equation(s) to use\n#' @param nsim Number of simulations used for the standard deviation\n#' \n#' @return If a single equations is given, a 2-column matrix giving the mean and standard\n#' deviation for k for each value in slope, velocity, etc. Otherwise a list of such matrices,\n#' one for each equation in eqn. Units for k are meters/day\n#' @references Raymond, P.A. et al. (2012). Scaling the gas transfer velocity and hydraulic\n#' \t\t geometry in streams and small rivers. Limnol. Oceanogr., 2, 41–53.\n#' @examples\n#' sl = 0.05\n#' v = 0.5\n#' q = 1\n#' d = 0.5\n#' eq1 = k_prior(sl, v, q, d, eqn = 1, nsim=3) ## normally run thousands of sims\n#' @export\nk_prior = function(slope, velocity, depth = NA, discharge = NA, eqn = 1:7, nsim = 5000) {\n\tif(!all(eqn %in% 1:7))\n\t\tstop(\"Unknown equation; ensure all values in eqn are in 1:7\")\n\tif(length(eqn) > 1) {\n\t\tlapply(eqn, function(e) k_prior(slope, velocity, depth, discharge, e, nsim))\n\t} else {\n\t\tparams = .k_parameters()\n\t\tsimpars = .k_generate_pars(params$pars[[eqn]], params$pars_se[[eqn]], 5000)\n\t\tsds = .k_sd_sim(slope, velocity, depth, discharge, simpars, eqn)\n\t\tmeans = .k_compute(slope, velocity, depth, discharge, params$pars[[eqn]], eqn)\n\t\tcbind(mean = means, stdev = sds)\n\t}\n}\n\n\n\n\n\n#' generate parameter sets (assuming independence) for k sims\n#' @keywords internal\n.k_generate_pars = function(par, se, n) {\n\tmapply(function(mu, stdev, N) rnorm(N, mu, stdev), par, se, n)\n}\n\n#' generate standard dev of k for a given set of stream variables\n#' @keywords internal\n.k_sd_sim = function(S, V, D, Q, pars, eqn) {\n\tgetsd = function(S, V, D, Q, p, eqn) sd(apply(p, 1, function(x)\n\t\t.k_compute(S, V, D, Q, x, eqn)))\n\tif(requireNamespace(\"parallel\")) {\n\t\tparallel::mcmapply(FUN = getsd, S = S, V = V, D = D, \n\t\t\tQ = Q, MoreArgs = list(p = pars, eqn = eqn))\n\t} else {\n\t\tmapply(FUN = getsd, S = S, V = V, D = D, \n\t\t\tQ = Q, MoreArgs = list(p = pars, eqn = eqn))\n\t}\n}\n\n# compute k\n#' @keywords internal\n.k_compute = function(S, V, D, Q, pars, eq) {\n\teq = paste0('eq', eq)\n\t# magic number is gravitational acceleration\n\tfr = V / sqrt(9.806*D)\n\tswitch(eq, \n\t\teq1 = (V*S)^(pars[1]) * D^(pars[2]) * pars[3],\n\t\teq2 = pars[1] * (1 - pars[2] * fr^2) * (V*S)^(pars[3]) * D^(pars[4]),\n\t\teq3 = pars[1] * S^pars[2] * V^pars[3],\n\t\teq4 = (V*S)^pars[1] * pars[2],\n\t\teq5 = V*S*pars[1] + pars[2],\n\t\teq6 = pars[1] * (V*S)^(pars[2]) * Q^pars[3],\n\t\teq7 = pars[1] * (V*S)^(pars[2]) * Q^pars[3] * D^pars[4]\n\t)\n}\n\n#' Constant for k parameters\n#' @keywords internal\n.k_parameters = function() {\n\t## parameters from Raymond\n\tpars = list(\n\t\teq1 = c(0.89, 0.54, 5037),\n\t\teq2 = c(5937, 2.54, 0.89, 0.58),\n\t\teq3 = c(1162, 0.77, 0.85),\n\t\teq4 = c(0.76, 951.5),\n\t\teq5 = c(2841, 2.02),\n\t\teq6 = c(929, 0.75, 0.011),\n\t\teq7 = c(4725, 0.86, -0.14, 0.66))\n\tpars_se = list(\n\t\teq1 = c(0.020, 0.030, 604),\n\t\teq2 = c(606, 0.223, 0.017, 0.027),\n\t\teq3 = c(192, 0.028, 0.045),\n\t\teq4 = c(0.027, 144),\n\t\teq5 = c(107, 0.209),\n\t\teq6 = c(141, 0.027, 0.016),\n\t\teq7 = c(445, 0.016, 0.012, 0.029))\n\tlist(pars = pars, pars_se = pars_se)\n}\n", "meta": {"hexsha": "535c084dd8e77ae461be4a3c79f5fc9ae96c8957", "size": 4597, "ext": "r", "lang": "R", "max_stars_repo_path": "R/priors.r", "max_stars_repo_name": "mtalluto/NSmetabolism", "max_stars_repo_head_hexsha": "1b179726cd1968f9562236799a82104ed5957d81", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/priors.r", "max_issues_repo_name": "mtalluto/NSmetabolism", "max_issues_repo_head_hexsha": "1b179726cd1968f9562236799a82104ed5957d81", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/priors.r", "max_forks_repo_name": "mtalluto/NSmetabolism", "max_forks_repo_head_hexsha": "1b179726cd1968f9562236799a82104ed5957d81", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.4841269841, "max_line_length": 95, "alphanum_fraction": 0.6571677181, "num_tokens": 1578, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7905303186696747, "lm_q2_score": 0.6688802537704063, "lm_q1q2_score": 0.5287701201649722}} {"text": "# With this skript I intend to analyze whether there is a significant\n# difference between the populations in the control (absolute gene\n# expression) samples and between the two stress temperatures (28°C\n# and 32°C)\n\nerror.bar <- function(x, y, upper, lower=upper, length=0.05,...){\n if(length(x) != length(y) | length(y) !=length(lower) | length(lower) != length(upper))\n stop(\"vectors must be same length\")\n arrows(x,upper, x, lower, angle=90, code=3, length=length, ...)\n }\n\n\nerror.bar2 <- function(x, y, upper, lower=upper, length=0.05,...){\n if(length(x) != length(y) | length(y) !=length(lower) | length(lower) != length(upper))\n stop(\"vectors must be same length\")\n arrows(x,y+upper, x, y-lower, angle=90, code=3, length=length, ...)\n }\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n#####################################################################\n#####################################################################\n########## Absolute expression of control samples ###################\n#####################################################################\n#####################################################################\nlibrary(\"nparLD\")\n\nAlldata32 <- read.table(\"/home/alj/Documents/org_files/data/ScientificJournal/2012ScientificJournal/Dec/2012-12-18-Fs_32_NormalizedAmount_OutliersRemoved.txt\",header=TRUE)\n\nAlldata28 <- read.table(\"/home/alj/Documents/org_files/data/ScientificJournal/2012ScientificJournal/Dec/2012-12-18-Fs_28_NormalizedAmount_OutliersRemoved.txt\",header=TRUE)\n\nlibrary(vegan)\n\n# Remove all data with FoldChange==NA\n# At this step I decide whether to take AllOutliersRemoved or AllBoxplotOutliersRemoved\nFT32All <- as.data.frame(as.matrix(Alldata32[is.na(Alldata32$NormalizedAmount)!=TRUE,]))\nData32.2 <- tapply(FT32All$MeanCT,list(FT32All$Gene,FT32All$Population,FT32All$Treatment),length)\n\nFT28All <- as.data.frame(as.matrix(Alldata28[is.na(Alldata28$NormalizedAmount)!=TRUE,]))\nData28.2 <- tapply(FT28All$MeanCT,list(FT28All$Gene,FT28All$Population,FT28All$Treatment),length)\n\n\n############################################################################\n############### Statistical analysis #######################################\n############################################################################\n\n# Do the analysis for the 32°C and the 28°C stress experiment and put\n# the results for the three Hsps in a list.\n\n\nFT <- rbind(FT28All,FT32All)\n# Combine the stress data\nFT$NormalizedAmount=as.numeric(as.vector(FT$NormalizedAmount))\nFT$Population=as.factor(FT$Population)\nFT$Treatment=as.factor(FT$Treatment)\n################### HSP70\n\nHSP70Compare <- subset(FT,FT$Gene==\"HSP70\")\nHSP90Compare <- subset(FT,FT$Gene==\"HSP90\")\nsHSPCompare <- subset(FT,FT$Gene==\"sHSP\")\n\nHSPs <- c(\"HSP70Compare\",\"HSP90Compare\",\"sHSPCompare\")\n\nMeanslist28 <- vector(\"list\",length(HSPs))\nSElist28upper <- vector(\"list\",length(HSPs))\nSElist28lower <- vector(\"list\",length(HSPs))\nsignificancelist28 <- vector(\"list\",length(HSPs))\nMeanslist32 <- vector(\"list\",length(HSPs))\nSElist32upper <- vector(\"list\",length(HSPs))\nSElist32lower <- vector(\"list\",length(HSPs))\nsignificancelist32 <- vector(\"list\",length(HSPs))\n\n#########################################\n######### Log transformation and ANOVA ##\n#########################################\nlibrary(\"lawstat\") # Provides the levene test to test for heteroscedasticity\n\n# The sHSP expression in the control samples from Spain (28°C) are\n# higher than for the other populations. I want to see if a\n# transformation of the data, followed by a normal ANOVA indicates\n# here a significant difference, since the PERMANOVA analysis does\n# not. I tested it and a log-transformation gives a statistical\n# significance if I apply an ANOVA on it.\n\n# I regard the normal ANOVA analysis only to be problematic when there\n# is an indication for non-normality or heteroscedasticity with a p-value<0.01\n\n# I do log-transformation only when necessary and then back-transform the data with exp(result)\n\n# Here I can change which gene shall be analyzed\nl=1\n###########\n### HSP 70\n###########\n\nd <- get(HSPs[l])\nHs.Pop0 <- subset(d,d$Treatment==0,drop=TRUE)\n\n# 28°C\nshapiro.test(log(Hs.Pop0$NormalizedAmount))\nlevene.test(log(Hs.Pop0$NormalizedAmount),Hs.Pop0$Population)\n# log-transformation necessary!\nmodel <- aov(log(Hs.Pop0$NormalizedAmount) ~ Hs.Pop0$Population*Hs.Pop0$Temperature)\nsummary(model)\nsignificances <- TukeyHSD(model)[[3]][,4]\n\n\nsignificancesTrue <- significances<0.05\nnames(significancesTrue) <- c(\"28F-28D\",\n \"28N-28D\",\n \"28S-28D\",\n \"32D-28D\",\n \"32F-28D\",\n \"32N-28D\",\n \"32S-28D\",\n \"28N-28F\",\n \"28S-28F\",\n \"32D-28F\",\n \"32F-28F\",\n \"32N-28F\",\n \"32S-28F\",\n \"28S-28N\",\n \"32D-28N\",\n \"32F-28N\",\n \"32N-28N\",\n \"32S-28N\",\n \"32D-28S\",\n \"32F-28S\",\n \"32N-28S\",\n \"32S-28S\",\n \"32F-32D\",\n \"32N-32D\",\n \"32S-32D\",\n \"32N-32F\",\n \"32S-32F\",\n \"32S-32N\")\nlibrary(\"multcompView\")\nComparisons <- c(\"28N\",\"28D\",\"28F\",\"28S\",\"32N\",\"32D\",\"32F\",\"32S\")\nLetters <- multcompLetters(significancesTrue)$Letters[match(c(\"28N\",\"28D\",\"28F\",\"28S\",\"32N\",\"32D\",\"32F\",\"32S\"),names(multcompLetters(significancesTrue)$Letters))]\n\n\n# Calculating the Means and SEs\n\nd28 <- d[d$Temperature==28,]\nd32 <- d[d$Temperature==32,]\n\n\nd28.Means <- as.matrix(tapply(as.numeric(as.vector(log(d28$NormalizedAmount))),list(as.factor(as.vector(d28$Population)),as.vector(d28$Treatment)),mean))\nd28.Means <- as.matrix(rbind(d28.Means[3,c(1,2,4,3)],d28.Means[1,c(1,2,4,3)],d28.Means[2,c(1,2,4,3)],d28.Means[4,c(1,2,4,3)]))\nrownames(d28.Means) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\nd28.Means2 <- d28.Means\nd28.Means <- exp(d28.Means)\n\nMeanslist28[[l]] <- d28.Means\n\nlibrary(plotrix)\n\nd28.SEs <- tapply(as.numeric(as.vector(log(d28$NormalizedAmount))),list(as.factor(as.vector(d28$Population)),as.vector(d28$Treatment)),std.error)\nd28.SEs <- as.matrix(rbind(d28.SEs[3,c(1,2,4,3)],d28.SEs[1,c(1,2,4,3)],d28.SEs[2,c(1,2,4,3)],d28.SEs[4,c(1,2,4,3)]))\nrownames(d28.SEs) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\n\nd28.UpperSEs <- exp(d28.Means2+d28.SEs)\nd28.LowerSEs <- exp(d28.Means2-d28.SEs)\n\nSElist28upper[[l]] <- d28.UpperSEs\nSElist28lower[[l]] <- d28.LowerSEs\n\n\nd32.Means <- as.matrix(tapply(as.numeric(as.vector(log(d32$NormalizedAmount))),list(as.factor(as.vector(d32$Population)),as.vector(d32$Treatment)),mean))\nd32.Means <- as.matrix(rbind(d32.Means[3,c(1,2,4,3)],d32.Means[1,c(1,2,4,3)],d32.Means[2,c(1,2,4,3)],d32.Means[4,c(1,2,4,3)]))\nrownames(d32.Means) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\nd32.Means2 <- d32.Means\nd32.Means <- exp(d32.Means)\n\nMeanslist32[[l]] <- d32.Means\n\nlibrary(plotrix)\n\nd32.SEs <- tapply(as.numeric(as.vector(log(d32$NormalizedAmount))),list(as.factor(as.vector(d32$Population)),as.vector(d32$Treatment)),std.error)\nd32.SEs <- as.matrix(rbind(d32.SEs[3,c(1,2,4,3)],d32.SEs[1,c(1,2,4,3)],d32.SEs[2,c(1,2,4,3)],d32.SEs[4,c(1,2,4,3)]))\nrownames(d32.SEs) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\n\nd32.UpperSEs <- exp(d32.Means2+d32.SEs)\nd32.LowerSEs <- exp(d32.Means2-d32.SEs)\n\nSElist32upper[[l]] <- d32.UpperSEs\nSElist32lower[[l]] <- d32.LowerSEs\n\n\nsignificancelist28[[l]] <- Letters[1:4]\nsignificancelist32[[l]] <- Letters[5:8]\n\n\n\n\n# Here I can change which gene shall be analyzed\nl=2\n###########\n### HSP 90\n###########\n\nd <- get(HSPs[l])\nHs.Pop0 <- subset(d,d$Treatment==0,drop=TRUE)\n\nshapiro.test(log(Hs.Pop0$NormalizedAmount))\nlevene.test(log(Hs.Pop0$NormalizedAmount),Hs.Pop0$Population)\n# log-transformation necessary!\nmodel <- aov(log(Hs.Pop0$NormalizedAmount) ~ Hs.Pop0$Population*Hs.Pop0$Temperature)\nsummary(model)\nsignificances <- TukeyHSD(model)[[3]][,4]\n\n\nsignificancesTrue <- significances<0.05\nnames(significancesTrue) <- c(\"28F-28D\",\n \"28N-28D\",\n \"28S-28D\",\n \"32D-28D\",\n \"32F-28D\",\n \"32N-28D\",\n \"32S-28D\",\n \"28N-28F\",\n \"28S-28F\",\n \"32D-28F\",\n \"32F-28F\",\n \"32N-28F\",\n \"32S-28F\",\n \"28S-28N\",\n \"32D-28N\",\n \"32F-28N\",\n \"32N-28N\",\n \"32S-28N\",\n \"32D-28S\",\n \"32F-28S\",\n \"32N-28S\",\n \"32S-28S\",\n \"32F-32D\",\n \"32N-32D\",\n \"32S-32D\",\n \"32N-32F\",\n \"32S-32F\",\n \"32S-32N\")\nlibrary(\"multcompView\")\nComparisons <- c(\"28N\",\"28D\",\"28F\",\"28S\",\"32N\",\"32D\",\"32F\",\"32S\")\nLetters <- multcompLetters(significancesTrue)$Letters[match(c(\"28N\",\"28D\",\"28F\",\"28S\",\"32N\",\"32D\",\"32F\",\"32S\"),names(multcompLetters(significancesTrue)$Letters))]\n\n\n# Calculating the Means and SEs\n\nd28 <- d[d$Temperature==28,]\nd32 <- d[d$Temperature==32,]\n\n\nd28.Means <- as.matrix(tapply(as.numeric(as.vector(log(d28$NormalizedAmount))),list(as.factor(as.vector(d28$Population)),as.vector(d28$Treatment)),mean))\nd28.Means <- as.matrix(rbind(d28.Means[3,c(1,2,4,3)],d28.Means[1,c(1,2,4,3)],d28.Means[2,c(1,2,4,3)],d28.Means[4,c(1,2,4,3)]))\nrownames(d28.Means) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\nd28.Means2 <- d28.Means\nd28.Means <- exp(d28.Means)\n\nMeanslist28[[l]] <- d28.Means\n\nlibrary(plotrix)\n\nd28.SEs <- tapply(as.numeric(as.vector(log(d28$NormalizedAmount))),list(as.factor(as.vector(d28$Population)),as.vector(d28$Treatment)),std.error)\nd28.SEs <- as.matrix(rbind(d28.SEs[3,c(1,2,4,3)],d28.SEs[1,c(1,2,4,3)],d28.SEs[2,c(1,2,4,3)],d28.SEs[4,c(1,2,4,3)]))\nrownames(d28.SEs) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\n\nd28.UpperSEs <- exp(d28.Means2+d28.SEs)\nd28.LowerSEs <- exp(d28.Means2-d28.SEs)\n\nSElist28upper[[l]] <- d28.UpperSEs\nSElist28lower[[l]] <- d28.LowerSEs\n\n\nd32.Means <- as.matrix(tapply(as.numeric(as.vector(log(d32$NormalizedAmount))),list(as.factor(as.vector(d32$Population)),as.vector(d32$Treatment)),mean))\nd32.Means <- as.matrix(rbind(d32.Means[3,c(1,2,4,3)],d32.Means[1,c(1,2,4,3)],d32.Means[2,c(1,2,4,3)],d32.Means[4,c(1,2,4,3)]))\nrownames(d32.Means) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\nd32.Means2 <- d32.Means\nd32.Means <- exp(d32.Means)\n\nMeanslist32[[l]] <- d32.Means\n\nlibrary(plotrix)\n\nd32.SEs <- tapply(as.numeric(as.vector(log(d32$NormalizedAmount))),list(as.factor(as.vector(d32$Population)),as.vector(d32$Treatment)),std.error)\nd32.SEs <- as.matrix(rbind(d32.SEs[3,c(1,2,4,3)],d32.SEs[1,c(1,2,4,3)],d32.SEs[2,c(1,2,4,3)],d32.SEs[4,c(1,2,4,3)]))\nrownames(d32.SEs) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\n\nd32.UpperSEs <- exp(d32.Means2+d32.SEs)\nd32.LowerSEs <- exp(d32.Means2-d32.SEs)\n\nSElist32upper[[l]] <- d32.UpperSEs\nSElist32lower[[l]] <- d32.LowerSEs\n\n\nsignificancelist28[[l]] <- Letters[1:4]\nsignificancelist32[[l]] <- Letters[5:8]\n\n\n\n# Here I can change which gene shall be analyzed\nl=3\n###########\n### sHSP\n###########\n\nd <- get(HSPs[l])\nHs.Pop0 <- subset(d,d$Treatment==0,drop=TRUE)\n\nshapiro.test(log(Hs.Pop0$NormalizedAmount))\nlevene.test(log(Hs.Pop0$NormalizedAmount),Hs.Pop0$Population)\n# log-transformation necessary!\nmodel <- aov(log(Hs.Pop0$NormalizedAmount) ~ Hs.Pop0$Population*Hs.Pop0$Temperature)\nsummary(model)\nsignificances <- TukeyHSD(model)[[3]][,4]\n\n\nsignificancesTrue <- significances<0.05\nnames(significancesTrue) <- c(\"28F-28D\",\n \"28N-28D\",\n \"28S-28D\",\n \"32D-28D\",\n \"32F-28D\",\n \"32N-28D\",\n \"32S-28D\",\n \"28N-28F\",\n \"28S-28F\",\n \"32D-28F\",\n \"32F-28F\",\n \"32N-28F\",\n \"32S-28F\",\n \"28S-28N\",\n \"32D-28N\",\n \"32F-28N\",\n \"32N-28N\",\n \"32S-28N\",\n \"32D-28S\",\n \"32F-28S\",\n \"32N-28S\",\n \"32S-28S\",\n \"32F-32D\",\n \"32N-32D\",\n \"32S-32D\",\n \"32N-32F\",\n \"32S-32F\",\n \"32S-32N\")\nlibrary(\"multcompView\")\nComparisons <- c(\"28N\",\"28D\",\"28F\",\"28S\",\"32N\",\"32D\",\"32F\",\"32S\")\nLetters <- multcompLetters(significancesTrue)$Letters[match(c(\"28N\",\"28D\",\"28F\",\"28S\",\"32N\",\"32D\",\"32F\",\"32S\"),names(multcompLetters(significancesTrue)$Letters))]\n\n\n# Calculating the Means and SEs\n\nd28 <- d[d$Temperature==28,]\nd32 <- d[d$Temperature==32,]\n\n\nd28.Means <- as.matrix(tapply(as.numeric(as.vector(log(d28$NormalizedAmount))),list(as.factor(as.vector(d28$Population)),as.vector(d28$Treatment)),mean))\nd28.Means <- as.matrix(rbind(d28.Means[3,c(1,2,4,3)],d28.Means[1,c(1,2,4,3)],d28.Means[2,c(1,2,4,3)],d28.Means[4,c(1,2,4,3)]))\nrownames(d28.Means) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\nd28.Means2 <- d28.Means\nd28.Means <- exp(d28.Means)\n\nMeanslist28[[l]] <- d28.Means\n\nlibrary(plotrix)\n\nd28.SEs <- tapply(as.numeric(as.vector(log(d28$NormalizedAmount))),list(as.factor(as.vector(d28$Population)),as.vector(d28$Treatment)),std.error)\nd28.SEs <- as.matrix(rbind(d28.SEs[3,c(1,2,4,3)],d28.SEs[1,c(1,2,4,3)],d28.SEs[2,c(1,2,4,3)],d28.SEs[4,c(1,2,4,3)]))\nrownames(d28.SEs) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\n\nd28.UpperSEs <- exp(d28.Means2+d28.SEs)\nd28.LowerSEs <- exp(d28.Means2-d28.SEs)\n\nSElist28upper[[l]] <- d28.UpperSEs\nSElist28lower[[l]] <- d28.LowerSEs\n\n\nd32.Means <- as.matrix(tapply(as.numeric(as.vector(log(d32$NormalizedAmount))),list(as.factor(as.vector(d32$Population)),as.vector(d32$Treatment)),mean))\nd32.Means <- as.matrix(rbind(d32.Means[3,c(1,2,4,3)],d32.Means[1,c(1,2,4,3)],d32.Means[2,c(1,2,4,3)],d32.Means[4,c(1,2,4,3)]))\nrownames(d32.Means) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\nd32.Means2 <- d32.Means\nd32.Means <- exp(d32.Means)\n\nMeanslist32[[l]] <- d32.Means\n\nlibrary(plotrix)\n\nd32.SEs <- tapply(as.numeric(as.vector(log(d32$NormalizedAmount))),list(as.factor(as.vector(d32$Population)),as.vector(d32$Treatment)),std.error)\nd32.SEs <- as.matrix(rbind(d32.SEs[3,c(1,2,4,3)],d32.SEs[1,c(1,2,4,3)],d32.SEs[2,c(1,2,4,3)],d32.SEs[4,c(1,2,4,3)]))\nrownames(d32.SEs) <- c(\"Norway\",\"Denmark\",\"France\",\"Spain\")\n#Back-transformation\n\nd32.UpperSEs <- exp(d32.Means2+d32.SEs)\nd32.LowerSEs <- exp(d32.Means2-d32.SEs)\n\nSElist32upper[[l]] <- d32.UpperSEs\nSElist32lower[[l]] <- d32.LowerSEs\n\n\nsignificancelist28[[l]] <- Letters[1:4]\nsignificancelist32[[l]] <- Letters[5:8]\n\n\n\n\n\n\nnames(Meanslist28) <- HSPs\nnames(SElist28upper) <- HSPs\nnames(SElist28lower) <- HSPs\nnames(significancelist28) <- HSPs\nnames(Meanslist32) <- HSPs\nnames(SElist32upper) <- HSPs\nnames(SElist32lower) <- HSPs\nnames(significancelist32) <- HSPs\n\n\nAbsMeanslist28 <- Meanslist28\nAbsSElist28upper <- SElist28upper\nAbsSElist28lower <- SElist28lower\nAbssignificancelist28 <- significancelist28\n\n\nAbsMeanslist32 <- Meanslist32\nAbsSElist32upper <- SElist32upper\nAbsSElist32lower <- SElist32lower\nAbssignificancelist32 <- significancelist32\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\nerror.bar <- function(x, y, upper, lower=upper, length=0.05,...){\n if(length(x) != length(y) | length(y) !=length(lower) | length(lower) != length(upper))\n stop(\"vectors must be same length\")\n arrows(x,upper, x, lower, angle=90, code=3, length=length, ...)\n }\n\n\nerror.bar2 <- function(x, y, upper, lower=upper, length=0.05,...){\n if(length(x) != length(y) | length(y) !=length(lower) | length(lower) != length(upper))\n stop(\"vectors must be same length\")\n arrows(x,y+upper, x, y-lower, angle=90, code=3, length=length, ...)\n }\n\n\n\n\n######################################################\n############# Plotting ###############################\n######################################################\n\ncnorway <- \"#fdfdfd\"\ncdenmark <- \"#808080\"\ncfrance <- \"#dfdfdf\"\ncspain <- \"#202020\"\n\n\n\n\n\n\n\n\n\n\n\n\n######################################################\n############# Plotting ###############################\n######################################################\n\n\n########## Plot for Control levels\npng(filename=\"/home/alj/Documents/writing/2012HSR/2012HeatShockResponse/Audioslides/images/Controls_Comparisons.png\",width=60,height=65,units='mm',res=100,pointsize=10)\npar(xpd=NA,mar=c(1,0,2,0),oma=c(3,6.5,0,0),mgp=c(4,2,1),font=1,ps=10,mfrow=c(2,1),bg=NA)\n\n\n\n# 28°C stress absolute control values\nMeans28 <- cbind(AbsMeanslist28$HSP70Compare[,1],AbsMeanslist28$HSP90Compare[,1],AbsMeanslist28$sHSPCompare[,1])\nSEs28upper <- cbind(AbsSElist28upper$HSP70Compare[,1],AbsSElist28upper$HSP90Compare[,1],AbsSElist28upper$sHSPCompare[,1])\nSEs28lower <- cbind(AbsSElist28lower$HSP70Compare[,1],AbsSElist28lower$HSP90Compare[,1],AbsSElist28lower$sHSPCompare[,1])\nsignificance28 <- c(Abssignificancelist28$HSP70Compare,Abssignificancelist28$HSP90Compare,Abssignificancelist28$sHSPCompare)\n \nbarx <- barplot(Means28[,3], beside=TRUE,col=c(rgb(51,102,0,maxColorValue=255),rgb(255,221,17,maxColorValue=255),rgb(102,51,153,maxColorValue=255),rgb(51,221,238,maxColorValue=255)),border=\"white\", ylim=c(0.001,10),axis.lty=1,lwd=1,cex.main=1,cex.lab=1,cex.axis=1,las=1,cex.names=1,cex.main=1.2,log=\"y\",axes=FALSE,main=\"\",names.arg=\"\",xaxt=\"n\")\naxis(2, at=c(0.001,0.1,10),labels=as.character(c(0.001,0.1,10)),las=1)\n\nlines(x=c(5.5,5.5),y=c(0.001,10000000),lty=2)\nlines(x=c(10.5,10.5),y=c(0.001,10000000),lty=2)\nerror.bar(barx,Means28[,3],upper=SEs28upper[,3],lower=SEs28lower[,3],lwd=1)\n\ndistances <- Means28[,3]+3\n\ntext(barx,distances,significance28[9:12],cex=1)\n\n\n# 32°C stress absolute control values\nMeans32 <- cbind(AbsMeanslist32$HSP70Compare[,1],AbsMeanslist32$HSP90Compare[,1],AbsMeanslist32$sHSPCompare[,1])\nSEs32upper <- cbind(AbsSElist32upper$HSP70Compare[,1],AbsSElist32upper$HSP90Compare[,1],AbsSElist32upper$sHSPCompare[,1])\nSEs32lower <- cbind(AbsSElist32lower$HSP70Compare[,1],AbsSElist32lower$HSP90Compare[,1],AbsSElist32lower$sHSPCompare[,1])\nsignificance32 <- c(Abssignificancelist32$HSP70Compare,Abssignificancelist32$HSP90Compare,Abssignificancelist32$sHSPCompare)\n \nbarx <- barplot(Means32[,3], beside=TRUE,col=c(rgb(51,102,0,maxColorValue=255),rgb(255,221,17,maxColorValue=255),rgb(102,51,153,maxColorValue=255),rgb(51,221,238,maxColorValue=255)),border=\"white\", ylim=c(0.001,10),axis.lty=1,xaxt=\"n\",lwd=1,cex.main=1,cex.lab=1,cex.axis=1,las=1,cex.names=1,cex.main=1.2,log=\"y\",axes=FALSE,main=\"\")\naxis(2, at=c(0.001,0.1,10),labels=as.character(c(0.001,0.1,10)),las=1)\n\n\nerror.bar(barx,Means32[,3],upper=SEs32upper[,3],lower=SEs32lower[,3],lwd=1)\n\ndistances <- Means32[,3]+3\n\ntext(barx,distances,significance32[9:12],cex=1)\n\n\ndev.off()\ndev.off(dev.prev())\n\n\n\n\n\n\n\n\n\n\n", "meta": {"hexsha": "97835e85d3af5f2317cbb73c76f05eefb6b00a3b", "size": 20174, "ext": "r", "lang": "R", "max_stars_repo_path": "figures/Control_Comparisons.r", "max_stars_repo_name": "alj1983/TrialLecture20192", "max_stars_repo_head_hexsha": "0eec1db27c6bb2ebb46fbbd967a2e3481f63e185", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "figures/Control_Comparisons.r", "max_issues_repo_name": "alj1983/TrialLecture20192", "max_issues_repo_head_hexsha": "0eec1db27c6bb2ebb46fbbd967a2e3481f63e185", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "figures/Control_Comparisons.r", "max_forks_repo_name": "alj1983/TrialLecture20192", "max_forks_repo_head_hexsha": "0eec1db27c6bb2ebb46fbbd967a2e3481f63e185", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.2076788831, "max_line_length": 344, "alphanum_fraction": 0.5835233469, "num_tokens": 6291, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.798186787341014, "lm_q2_score": 0.6619228825191872, "lm_q1q2_score": 0.5283380990654934}} {"text": "model_penman <- function (evapoTranspirationPriestlyTaylor = 449.367,\n hslope = 0.584,\n VPDair = 2.19,\n psychrometricConstant = 0.66,\n Alpha = 1.5,\n lambdaV = 2.454,\n rhoDensityAir = 1.225,\n specificHeatCapacityAir = 0.00101,\n conductance = 598.685){\n #'- Name: Penman -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: Penman Model\n #' * Author: Pierre Martre\n #' * Reference: Modelling energy balance in the wheat crop model SiriusQuality2:\n #' Evapotranspiration and canopy and soil temperature calculations\n #' * Institution: INRA/LEPSE Montpellier\n #' * Abstract: This method is used when wind and vapor pressure daily data are available\n #' \n #'- inputs:\n #' * name: evapoTranspirationPriestlyTaylor\n #' ** description : evapoTranspiration of Priestly Taylor \n #' ** variablecategory : rate\n #' ** datatype : DOUBLE\n #' ** default : 449.367\n #' ** min : 0\n #' ** max : 10000\n #' ** unit : g m-2 d-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' * name: hslope\n #' ** description : the slope of saturated vapor pressure temperature curve at a given temperature \n #' ** variablecategory : auxiliary\n #' ** datatype : DOUBLE\n #' ** default : 0.584\n #' ** min : 0\n #' ** max : 1000\n #' ** unit : hPa °C-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' * name: VPDair\n #' ** description : vapour pressure density\n #' ** variablecategory : auxiliary\n #' ** datatype : DOUBLE\n #' ** default : 2.19\n #' ** min : 0\n #' ** max : 1000\n #' ** unit : hPa\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' * name: psychrometricConstant\n #' ** description : psychrometric constant\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.66\n #' ** min : 0\n #' ** max : 1\n #' ** unit : \n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: Alpha\n #' ** description : Priestley-Taylor evapotranspiration proportionality constant\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 1.5\n #' ** min : 0\n #' ** max : 100\n #' ** unit : \n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: lambdaV\n #' ** description : latent heat of vaporization of water\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 2.454\n #' ** min : 0\n #' ** max : 10\n #' ** unit : \n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: rhoDensityAir\n #' ** description : Density of air\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 1.225\n #' ** unit : \n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: specificHeatCapacityAir\n #' ** description : Specific heat capacity of dry air\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.00101\n #' ** min : 0\n #' ** max : 1\n #' ** unit : \n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: conductance\n #' ** description : conductance\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 10000\n #' ** default : 598.685\n #' ** unit : m d-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #'- outputs:\n #' * name: evapoTranspirationPenman\n #' ** description : evapoTranspiration of Penman Monteith\n #' ** variablecategory : rate\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 5000\n #' ** unit : g m-2 d-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n evapoTranspirationPenman <- evapoTranspirationPriestlyTaylor / Alpha + (1000.0 * (rhoDensityAir * specificHeatCapacityAir * VPDair * conductance / (lambdaV * (hslope + psychrometricConstant))))\n return (list('evapoTranspirationPenman' = evapoTranspirationPenman))\n}", "meta": {"hexsha": "0ea533e2217b178249da9b18ecde095682350dd5", "size": 6753, "ext": "r", "lang": "R", "max_stars_repo_path": "test/Models/energybalance_pkg/src/r/Penman.r", "max_stars_repo_name": "brichet/PyCrop2ML", "max_stars_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-06-21T18:58:04.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-29T21:32:28.000Z", "max_issues_repo_path": "test/Models/energybalance_pkg/src/r/Penman.r", "max_issues_repo_name": "brichet/PyCrop2ML", "max_issues_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 27, "max_issues_repo_issues_event_min_datetime": "2018-12-04T15:35:44.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-11T08:25:03.000Z", "max_forks_repo_path": "test/Models/energybalance_pkg/src/r/Penman.r", "max_forks_repo_name": "brichet/PyCrop2ML", "max_forks_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 7, "max_forks_repo_forks_event_min_datetime": "2019-04-20T02:25:22.000Z", "max_forks_repo_forks_event_max_datetime": "2021-11-04T07:52:35.000Z", "avg_line_length": 56.7478991597, "max_line_length": 197, "alphanum_fraction": 0.3822005035, "num_tokens": 1419, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8856314647623016, "lm_q2_score": 0.5964331462646254, "lm_q1q2_score": 0.5282199609591283}} {"text": "# script for cv mortality \n# extract the data from the information \n# save the datasets after combining the data.\n\nlibrary(\"tidyverse\")\nlibrary(\"survival\")\nlibrary(\"IPDfromKM\")\nlibrary(\"metaRMST\")\nlibrary(\"broom\")\nlibrary(\"survminer\")\n\n# get the data 1 trial at a time and then format it to extract information.\n\n# leader \n# liraglutide arm - \n\nli <- read.table(\"F:/GLP1_agonists/data/leader/cvmort_liraglutide.txt\")\n\nsummary(li)\n\n# convert V2 to survival - \n\nli$V2 <- with(li, ifelse(V2 == -0.0479, 0, V2)) \n\nli$V2 <- 100 - li$V2\n\nli$V2 <- with(li, ifelse(V2 > 100, 100, V2))\n\nsummary(li)\n\n# now to extract information \n\nnrisk_l <- c(4668, 4641, 4599, 4558, 4505, 4445, 4382, 4322, 1723, 484)\n\ntrisk_l <- c(0,6, 12, 18, 24, 32, 36, 42, 48, 54)\n\nleader_l <- preprocess(dat = li,\n trisk = trisk_l,\n nrisk = nrisk_l,\n maxy = 100)\n\n\n\n\nleader_ipd_li <- getIPD(prep = leader_l,\n armID = 1)\n\n\nplot(leader_ipd_li)\n\ncv_mort_li <- leader_ipd_li$IPD\n\n\n# control arm \n\n\ncontrol <- \nread.table(\"F:/GLP1_agonists/data/leader/cvmort_control.txt\")\n\nsummary(control)\n\n# convert V2 to survival - \n\ncontrol$V2 <- 100 - control$V2\n\nsummary(control)\n\n# now to extract information \n\nnrisk_c <- c(4672, 4648, 4601, 4479, 4407, 4338, 4267, 1709, 465)\n\ntrisk_c <- c(0,6, 12, 18, 24, 32, 36, 42, 48, 54)\n\nleader_c <- preprocess(dat = control,\n trisk = trisk_c,\n nrisk = nrisk_c,\n maxy = 100)\n\n\n\n\nleader_ipd_c <- getIPD(prep = leader_c,\n armID = 0)\n\n\nplot(leader_ipd_c)\n\ncvmort_control <- leader_ipd_c$IPD\n\n\n# combine data and then plot \n\ndf_leader_cvmort <- rbind(cv_mort_li, cvmort_control)\n\ncvmort_leader_s <- \n survfit(Surv(time, status) ~ treat, data = df_leader_cvmort)\n\n\nggsurvplot(cvmort_leader_s, \n fun = \"event\",\n ylim = c(0,0.20),\n xlim = c(0,54),\n break.x.y = 6,\n break.y.by = 0.05,\n censor.size = 0,\n palette = c(\"gray\",\"blue2\"))\n\n# time already in months.\n\nwrite_csv(df_leader_cvmort,\n 'F:\\\\GLP1_agonists\\\\data\\\\pooled_data\\\\cvmort_leader.csv'\n)\n\n# sustain 6 \n\nsema <- read.table(\"F:/GLP1_agonists/data/sustain6/cvmort_semaglutide.txt\")\n\nsummary(sema)\n\n# convert V2 to survival - \n\nsema$V2 <- 100 - sema$V2\n\nsummary(sema)\n\n# now to extract information \n\nnrisk_s <- c(1648, 1634, 1627, 1617, 1607, 1589, 1579)\n\ntrisk_s <- c(0,16, 32, 48, 64, 80, 96)\n\nsustain6_s <- preprocess(dat = sema,\n trisk = trisk_s,\n nrisk = nrisk_s,\n maxy = 100)\n\n\nsustain6_ipd_sema <- getIPD(prep = sustain6_s,\n armID = 1)\n\n\nplot(sustain6_ipd_sema)\n\ncv_mort_sema <- sustain6_ipd_sema$IPD\n\n\n# now to get control arm data \n\n\ncontrol <- \nread.table(\"F:/GLP1_agonists/data/sustain6/cvmort_control.txt\")\n\nsummary(control)\n\n# convert V2 to survival - \n\ncontrol$V2 <- 100 - control$V2\n\nsummary(control)\n\n# now to extract information \n\nnrisk_c <- c(1649, 1637, 1623, 1617, 1600, 1584, 1566)\n\ntrisk_c <- c(0,16, 32, 48, 64, 80, 96)\n\nsustain6_c <- preprocess(dat = control,\n trisk = trisk_c,\n nrisk = nrisk_c,\n maxy = 100)\n\n\nsustain6_ipd_control <- getIPD(prep = sustain6_c,\n armID = 0)\n\n\nplot(sustain6_ipd_control)\n\ncv_mort_control <- sustain6_ipd_control$IPD\n\n\n# combine the dataset and then plot again \n\ncvmort_sustain6 <- rbind(cv_mort_sema, cv_mort_control)\n\nsurv_sustain6 <- survfit(Surv(time, status) ~ treat, \n data = cvmort_sustain6)\n\n# plot graph \n\n\nggsurvplot(surv_sustain6,\n data = cvmort_sustain6,\n fun = \"event\",\n ylim = c(0,0.05),\n break.y.by = 0.01,\n xim = c(0, 104),\n break.x.by = 8,\n palette = c(\"gray\",\"blue2\"),\n censor.size = 0)\n\n# now to convert weeks to months\n# save with time in months \n\ncvmort_sustain6$time <- cvmort_sustain6$time/4\n\n# save dataaset\n\nwrite_csv(cvmort_sustain6,\n 'F:\\\\GLP1_agonists\\\\data\\\\pooled_data\\\\cvmort_sustain6.csv')\n\n\n# EXCSEL \n\n\nexe <- \nread.table(\"F:/GLP1_agonists/data/excsel/cvmort_exenatide.txt\")\n\nsummary(exe)\n\n# convert V2 to survival - \n\nexe$V2 <- 100 - exe$V2\n\nsummary(exe)\n\n# now to extract information \n\nnrisk_e <- c(7356, 7234, 6433, 4095, 2698, 892)\n\ntrisk_e <- c(0,1,2,3,4,5)\n\nexcsel_exe <- preprocess(dat = exe,\n trisk = trisk_e,\n nrisk = nrisk_e,\n maxy = 100)\n\n\nexcsel_ipd_exe <- getIPD(prep = excsel_exe,\n armID = 1)\n\n\nplot(excsel_ipd_exe)\n\ncv_mort_exe <- excsel_ipd_exe$IPD\n\n# control arm of EXCSEL \n\ncontrol <- \n read.table(\"F:/GLP1_agonists/data/excsel/cvmort_control.txt\")\n\nsummary(control)\n\n# convert V2 to survival - \n\ncontrol$V2 <- 100 - control$V2\n\nsummary(control)\n\n# now to extract information \n\nnrisk_c <- c(7396, 7278, 6470, 4091, 2666, 907)\n\ntrisk_c <- c(0,1,2,3,4,5)\n\nexcsel_c <- preprocess(dat = control,\n trisk = trisk_c,\n nrisk = nrisk_c,\n maxy = 100)\n\n\nexcsel_ipd_c <- getIPD(prep = excsel_c,\n armID = 0)\n\n\nplot(excsel_ipd_c)\n\ncv_mort_control <- excsel_ipd_c$IPD\n\n# combine dataset and then plot \n\ncvmort_excsel <- rbind(cv_mort_exe, cv_mort_control)\n\n\ncvmort_excsel_s <- survfit(Surv(time, status) ~ treat,\n data = cvmort_excsel)\n\nggsurvplot(cvmort_excsel_s,\n data = cvmort_excsel,\n xlim = c(0,5),\n ylim = c(0,0.18),\n break.y.by = 0.03,\n censor.size = 0,\n fun = \"event\",\n palette = c(\"blue\",\"red\"),\n linetype = c(2,1))\n\n\n# convert time to months\n# save dataset\n\ncvmort_excsel$time <- cvmort_excsel$time*12\n\nwrite_csv(cvmort_excsel,\n 'F:\\\\GLP1_agonists\\\\data\\\\pooled_data\\\\cvmort_excsel.csv'\n)\n\n\n# Harmony Outcomes \n\n\nalbi <- \n read.table(\"F:/GLP1_agonists/data/Harmony/cvmort_albiglutide.txt\")\n\nsummary(albi)\n\n# convert V2 to survival - \n\nalbi$V2 <- 100 - albi$V2\n\nsummary(albi)\n\n# now to extract information \n\nnrisk_a <- c(4731, 4681, 4611, 4379, 3274, 2234, 1121)\n\ntrisk_a <- c(0, 4, 8, 12, 16, 20, 24 )\n\nharmony_a <- preprocess(dat = albi,\n trisk = trisk_a,\n nrisk = nrisk_a,\n maxy = 100)\n\n\nharmony_ipd_a <- getIPD(prep = harmony_a,\n armID = 1)\n\n\nplot(harmony_ipd_a)\n\ncv_mort_albi <- harmony_ipd_a$IPD\n\n\n# control arm in Harmony Outcomes \n\n\ncontrol <- \nread.table(\"F:/GLP1_agonists/data/Harmony/cvmort_control.txt\")\n\nsummary(control)\n\n# convert V2 to survival - \n\ncontrol$V2 <- 100 - control$V2\n\nsummary(control)\n\n# now to extract information \n\nnrisk_c <- c(4732, 4662, 4580, 4373, 3245, 2261, 1121)\n\ntrisk_c <- c(0, 4, 8, 12, 16, 20, 24 )\n\nharmony_c <- preprocess(dat = control,\n trisk = trisk_c,\n nrisk = nrisk_c,\n maxy = 100)\n\n\nharmony_ipd_c <- getIPD(prep = harmony_c,\n armID = 0)\n\n\nplot(harmony_ipd_c)\n\ncv_mort_harmony_c <- harmony_ipd_c$IPD\n\n# combine dataasets\n\ncvmort_harmony <- rbind(cv_mort_albi, cv_mort_harmony_c)\n\ncvmort_hs <- survfit(Surv(time, status) ~ treat, \n data = cvmort_harmony)\n\nggsurvplot(\n cvmort_hs,\n data = cvmort_harmony,\n fun = \"event\",\n xlim = c(0,28),\n ylim = c(0, 0.18),\n break.y.by = 0.02,\n break.x.by = 4,\n censor.size = 0,\n palette = c(\"blue2\",\"red\")\n)\n\n\n# convert time to months from weeks\n\ncvmort_harmony$time <- cvmort_harmony$time*4\n\nwrite_csv(\n cvmort_harmony,\n 'F:\\\\GLP1_agonists\\\\data\\\\pooled_data\\\\cvmort_harmony.csv'\n)\n\n# REWIND\n\n\ndula <- \nread.table(\"F:/GLP1_agonists/data/rewind/cvmort_dulaglutide.txt\")\n\nsummary(dula)\n\ndula$V1 <- with(dula, ifelse(V1 < 0, 0 , V1))\n\ndula$V2 <- with(dula, ifelse(V2 < 0, 0, V2))\n\n# convert V2 to survival - \n\ndula$V2 <- 100 - dula$V2\n\nsummary(dula)\n\n# now to extract information \n\nnrisk_d <- c(4949, 4866, 4773, 4663, 4556, 3887, 807)\n\ntrisk_d <- c(0, 1, 2, 3, 4, 5, 6)\n\nrewind_d <- preprocess(dat = dula,\n trisk = trisk_d,\n nrisk = nrisk_d,\n maxy = 100)\n\n\nrewind_ipd_d <- getIPD(prep = rewind_d,\n armID = 1)\n\n\nplot(rewind_ipd_d)\n\ncvmort_rewind_d <- rewind_ipd_d$IPD\n\n# control arm of rewind \n\ncontrol <- \n \nread.table(\n \"F:/GLP1_agonists/data/rewind/cvmort_control.txt\"\n)\n\nsummary(control)\n\ncontrol$V1 <- with(control, ifelse(V1 < 0, 0 , V1))\n\n# convert V2 to survival - \n\ncontrol$V2 <- 100 - control$V2\n\nsummary(control)\n\n# now to extract information \n\nnrisk_c <- c(4952, 4854, 4748, 4617, 4499, 3813, 802)\n\ntrisk_c <- c(0, 1, 2, 3, 4, 5, 6)\n\nrewind_c <- preprocess(dat = control,\n trisk = trisk_c,\n nrisk = nrisk_c,\n maxy = 100)\n\n\nrewind_ipd_c <- getIPD(prep = rewind_c,\n armID = 0)\n\n\nplot(rewind_ipd_c)\n\ncvmort_rewind_c <- rewind_ipd_c$IPD\n\n# combine dataset\n\ncvmort_rewind <- rbind(cvmort_rewind_d, cvmort_rewind_c)\n\n# create survobject and plot \n\ncvmort_rewind_s <- survfit(Surv(time, status) ~ treat, \n data = cvmort_rewind)\n\nplot(cvmort_rewind_s)\n\nggsurvplot(cvmort_rewind_s,\n data = cvmort_rewind,\n fun = \"event\",\n ylim = c(0, 0.18),\n censor.size = 0,\n break.y.by = 0.03)\n\n# convert years to months \n\ncvmort_rewind$time <- cvmort_rewind*12\n\nwrite_csv(\n cvmort_rewind,\n 'F:\\\\GLP1_agonists\\\\data\\\\pooled_data\\\\cvmort_rewind.csv'\n)\n\n# PIONEER 6 \n\npsema <- \n read.table(\n \"F:/GLP1_agonists/data/pioneer6/cvmort_semaglutide.txt\")\n\n\nsummary(psema)\n\n\npsema$V1 <- with(psema, ifelse(V1 < 0, 0, V1))\n\npsema$V2 <- with(psema, ifelse(V2 < 0, 0, V2))\n\n# convert V2 to survival - \n\npsema$V2 <- 100 - psema$V2\n\nsummary(psema)\n\n\n\n# now to extract information \n\npioneer6_sema <- preprocess(dat = psema,\n totalpts = 1591,\n maxy = 100)\n\n\npioneer6_ipd_sema <- getIPD(prep = pioneer6_sema,\n armID = 1)\n\n\nplot(pioneer6_ipd_sema)\n\ncvmort_pioneer6_sema <- pioneer6_ipd_sema$IPD\n\n\n# control arm for pioneer6 \n\n\np_control <- \n read.table(\n \"F:/GLP1_agonists/data/pioneer6/cvmort_control.txt\")\n\n\nsummary(p_control)\n\n\np_control$V1 <- with(p_control, ifelse(V1 < 0, 0, V1))\n\n# convert V2 to survival - \n\np_control$V2 <- 100 - p_control$V2\n\nsummary(p_control)\n\n\n# now to extract information \n\npioneer6_control <- preprocess(dat = p_control,\n totalpts = 1592,\n maxy = 100)\n\n\npioneer6_ipd_control <- getIPD(prep = pioneer6_control,\n armID = 0)\n\n\ncvmort_pioneer6_control <- pioneer6_ipd_control$IPD\n\ncvmort_pioneer6 <- rbind(cvmort_pioneer6_sema, cvmort_pioneer6_control)\n\n# plot graph \n\ncvmort_pioneer6_s <- survfit(Surv(time, status) ~ treat, \n data = cvmort_pioneer6)\n\n\nggsurvplot(cvmort_pioneer6_s,\n data = cvmort_pioneer6,\n xlim = c(0, 83),\n ylim = c(0, 0.1),\n break.x.by = 9,\n break.y.by = 0.01,\n censor.size = 0,\n palette = c(\"gray\",\"blue2\"),\n fun = \"event\")\n\n# convert weeks to months \n# save combined dataset\n\ncvmort_pioneer6$time <- cvmort_pioneer6$time/4\n\nwrite_csv(cvmort_pioneer6,\n 'F:\\\\GLP1_agonists\\\\data\\\\pooled_data\\\\cvmort_pioneer6.csv')\n\n\n", "meta": {"hexsha": "3ebb074a153cfcb5f747899bd3aad2695360b79f", "size": 11583, "ext": "r", "lang": "R", "max_stars_repo_path": "script1_cvmortality.r", "max_stars_repo_name": "svd09/RMST_glp1a", "max_stars_repo_head_hexsha": "987971a19f7678cb782f9c7a2041dc0a4f10a860", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "script1_cvmortality.r", "max_issues_repo_name": "svd09/RMST_glp1a", "max_issues_repo_head_hexsha": "987971a19f7678cb782f9c7a2041dc0a4f10a860", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "script1_cvmortality.r", "max_forks_repo_name": "svd09/RMST_glp1a", "max_forks_repo_head_hexsha": "987971a19f7678cb782f9c7a2041dc0a4f10a860", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 19.0509868421, "max_line_length": 75, "alphanum_fraction": 0.5971682638, "num_tokens": 3763, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8198933271118222, "lm_q2_score": 0.6442251064863697, "lm_q1q2_score": 0.5281958659660776}} {"text": "#' add\n#' \n#' Add two matrices: \\code{ret = alpha*x + beta*y}.\n#' \n#' @param transx,transy Should x/y be transposed?\n#' @param alpha,beta Scalars.\n#' @param x,y Input data.\n#' @param ret Either \\code{NULL} or an already allocated fml matrix of the same\n#' class and type as \\code{x}.\n#' @return Returns the matrix sum.\n#' \n#' @rdname linalg-add\n#' @name add\n#' \n#' @useDynLib fmlr R_linalg_add\n#' \n#' @export\nlinalg_add = function(transx=FALSE, transy=FALSE, alpha=1, beta=1, x, y, ret=NULL)\n{\n check_is_mat(x)\n check_is_mat(y)\n \n transx = as.logical(transx)\n transy = as.logical(transy)\n \n alpha = as.double(alpha)\n beta = as.double(beta)\n \n invisiret = check_inputs(ret, x, y)\n \n if (is.null(ret))\n ret = setret(x)\n \n .Call(R_linalg_add, get_backend(x), x$get_type(), transx, transy, alpha, beta, x$data_ptr(), y$data_ptr(), ret$data_ptr())\n \n if (invisiret)\n invisible(ret)\n else\n ret\n}\n\n\n\n#' matmult\n#' \n#' Multiply two matrices: \\code{ret = alpha*x*y}.\n#' \n#' @param transx,transy Should x/y be transposed?\n#' @param alpha Scalar.\n#' @param x,y Input data.\n#' @param ret Either \\code{NULL} or an already allocated fml matrix of the same\n#' class and type as \\code{x}.\n#' @return Returns the matrix product.\n#' \n#' @rdname linalg-matmult\n#' @name matmult\n#' \n#' @useDynLib fmlr R_linalg_matmult\n#' \n#' @export\nlinalg_matmult = function(transx=FALSE, transy=FALSE, alpha=1, x, y, ret=NULL)\n{\n if (!is_mat(x) && !is_vec(x))\n stop(\"argument 'x' must be a matrix or vector type\")\n if (!is_mat(y) && !is_vec(y))\n stop(\"argument 'y' must be a matrix or vector type\")\n \n transx = as.logical(transx)\n transy = as.logical(transy)\n \n alpha = as.double(alpha)\n \n \n invisiret = check_inputs(ret, x, y, check_class=FALSE)\n \n if (is.null(ret))\n {\n vec = is_vec(x) || is_vec(y)\n ret = setret(x, vec=vec)\n }\n \n .Call(R_linalg_matmult, get_backend(x), x$get_type(), transx, transy, alpha, x$get_class(), x$data_ptr(), y$get_class(), y$data_ptr(), ret$data_ptr())\n \n if (invisiret)\n invisible(ret)\n else\n ret\n}\n\n\n\n#' @useDynLib fmlr R_linalg_crossprod\nlinalg_crossprods = function(x, ret, alpha, xpose)\n{\n check_is_mat(x)\n \n xpose = as.logical(xpose)\n alpha = as.double(alpha)\n \n invisiret = check_inputs(ret, x)\n \n if (is.null(ret))\n ret = setret(x)\n \n .Call(R_linalg_crossprod, get_backend(x), x$get_type(), xpose, alpha, x$data_ptr(), ret$data_ptr())\n \n if (invisiret)\n invisible(ret)\n else\n ret\n}\n\n#' crossprod\n#' \n#' Compute crossproducts.\n#' \n#' @param alpha Number to scale the crossproduct by.\n#' @param x Input data.\n#' @param ret Either \\code{NULL} or an already allocated fml matrix of the same\n#' class and type as \\code{x}.\n#' @return Returns the crossproduct.\n#' \n#' @rdname linalg-crossprod\n#' @name crossprod\nNULL\n\n#' @rdname linalg-crossprod\n#' @export\nlinalg_crossprod = function(alpha=1, x, ret=NULL)\n{\n linalg_crossprods(x, ret, alpha, xpose=FALSE)\n}\n\n#' @rdname linalg-crossprod\n#' @export\nlinalg_tcrossprod = function(alpha=1, x, ret=NULL)\n{\n linalg_crossprods(x, ret, alpha, xpose=TRUE)\n}\n\n\n\n#' xpose\n#' \n#' Matrix transpose.\n#' \n#' @param x Input data.\n#' @param ret Either \\code{NULL} or an already allocated fml matrix of the same\n#' class and type as \\code{x}.\n#' @return Returns the xpose.\n#' \n#' @rdname linalg-xpose\n#' @name xpose\n#' @useDynLib fmlr R_linalg_xpose\n#' \n#' @export\nlinalg_xpose = function(x, ret=NULL)\n{\n check_is_mat(x)\n invisiret = check_inputs(ret, x)\n \n if (is.null(ret))\n ret = setret(x)\n \n .Call(R_linalg_xpose, get_backend(x), x$get_type(), x$data_ptr(), ret$data_ptr())\n \n if (invisiret)\n invisible(ret)\n else\n ret\n}\n\n\n\n#' lu\n#' \n#' LU factorization. The factorization occurs in-place.\n#' \n#' @param x Input data, overwritten by its LU factorization.\n#' @return Returns \\code{NULL}.\n#' \n#' @rdname linalg-lu\n#' @name lu\n#' @useDynLib fmlr R_linalg_lu\n#' \n#' @export\nlinalg_lu = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_lu, get_backend(x), x$get_type(), x$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' det\n#' \n#' Determinant\n#' \n#' @param x Input data, overwritten by its LU factorization.\n#' \n#' @return Returns a list containing the modulus and the sign.\n#' \n#' @rdname linalg-det\n#' @name det\n#' @useDynLib fmlr R_linalg_det\n#' \n#' @export\nlinalg_det = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_det, get_backend(x), x$get_type(), x$data_ptr())\n}\n\n\n\n#' trace\n#' \n#' Matrix trace, i.e. theh sum of the diagonal elements.\n#' \n#' @param x Input data.\n#' @return Returns the trace.\n#' \n#' @rdname linalg-trace\n#' @name trace\n#' @useDynLib fmlr R_linalg_trace\n#' \n#' @export\nlinalg_trace = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_trace, get_backend(x), x$get_type(), x$data_ptr())\n}\n\n\n\n#' svd\n#' \n#' Computes the singular value decomposition.\n#' \n#' @details\n#' You will need to initialize the return objects \\code{s} and/or \\code{u} and\n#' \\code{vt}.\n#' manually. See the example.\n#' \n#' @param x Input data. The input values are overwritten.\n#' @param s Singular values.\n#' @param u,vt The left/right singular vectors. Should both be \\code{NULL} or\n#' matrices of the same backend and fundamental type as \\code{x}.\n#' \n#' @examples\n#' suppressMessages(library(fmlr))\n#' x = cpumat(3, 2)\n#' x$fill_linspace(1, 6)\n#' \n#' s = cpuvec()\n#' linalg_svd(x, s)\n#' s$info()\n#' s$print()\n#' \n#' @rdname linalg-svd\n#' @name svd\n#' @useDynLib fmlr R_linalg_svd\n#' \n#' @export\nlinalg_svd = function(x, s, u=NULL, vt=NULL)\n{\n check_is_mat(x)\n check_is_vec(s)\n \n check_type_consistency(x, s)\n if (!is.null(u) && !is.null(vt))\n check_inputs(x, u, vt)\n else if (!is.null(u) || !is.null(vt))\n stop(\"must pass neither u and vt or both u and vt\")\n \n if (is.null(u))\n .Call(R_linalg_svd, get_backend(x), x$get_type(), x$data_ptr(), s$data_ptr(), NULL, NULL)\n else\n .Call(R_linalg_svd, get_backend(x), x$get_type(), x$data_ptr(), s$data_ptr(), u$data_ptr(), vt$data_ptr())\n \n invisible(NULL)\n}\n\n\n\n#' eigen\n#' \n#' Computes the eigenvalues and/or eigenvectors\n#' \n#' @details\n#' You will need to initialize the return objects \\code{values} and/or\n#' \\code{vectors}.\n#' manually. See the example.\n#' \n#' @param x Input data. The input values are overwritten.\n#' @param values The eigenvalues.\n#' @param vectors The eigenvectors.\n#' \n#' @rdname linalg-eigen\n#' @name eigen\n#' @useDynLib fmlr R_linalg_eigen_sym\n#' \n#' @export\nlinalg_eigen_sym = function(x, values, vectors=NULL)\n{\n check_is_mat(x)\n check_is_vec(values)\n check_type_consistency(x, values)\n if (!is.null(vectors))\n check_inputs(x, vectors)\n \n if (is.null(vectors))\n .Call(R_linalg_eigen_sym, get_backend(x), x$get_type(), x$data_ptr(), values$data_ptr(), NULL)\n else\n .Call(R_linalg_eigen_sym, get_backend(x), x$get_type(), x$data_ptr(), values$data_ptr(), vectors$data_ptr())\n \n invisible(NULL)\n}\n\n\n\n#' invert\n#' \n#' Invert a matrix.\n#' \n#' @param x Input data, overwritten by the inverse.\n#' @return Returns \\code{NULL}.\n#' \n#' @rdname linalg-invert\n#' @name invert\n#' @useDynLib fmlr R_linalg_invert\n#' \n#' @export\nlinalg_invert = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_invert, get_backend(x), x$get_type(), x$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' solve\n#' \n#' Solve a system of equations.\n#' \n#' @param x Input data. The input values are overwritten.\n#' @param y The RHS, overwritten by the solution.\n#' \n#' @rdname linalg-solve\n#' @name solve\n#' @useDynLib fmlr R_linalg_solve\n#' \n#' @export\nlinalg_solve = function(x, y)\n{\n check_is_mat(x)\n check_type_consistency(x, y)\n \n if (is_vec(y))\n {\n if (is_mpimat(x))\n stop(\"can not mix vector with mpimat\")\n \n class = CLASS_VEC\n }\n else\n {\n check_class_consistency(x, y)\n class = CLASS_MAT\n }\n \n .Call(R_linalg_solve, get_backend(x), x$get_type(), x$data_ptr(), class, y$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' qr\n#' \n#' Computes the compact QR factorization.\n#' \n#' @param x Input data. The input values are overwritten.\n#' @param qraux Auxilliary data for the compact QR.\n#' \n#' @rdname linalg-qr\n#' @name qr\n#' @useDynLib fmlr R_linalg_qr\n#' \n#' @export\nlinalg_qr = function(x, qraux)\n{\n check_is_mat(x)\n check_is_vec(qraux)\n check_type_consistency(x, qraux)\n \n .Call(R_linalg_qr, get_backend(x), x$get_type(), x$data_ptr(), qraux$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' qr_Q\n#' \n#' Computes the Q matrix from the compact QR factorization.\n#' \n#' @param QR The compact QR factorization. The return from \\code{qr()}.\n#' @param qraux Auxilliary data for the compact QR.\n#' @param Q The output Q matrix.\n#' @param work A workspace vector.\n#' \n#' @rdname linalg-qr-Q\n#' @name qr_Q\n#' @useDynLib fmlr R_linalg_qr_Q\n#' \n#' @export\nlinalg_qr_Q = function(QR, qraux, Q, work)\n{\n check_inputs(QR, Q)\n check_is_vec(qraux)\n check_is_vec(work)\n check_type_consistency(QR, qraux, work)\n \n .Call(R_linalg_qr_Q, get_backend(QR), QR$get_type(), QR$data_ptr(), qraux$data_ptr(), Q$data_ptr(), work$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' qr_R\n#' \n#' Computes the R matrix from the compact QR factorization.\n#' \n#' @param QR The compact QR factorization. The return from \\code{qr()}.\n#' @param R The output R matrix.\n#' \n#' @rdname linalg-qr-R\n#' @name qr_R\n#' @useDynLib fmlr R_linalg_qr_R\n#' \n#' @export\nlinalg_qr_R = function(QR, R)\n{\n check_inputs(QR, R)\n .Call(R_linalg_qr_R, get_backend(QR), QR$get_type(), QR$data_ptr(), R$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' lq\n#' \n#' Computes the compact LQ factorization.\n#' \n#' @param x Input data. The input values are overwritten.\n#' @param lqaux Auxilliary data for the compact LQ.\n#' \n#' @rdname linalg-lq\n#' @name lq\n#' @useDynLib fmlr R_linalg_lq\n#' \n#' @export\nlinalg_lq = function(x, lqaux)\n{\n check_is_mat(x)\n check_is_vec(lqaux)\n check_type_consistency(x, lqaux)\n \n .Call(R_linalg_lq, get_backend(x), x$get_type(), x$data_ptr(), lqaux$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' lq_L\n#' \n#' Computes the L matrix from the compact LQ factorization.\n#' \n#' @param LQ The compact LQ factorization. The return from \\code{lq()}.\n#' @param L The output L matrix.\n#' \n#' @rdname linalg-qr-R\n#' @name lq_L\n#' @useDynLib fmlr R_linalg_lq_L\n#' \n#' @export\nlinalg_lq_L = function(LQ, L)\n{\n check_inputs(LQ, L)\n .Call(R_linalg_lq_L, get_backend(LQ), LQ$get_type(), LQ$data_ptr(), L$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' lq_Q\n#' \n#' Computes the Q matrix from the compact LQ factorization.\n#' \n#' @param LQ The compact LQ factorization. The return from \\code{lq()}.\n#' @param lqaux Auxilliary data for the compact LQ.\n#' @param Q The output Q matrix.\n#' @param work A workspace vector.\n#' \n#' @rdname linalg-lq-Q\n#' @name lq_Q\n#' @useDynLib fmlr R_linalg_lq_Q\n#' \n#' @export\nlinalg_lq_Q = function(LQ, lqaux, Q, work)\n{\n check_inputs(LQ, Q)\n check_is_vec(lqaux)\n check_is_vec(work)\n check_type_consistency(LQ, lqaux, work)\n \n .Call(R_linalg_lq_Q, get_backend(LQ), LQ$get_type(), LQ$data_ptr(), lqaux$data_ptr(), Q$data_ptr(), work$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' qrsvd\n#' \n#' QR/LQ-based SVD. Useful for very tall/skinny or short/wide data.\n#' \n#' @param x Input data. The input values are overwritten.\n#' @param s Singular values.\n#' @param u,vt The left/right singular vectors. Should both be \\code{NULL} or\n#' matrices of the same backend and fundamental type as \\code{x}.\n#' \n#' @rdname linalg-qrsvd\n#' @name qrsvd\n#' @useDynLib fmlr R_linalg_qrsvd\n#' \n#' @export\nlinalg_qrsvd = function(x, s, u=NULL, vt=NULL)\n{\n check_is_mat(x)\n check_is_vec(s)\n \n check_type_consistency(x, s)\n if (!is.null(u) && !is.null(vt))\n check_inputs(x, u, vt)\n else if (!is.null(u) || !is.null(vt))\n stop(\"must pass neither u and vt or both u and vt\")\n \n if (is.null(u))\n .Call(R_linalg_qrsvd, get_backend(x), x$get_type(), x$data_ptr(), s$data_ptr(), NULL, NULL)\n else\n .Call(R_linalg_qrsvd, get_backend(x), x$get_type(), x$data_ptr(), s$data_ptr(), u$data_ptr(), vt$data_ptr())\n \n invisible(NULL)\n}\n\n\n\n#' cpsvd\n#' \n#' \"Crossproducts\" SVD.\n#' \n#' @details\n#' Computes the approximate SVD via the eigenvalue decomposition of\n#' \\code{crossprod(x)} if the input is tall/skinny and \\code{tcrossprod(x)}\n#' otherwise.\n#' \n#' @param x Input data. The input values are overwritten.\n#' @param s Singular values.\n#' @param u,vt The left/right singular vectors. Should both be \\code{NULL} or\n#' matrices of the same backend and fundamental type as \\code{x}.\n#' \n#' @rdname linalg-cpsvd\n#' @name cpsvd\n#' @useDynLib fmlr R_linalg_cpsvd\n#' \n#' @export\nlinalg_cpsvd = function(x, s, u=NULL, vt=NULL)\n{\n check_is_mat(x)\n check_is_vec(s)\n \n check_type_consistency(x, s)\n if (!is.null(u) && !is.null(vt))\n check_inputs(x, u, vt)\n else if (!is.null(u) || !is.null(vt))\n stop(\"must pass neither u and vt or both u and vt\")\n \n if (is.null(u))\n .Call(R_linalg_cpsvd, get_backend(x), x$get_type(), x$data_ptr(), s$data_ptr(), NULL, NULL)\n else\n .Call(R_linalg_cpsvd, get_backend(x), x$get_type(), x$data_ptr(), s$data_ptr(), u$data_ptr(), vt$data_ptr())\n \n invisible(NULL)\n}\n\n\n\n#' chol\n#' \n#' Compute the lower-triangular Cholesky factor.\n#' \n#' @param x Input data, overwritten by the inverse.\n#' @return Returns \\code{NULL}.\n#' \n#' @rdname linalg-chol\n#' @name chol\n#' @useDynLib fmlr R_linalg_chol\n#' \n#' @export\nlinalg_chol = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_chol, get_backend(x), x$get_type(), x$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' norms\n#' \n#' Norms.\n#' \n#' @param x Input data. The data is un-modified except for \\code{norm_2()}.\n#' \n#' @return The requested norm.\n#' \n#' @rdname linalg-norm\n#' @name norm\n#' \n#' @useDynLib fmlr R_linalg_norm\nNULL\n\n#' @rdname linalg-norm\n#' @export\nlinalg_norm_1 = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_norm, get_backend(x), x$get_type(), x$data_ptr(), \"1\")\n}\n\n#' @rdname linalg-norm\n#' @export\nlinalg_norm_I = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_norm, get_backend(x), x$get_type(), x$data_ptr(), \"I\")\n}\n\n#' @rdname linalg-norm\n#' @export\nlinalg_norm_F = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_norm, get_backend(x), x$get_type(), x$data_ptr(), \"F\")\n}\n\n#' @rdname linalg-norm\n#' @export\nlinalg_norm_M = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_norm, get_backend(x), x$get_type(), x$data_ptr(), \"M\")\n}\n\n#' @rdname linalg-norm\n#' @export\nlinalg_norm_2 = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_norm, get_backend(x), x$get_type(), x$data_ptr(), \"2\")\n}\n\n\n\n#' Condition Number\n#' \n#' Condition numbers.\n#' \n#' @param x Input data. The data is modified in each case.\n#' \n#' @return The requested condition number.\n#' \n#' @rdname linalg-cond\n#' @name cond\n#' \n#' @useDynLib fmlr R_linalg_cond\nNULL\n\n#' @rdname linalg-cond\n#' @export\nlinalg_cond_1 = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_cond, get_backend(x), x$get_type(), x$data_ptr(), \"1\")\n}\n\n#' @rdname linalg-cond\n#' @export\nlinalg_cond_I = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_cond, get_backend(x), x$get_type(), x$data_ptr(), \"I\")\n}\n\n#' @rdname linalg-cond\n#' @export\nlinalg_cond_2 = function(x)\n{\n check_is_mat(x)\n .Call(R_linalg_cond, get_backend(x), x$get_type(), x$data_ptr(), \"2\")\n}\n\n\n\n#' dot\n#' \n#' Compute vector dot product i.e. \\code{sum(x*y)}.\n#' \n#' @param x,y Input data.\n#' class and type as \\code{x}.\n#' @return Returns the dot product.\n#' \n#' @rdname linalg-dot\n#' @name dot\n#' \n#' @useDynLib fmlr R_linalg_dot\n#' \n#' @export\nlinalg_dot = function(x, y=NULL)\n{\n check_is_vec(x)\n \n if (!is.null(y))\n {\n check_is_vec(y)\n check_backend_consistency(x, y)\n .Call(R_linalg_dot, get_backend(x), x$get_type(), x$data_ptr(), y$data_ptr())\n }\n else\n .Call(R_linalg_dot, get_backend(x), x$get_type(), x$data_ptr(), NULL)\n}\n\n\n\n#' trinv\n#' \n#' Invert a triangular matrix.\n#' \n#' @param upper Is the matrix upper triangular? Otherwise only the lower\n#' triangle will be referenced.\n#' @param unit_diag Is the matrix unit diagonal?\n#' @param x Input data, overwritten by the inverse.\n#' @return Returns \\code{NULL}.\n#' \n#' @rdname linalg-trinv\n#' @name trinv\n#' @useDynLib fmlr R_linalg_trinv\n#' \n#' @export\nlinalg_trinv = function(upper, unit_diag, x)\n{\n check_is_mat(x)\n \n upper = as.logical(upper)\n unit_diag = as.logical(unit_diag)\n \n .Call(R_linalg_trinv, get_backend(x), x$get_type(), upper, unit_diag, x$data_ptr())\n invisible(NULL)\n}\n\n\n\n#' rsvd\n#' \n#' SVD approximation via random projections.\n#' \n#' @param seed Seed for the random generator.\n#' @param k The number of components to estimate. Integer > 0.\n#' @param q Exponent (see paper if you really care). Values of 1 or 2 are good.\n#' @param x Input data. The input values are overwritten.\n#' @param s Singular values.\n#' @param u,vt The left/right singular vectors. Should both be \\code{NULL} or\n#' matrices of the same backend and fundamental type as \\code{x}.\n#' \n#' @references Halko, Nathan, Per-Gunnar Martinsson, and Joel A. Tropp. \"Finding\n#' structure with randomness: Probabilistic algorithms for constructing\n#' approximate matrix decompositions.\" SIAM review 53, no. 2 (2011): 217-288.\n#' \n#' @rdname linalg-rsvd\n#' @name rsvd\n#' @useDynLib fmlr R_linalg_rsvd\n#' \n#' @export\nlinalg_rsvd = function(seed, k, q, x, s, u=NULL, vt=NULL)\n{\n seed = as.integer(seed)\n k = as.integer(k)\n q = as.integer(q)\n \n check_is_mat(x)\n check_is_vec(s)\n \n check_type_consistency(x, s)\n if (!is.null(u) && !is.null(vt))\n check_inputs(x, u, vt)\n else if (!is.null(u) || !is.null(vt))\n stop(\"must pass neither u and vt or both u and vt\")\n \n if (is.null(u))\n .Call(R_linalg_rsvd, get_backend(x), x$get_type(), seed, k, q, x$data_ptr(), s$data_ptr(), NULL, NULL)\n else\n .Call(R_linalg_rsvd, get_backend(x), x$get_type(), seed, k, q, x$data_ptr(), s$data_ptr(), u$data_ptr(), vt$data_ptr())\n \n invisible(NULL)\n}\n", "meta": {"hexsha": "4a2a4c508b6bcd57adff06495b298a263f3bca97", "size": 17680, "ext": "r", "lang": "R", "max_stars_repo_path": "R/linalg.r", "max_stars_repo_name": "fml-fam/fmlr", "max_stars_repo_head_hexsha": "7a9c8030435b9921fc832b27ef5f174a40c7792b", "max_stars_repo_licenses": ["BSL-1.0"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-02-06T21:06:14.000Z", "max_stars_repo_stars_event_max_datetime": "2020-06-23T22:34:08.000Z", "max_issues_repo_path": "R/linalg.r", "max_issues_repo_name": "wrathematics/fmlr", "max_issues_repo_head_hexsha": "7a9c8030435b9921fc832b27ef5f174a40c7792b", "max_issues_repo_licenses": ["BSL-1.0"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2020-02-19T17:27:46.000Z", "max_issues_repo_issues_event_max_datetime": "2020-06-09T00:30:36.000Z", "max_forks_repo_path": "R/linalg.r", "max_forks_repo_name": "wrathematics/fmlr", "max_forks_repo_head_hexsha": "7a9c8030435b9921fc832b27ef5f174a40c7792b", "max_forks_repo_licenses": ["BSL-1.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 21.6401468788, "max_line_length": 152, "alphanum_fraction": 0.6642533937, "num_tokens": 5548, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.785308580887758, "lm_q2_score": 0.6723317057447908, "lm_q1q2_score": 0.5279878577242874}} {"text": " C02AMJ Example Program Results\n\n Roots of cubic equation Error estimates\n (machine-dependent)\n\n z = -2.0000E+00 +3.0000E+00*i 1.3E-15\n z = 1.0000E+00 -2.0000E+00*i 2.2E-15\n z = 3.0000E+00 -4.0000E+00*i 2.1E-15\n", "meta": {"hexsha": "c2c70b446dfb0af350db8fc96811db651b5b6602", "size": 294, "ext": "r", "lang": "R", "max_stars_repo_path": "simple_examples/baseresults/c02amje.r", "max_stars_repo_name": "numericalalgorithmsgroup/NAGJavaExamples", "max_stars_repo_head_hexsha": "f625b3f043c5c14a88d7ecbc04374acf75d63c82", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2021-07-03T22:53:20.000Z", "max_stars_repo_stars_event_max_datetime": "2021-08-04T01:44:03.000Z", "max_issues_repo_path": "simple_examples/baseresults/c02amje.r", "max_issues_repo_name": "numericalalgorithmsgroup/NAGJavaExamples", "max_issues_repo_head_hexsha": "f625b3f043c5c14a88d7ecbc04374acf75d63c82", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "simple_examples/baseresults/c02amje.r", "max_forks_repo_name": "numericalalgorithmsgroup/NAGJavaExamples", "max_forks_repo_head_hexsha": "f625b3f043c5c14a88d7ecbc04374acf75d63c82", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2021-07-03T22:55:16.000Z", "max_forks_repo_forks_event_max_datetime": "2022-02-02T01:00:53.000Z", "avg_line_length": 32.6666666667, "max_line_length": 55, "alphanum_fraction": 0.4965986395, "num_tokens": 112, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7772998611746911, "lm_q2_score": 0.679178692681616, "lm_q1q2_score": 0.5279255035342283}} {"text": "# Get command line args with file names for processing\nargs <- commandArgs(TRUE)\n\n# MBE PG functions\nsource(\"/ufrc/barbazuk/lboat/T.lamottei_T.crocifolius/Complete_data_set/align_reads_for_samcompare/scripts/emp_bayesian/Subroutines_model2_experimental.R\")\nsource(\"/ufrc/barbazuk/lboat/T.lamottei_T.crocifolius/Complete_data_set/align_reads_for_samcompare/scripts/emp_bayesian/AI_poissongamma_functions.R\")\n\n# Prepare output file\nfileout = args[2]\nheaders_out = \"commonID,q4_mean_theta,q4_q025,q4_q975,q5_mean_theta,q5_q025,q5_q975,q6_mean_theta,q6_q025,q6_q975\"\ncat(headers_out,file=fileout,append=FALSE,sep=\"\\n\")\n\n# Make Connection to input\ncon = file(args[1],\"r\")\nnewline<-readLines(con,n=1) # Go to header line\nheaders_in=strsplit(newline,split=\",\")[[1]]\n\nmydata=rep(NA,length(headers_in))\nnames(mydata)=headers_in\nm=1\n\nwhile(length(newline) != 0 ){\n # Move past fusions that are not flag_analyze\n flaganalyze=0\n while(flaganalyze==0){\n newline<-readLines(con,n=1) \n if(length(newline)==0){break}\n mydata<-as.vector(strsplit(newline,split=\",\")[[1]])\n names(mydata)=headers_in\n flaganalyze=as.numeric(mydata[\"flag_analyze\"])\n }\n if(length(newline)==0){break}\n\n print(paste(\"------------Analyzing\",mydata['line'],mydata['mating_status'],mydata['fusion_id'], \"--------------\"));\n m=m+1\n X_RNA <- as.numeric(mydata[c(\"LINE_TOTAL_1\",\"LINE_TOTAL_2\",\"LINE_TOTAL_3\",\"LINE_TOTAL_4\",\"LINE_TOTAL_5\")])\n X_RNA <- X_RNA[!is.na(X_RNA)]\n Y_RNA <- as.numeric(mydata[c(\"TESTER_TOTAL_1\",\"TESTER_TOTAL_2\",\"TESTER_TOTAL_3\",\"TESTER_TOTAL_4\",\"TESTER_TOTAL_5\")])\n Y_RNA <- Y_RNA[!is.na(Y_RNA)]\n n_i <- X_RNA + Y_RNA\n\n # Create storage vectors for Quantiles and posterior means for theta (proportion of male reads) under different models\n means=q_025=q_975=rep(NA,3)\n names(means)=names(q_025)=names(q_975)=c(\"PG_4\",\"PG_5\",\"PG_6\")\n\n #Poisson Gamma models q = .4\n tem_PG=gibbs_poissongamma(nsim=1000,nburnin=1000,lag=10,x=X_RNA,y=Y_RNA,both=c(0,0,0),a_mu=1/2,b_mu=1/2,a_alpha=1/2,b_alpha=1/2,a_beta=1/2,b_beta=1/2,\n q_=.4,#<-------As before but with q=q_sim \n abundance=FALSE)\n\n thetas=tem_PG$alphas/(1+tem_PG$alphas)\n q_025[\"PG_4\"]= quantile(thetas,c(.025),na.rm=TRUE)\n q_975[\"PG_4\"]= quantile(thetas,c(.975),na.rm=TRUE)\n means[\"PG_4\"]= mean(thetas)\n\n #Poisson Gamma models q = .5\n tem_PG=gibbs_poissongamma(nsim=1000,nburnin=1000,lag=10,x=X_RNA,y=Y_RNA,both=c(0,0,0),a_mu=1/2,b_mu=1/2,a_alpha=1/2,b_alpha=1/2,a_beta=1/2,b_beta=1/2,\n q_=.5,#<-------As before but with q=q_sim \n abundance=FALSE)\n\n thetas=tem_PG$alphas/(1+tem_PG$alphas)\n q_025[\"PG_5\"]= quantile(thetas,c(.025),na.rm=TRUE)\n q_975[\"PG_5\"]= quantile(thetas,c(.975),na.rm=TRUE)\n means[\"PG_5\"]= mean(thetas)\n\n #Poisson Gamma models q = .6\n tem_PG=gibbs_poissongamma(nsim=1000,nburnin=1000,lag=10,x=X_RNA,y=Y_RNA,both=c(0,0,0),a_mu=1/2,b_mu=1/2,a_alpha=1/2,b_alpha=1/2,a_beta=1/2,b_beta=1/2,\n q_=.6,#<-------As before but with q=q_sim \n abundance=FALSE)\n\n thetas=tem_PG$alphas/(1+tem_PG$alphas)\n q_025[\"PG_6\"]= quantile(thetas,c(.025),na.rm=TRUE)\n q_975[\"PG_6\"]= quantile(thetas,c(.975),na.rm=TRUE)\n means[\"PG_6\"]= mean(thetas)\n\n # Create output and write to table\n SNPout = paste(mydata[\"commonID\"],paste(round(c(means[\"PG_4\"],q_025[\"PG_4\"],q_975[\"PG_4\"]),3),collapse=\",\"),paste(round(c(means[\"PG_5\"],q_025[\"PG_5\"],q_975[\"PG_5\"]),3),collapse=\",\"),paste(round(c(means[\"PG_6\"],q_025[\"PG_6\"],q_975[\"PG_6\"]),3),collapse=\",\"),sep=\",\")\n cat(SNPout,file=fileout,append=TRUE,sep=\"\\n\")\n}\n", "meta": {"hexsha": "1ad997617db9ff11ae203f52e1a9d6e24fd17f17", "size": 3581, "ext": "r", "lang": "R", "max_stars_repo_path": "scripts/emp_bayesian/PG_model_empirical_q456_5reps.r", "max_stars_repo_name": "jlboat/Tragopogon_castellanus", "max_stars_repo_head_hexsha": "722944a260d965e20a0d008ae88f69061e9b205d", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "scripts/emp_bayesian/PG_model_empirical_q456_5reps.r", "max_issues_repo_name": "jlboat/Tragopogon_castellanus", "max_issues_repo_head_hexsha": "722944a260d965e20a0d008ae88f69061e9b205d", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "scripts/emp_bayesian/PG_model_empirical_q456_5reps.r", "max_forks_repo_name": "jlboat/Tragopogon_castellanus", "max_forks_repo_head_hexsha": "722944a260d965e20a0d008ae88f69061e9b205d", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 44.7625, "max_line_length": 268, "alphanum_fraction": 0.6889137113, "num_tokens": 1282, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104789178257654, "lm_q2_score": 0.6513548511303336, "lm_q1q2_score": 0.5279093748646754}} {"text": "model_snowdensity <- function (ps_t1 = 0.0,\n Sdepth_t1 = 0.0,\n Sdry_t1 = 0.0,\n Swet_t1 = 0.0){\n #'- Name: SnowDensity -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: Density of snow cover calculation\n #' * Author: STICS\n #' * Reference: doi:http://dx.doi.org/10.1016/j.agrformet.2014.05.002\n #' * Institution: INRA\n #' * Abstract: density of snow cover\n #'- inputs:\n #' * name: ps_t1\n #' ** description : density of snow cover in previous day\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 100.0\n #' ** unit : kg/m**3\n #' ** uri : \n #' * name: Sdepth_t1\n #' ** description : snow cover depth Calculation in previous day\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 5000.0\n #' ** unit : m\n #' ** uri : \n #' * name: Sdry_t1\n #' ** description : water in solid state in the snow cover in previous day\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 500.0\n #' ** unit : mmW\n #' ** uri : \n #' * name: Swet_t1\n #' ** description : water in liquid state in the snow cover\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 100.0\n #' ** unit : mmW\n #' ** uri : \n #'- outputs:\n #' * name: ps\n #' ** description : density of snow cover\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** min : 0.0\n #' ** max : 500.0\n #' ** unit : kg/m**3\n #' ** uri : \n ps <- 0.0\n if (abs(Sdepth_t1) > 0.0)\n {\n if (abs(Sdry_t1 + Swet_t1) > 0.0)\n {\n ps <- (Sdry_t1 + Swet_t1) / Sdepth_t1\n }\n else\n {\n ps <- ps_t1\n }\n }\n return (list('ps' = ps))\n}", "meta": {"hexsha": "81ce373d746e24fdb3fe9337342659a26eb5abab", "size": 3255, "ext": "r", "lang": "R", "max_stars_repo_path": "src/r/STICS_SNOW/Snowdensity.r", "max_stars_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_stars_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/r/STICS_SNOW/Snowdensity.r", "max_issues_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_issues_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/r/STICS_SNOW/Snowdensity.r", "max_forks_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_forks_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 43.4, "max_line_length": 103, "alphanum_fraction": 0.3078341014, "num_tokens": 736, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104788903594355, "lm_q2_score": 0.651354857898194, "lm_q1q2_score": 0.5279093624595561}} {"text": "#'@title calcSpeedUnitConversion\n#'\n#'@description Convert speed in knots to m/s.\n#'\n#'@param shipSpeed Ship speed (vector of numericals, knots)\n#'\n#'@return speed (vector of numericals, m/s)\n#'\n#'@examples\n#'calcSpeedUnitConversion(seq(10,15,1))\n#'\n#'@export\n\ncalcSpeedUnitConversion<- function(shipSpeed){\n\n speed<- shipSpeed*0.5144\n\n return(speed)\n}\n", "meta": {"hexsha": "5c690a3e9f5275f0bfa05a9b0775122ce7400443", "size": 355, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcSpeedUnitConversion.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcSpeedUnitConversion.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcSpeedUnitConversion.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 17.75, "max_line_length": 59, "alphanum_fraction": 0.7154929577, "num_tokens": 103, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.787931185683219, "lm_q2_score": 0.668880247169804, "lm_q1q2_score": 0.5270316062325883}} {"text": "# This is a special version of K-dO rq modified to compute the LR process\n# Note that the sol array is now p+3 by J and the third row contains the\n# value of the objective function at the tau specified in row one.\n#\n# This software is in the public domain and may be freely\n# used and redistributed for non-commercial purposes. No guarantees\n# are offered or implied. comments, bug reports, etc are welcome\n# and should be sent to roger@ysidro.econ.uiuc.edu or to\n#\n# roger koenker\n# department of economics\n# university of illinois\n# champaign, illinois, 61820\n#\n#\n z <- .Fortran(\"rqbr\", as.integer(n), as.integer(p), as.integer(n +\n 5), as.integer(p + 3), as.integer(p + 4), as.double(x),\n as.double(y), as.double(tau), as.double(tol), flag = as.integer(1),\n coef = double(p), resid = double(n), integer(n), double((n +\n 5) * (p + 4)), double(n), as.integer(nsol), as.integer(ndsol),\n sol = double((p + 3) * nsol), dsol = double(n * ndsol),\n lsol = as.integer(0), h = integer(p * nsol), qn = as.double(qn),\n cutoff = as.double(cutoff), ci = double(4 * p), tnmat = double(4 *\n p), as.double(big), as.logical(lci1))\n\n# If I can get this to work and not crash\n# then should just be a matter of porting the wrapper into a SKlearn style interface\n# for br and fn. ppro is harder because it is more of an r source port.\n\n# Also important to use 'conquer' package as that's reputed to be better than BR/FN\n# And updated preprocessing when available\n\nCf2py intent(in) m\nCf2py intent(in) nn\nCf2py intent(in) m5\nCf2py intent(in) n3\nCf2py intent(in) n4\nCf2py intent(in) a\nCf2py intent(in) b\nCf2py intent(in) t\nCf2py intent(in) toler\nCf2py intent(out) ift\nCf2py intent(out) x\nCf2py intent(out) e\nCf2py intent(out) s\nCf2py intent(out) wa\nCf2py intent(out) wb\nCf2py intent(out) nsol\nCf2py intent(out) ndsol\nCf2py intent(out) sol\nCf2py intent(out) dsol\nCf2py intent(out) lsol\nCf2py intent(out) h\nCf2py intent(out) qn\nCf2py intent(out) cutoff\nCf2py intent(out) ci\nCf2py intent(out) tnmat\nCf2py intent(out) big\nCf2py intent(out) lci1\n\nsubroutine rqbr(m,nn,m5,n3,n4,a,b,t,toler,ift,x,e,s,wa,wb,nsol,ndsol,sol,dsol,lsol,h,qn,cutoff,ci,tnmat,big,lci1)\n#\n# m = number of observations\n# n = number of parameters\n# m5 = m+5\n# n3 = n+3\n# n4 = n+4\n# a is the x matrix\n# b is the y vector\n# t, the desired quantile\n# toler, smallest detectable |x-y|/x machine precision to the 2/3\n# ift exit code:\n# \t\t0-ok\n# \t\telse dimensions inconsistent see below\n# x the parameter estimate betahat\n# e is the residual vector\n# s is an integer work array\n# wa is a real work array\n# wb is another real work array\n# nsol is an estimated (row) dimension of the primal solution array say 3*m; minimum value = 2\n# ndsol is an estimated (row) dimension of the dual solution array say 3*m; minimum value = 2\n# sol is the primal solution array\n# dsol is the dual solution arry\n# lsol is the actual dimension of the solution arrays\n# h is the matrix of basic observations indices\n# qn is the vector of residuals variances from the projection of each column of the x matrix on the remaining columns\n# cutoff, the critical point for N(0,1)\n# ci is the matrix of confidence intervals\n# tnmat is the matrix of the JGPK rank test statistics\n# big, large positive finite floating-point number\n# utilization: if you just want a solution at a single quantile you\n# neednt bother with sol, nsol, etc, if you want all the solutions\n# then set theta to something <0 and sol and dsol will return all the\n# estimated quantile solutions.\n# the algorithm is a slightly modified version of algorithm as 229\n# described in koenker and dorey, computing regression quantiles,\n# applied statistics, pp. 383-393.\n#\ninteger i,j,k,kl,kount,kr,l,lsol,m,m1,m2,m3,m4,m5,ift\ninteger n,n1,n2,n3,nsol,ndsol,out,s(m),h(nn,nsol)\ninteger nn,n4,idxcf\nlogical stage,test,init,iend,lup\nlogical lci1,lci2,skip\ndouble precision a1,aux,b1,big,d,dif,pivot,smax,t,t0,t1,tnt,dk\ndouble precision min,max,toler,zero,half,one,two\ndouble precision b(m),sol(n3,nsol),a(m,nn),x(nn),wa(m5,n4),wb(m)\ndouble precision sum,e(m),dsol(m,ndsol)\ndouble precision qn(nn),cutoff,ci(4,nn),tnmat(4,nn),tnew,told,tn\nparameter( zero = 0.d0)\nparameter( one = 1.d0)\nparameter( two = 2.d0)\n#\n# check dimension parameters\n#\nn=nn\nift = 0\nwa(m+2,nn+1) = one\nif (m5!=m+5)\n ift = 3\nif (n3!=n+3)\n ift = 4\nif (n4!=n+4)\n ift = 5\nif (m<=zero||n<=zero)\n ift = 6\nif (ift<=two) {\n#\n# initialization\n#\n half = one/two\n iend = .true.\n lci2 = .false.\n lup = .true.\n skip = .false.\n idxcf = 0\n tnew = zero\n tn = zero\n m1 = m+1\n n1 = n+1\n n2 = n+2\n m2 = m+2\n m3 = m+3\n m4 = m+4\n do j = 1,n{\n x(j) = zero\n }\n do i = 1,m\n e(i) = zero\n if (tone) {\n t0 = one/(two*float(m))\n t1 = one-t0\n t = t0\n iend = .false.\n lci1 = .false.\n }\n repeat { #0\n do i = 1,m {\n k = 1\n do j = 1,nn\n if (k<=nn){\n if(j==idxcf)\n skip = .true.\n else\n skip = .false.\n if(!skip){\n wa(i,k) = a(i,j)\n k = k+1\n }\n } \n wa(i,n4) = n+i\n wa(i,n2) = b(i)\n if (idxcf != 0)\n wa(i,n3) = tnew*a(i,idxcf)\n else\n wa(i,n3) = zero\n wa(i,n1) = wa(i,n2)-wa(i,n3)\n if (wa(i,n1)(-two))\n next 1\n d = -d-two\n }\n if (d>max) {\n max = d\n in = j\n }\n }\n if (max<=toler)\n break 2\n if (wa(m1,in)<=zero) {\n do i = 1,m4\n wa(i,in) = -wa(i,in)\n wa(m1,in) = wa(m1,in)-two\n wa(m2,in) = wa(m2,in)-two\n }\n repeat { #4\n#\n# determine the vector to leave the basis\n#\n k = 0\n do i = kl,m {\n d = wa(i,in)\n if (d>toler) {\n k = k+1\n wb(k) = wa(i,n1)/d\n s(k) = i\n test = .true.\n }\n }\n repeat { #5\n if (k<=0)\n test = .false.\n else {\n min = big\n do i = 1,k\n if (wb(i)max) {\n max = d\n in = j\n }\n }\n if (wa(m1,in)zero)\n dsol(kd,lsol) = one\n }\n if (!lci2){\n sol(1,lsol) = smax\n sol(2,lsol) = sum\n sum = zero\n do j=kl,m{\n d = wa(j,n1)*sign(one,wa(j,n4))\n sum = sum + d*(smax + half*(sign(one,d) - one))\n\t }\n sol(3,lsol) = sum\n do i=1,m\n dsol(i,lsol+1) = dsol(i,lsol)\n }\n if (lci2){\n# compute next theta\n a1 = zero\n do i = 1,m {\n a1 = a1+a(i,idxcf)*(dsol(i,lsol)+t-one)\n }\n tn = a1/sqrt(qn(idxcf)*t*(one-t))\n if (abs(tn)tnew)\n if (tntsmax){\n smax = tnt\n out = i\n }\n }\n }\n if (lup){\n told = tnew \n tnew = smax + toler\n ci(3,idxcf) = told - toler\n tnmat(3,idxcf) = tn\n if (!(tnew < big-toler)){\n ci(3,idxcf) = big\n ci(4,idxcf) = big\n tnmat(3,idxcf) = tn\n tnmat(4,idxcf) = tn\n lup = .false.\n go to 70\n }\n }\n else{\n told = tnew \n tnew = smax - toler\n ci(2,idxcf) = told + toler\n tnmat(2,idxcf) = tn\n if (!(tnew > -big+toler)){\n ci(2,idxcf) = -big\n ci(1,idxcf) = -big\n tnmat(2,idxcf) = tn\n tnmat(1,idxcf) = tn\n lup = .true.\n go to 60\n }\n }\n#update the new marginal cost\n do i = 1,m{\n wa(i,n3) = wa(i,n3)/told*tnew\n wa(i,n1) = wa(i,n2) - wa(i,n3)\n }\n do j = kr,n3{\n d = wa(out,j)\n wa(m1,j) = wa(m1,j) -d -d\n wa(m2,j) = wa(m2,j) -d -d\n wa(out,j) = -d\n }\n wa(out,n4) = -wa(out,n4)\n init = .true.\n }\n else{\n if (lup){\n ci(4,idxcf) = tnew - toler\n tnmat(4,idxcf) = tn\n lup = .false.\n go to 70\n }\n else{\n ci(1,idxcf) = tnew + toler\n tnmat(1,idxcf) = tn\n lup = .true.\n go to 60\n }\n }\n }\n if ((iend)&&(!lci2))\n go to 40\n if (!lci2){\n init = .true.\n lsol = lsol+1\n do i = 1,m\n s(i) = zero\n do j = 1,n\n x(j) = zero\n#\n# compute next t\n#\n smax = two\n do j = 1,n {\n b1 = wa(m3,j)\n a1 = (-two-wa(m2,j))/b1\n b1 = -wa(m2,j)/b1\n if (a1>=t)\n if (a1t)\n if (b1=t1+toler)\n iend = .true.\n t = tnt\n if (iend)\n t = t1\n }\n } #1\n wa(m2,nn+1) = two\n ift = 2\n go to 50\n 40 if (lsol>2) {\n sol(1,1) = zero\n #sol(2,1) = zero\n sol(3,1) = zero\n sol(1,lsol) = one\n #sol(2,lsol) = zero\n sol(3,lsol) = zero\n do i = 1,m {\n dsol(i,1) = one\n dsol(i,lsol) = zero\n dsol(i,lsol+1) = zero\n }\n }\n l = kl-1\n do i = 1,l\n if (wa(i,n1)nn){\n break 1\n\t}\n 70 if (lup){\n tnew = x(idxcf)+toler\n told = tnew\n ci(3,idxcf) = x(idxcf)\n tnmat(3,idxcf) = zero\n }\n else{\n tnew = x(idxcf)-toler\n told = tnew\n ci(2,idxcf) = x(idxcf)\n tnmat(2,idxcf) = zero\n }\n }\n } #0\n # restore the original value of dual when ci is true\n do i=1,m\n dsol(i,lsol) = dsol(i,lsol+1)\n }\nreturn\nend\n", "meta": {"hexsha": "d434f8a97329e9b90725cfa5df787a0dcf15a076", "size": 15764, "ext": "r", "lang": "R", "max_stars_repo_path": "fortran/ratfor/rqbr.r", "max_stars_repo_name": "mhoward2718/quantreg", "max_stars_repo_head_hexsha": "48478aee30b3dcbd6204c287f99cb287f005d6c1", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2022-01-02T22:15:54.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-02T22:15:54.000Z", "max_issues_repo_path": "fortran/ratfor/rqbr.r", "max_issues_repo_name": "mhoward2718/quantreg", "max_issues_repo_head_hexsha": "48478aee30b3dcbd6204c287f99cb287f005d6c1", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "fortran/ratfor/rqbr.r", "max_forks_repo_name": "mhoward2718/quantreg", "max_forks_repo_head_hexsha": "48478aee30b3dcbd6204c287f99cb287f005d6c1", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 25.674267101, "max_line_length": 121, "alphanum_fraction": 0.4314894697, "num_tokens": 5210, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.867035752930664, "lm_q2_score": 0.6076631698328916, "lm_q1q2_score": 0.5268656939842952}} {"text": "# Latent class models with one (loglinear independence) to three classes\ndata(election)\nf <- cbind(MORALG,CARESG,KNOWG,LEADG,DISHONG,INTELG,\n MORALB,CARESB,KNOWB,LEADB,DISHONB,INTELB)~1\nnes1 <- poLCA(f,election,nclass=1) # log-likelihood: -18647.31\nnes2 <- poLCA(f,election,nclass=2) # log-likelihood: -17344.92\nnes3 <- poLCA(f,election,nclass=3) # log-likelihood: -16714.66\n# Three-class model with a single covariate (party)\nf2a <- cbind(MORALG,CARESG,KNOWG,LEADG,DISHONG,INTELG,\n MORALB,CARESB,KNOWB,LEADB,DISHONB,INTELB)~PARTY\nnes2a <- poLCA(f2a,election,nclass=3,nrep=5) # log-likelihood: -16222.32\npidmat <- cbind(1,c(1:7))\nexb <- exp(pidmat %*% nes2a$coeff)\nmatplot(c(1:7),(cbind(1,exb)/(1+rowSums(exb))),ylim=c(0,1),type=\"l\",\n main=\"Party ID as a predictor of candidate affinity class\",\n xlab=\"Party ID: strong Democratic (1) to strong Republican (7)\",\n ylab=\"Probability of latent class membership\",lwd=2,col=1)\ntext(5.9,0.35,\"Other\")\ntext(5.4,0.7,\"Bush affinity\")\ntext(1.8,0.6,\"Gore affinity\")\n", "meta": {"hexsha": "151747b7a21aaaa70a90372b80052e62b87e04e2", "size": 1047, "ext": "r", "lang": "R", "max_stars_repo_path": "poLCA/poLCA_election_demo.r", "max_stars_repo_name": "matthew9602/R-demo", "max_stars_repo_head_hexsha": "301e343750b6e6874985d6db9770379443ff3e86", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "poLCA/poLCA_election_demo.r", "max_issues_repo_name": "matthew9602/R-demo", "max_issues_repo_head_hexsha": "301e343750b6e6874985d6db9770379443ff3e86", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "poLCA/poLCA_election_demo.r", "max_forks_repo_name": "matthew9602/R-demo", "max_forks_repo_head_hexsha": "301e343750b6e6874985d6db9770379443ff3e86", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 49.8571428571, "max_line_length": 72, "alphanum_fraction": 0.6991404011, "num_tokens": 404, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8128673087708699, "lm_q2_score": 0.647798211152541, "lm_q1q2_score": 0.5265739885261497}} {"text": "#R version 3.3.2 \n\ni <- 1\n\n# O(maxN) \nwhile (i <= 1000000000) {\n if (i %% 2 == 0 && i %% 3 == 0 && i %% 4 == 0 && i %% 5 == 0 && i %% 6 == 0 && i %% 7 == 0 && i %% 8 == 0 &&\n i %% 9 == 0 && i %% 10 == 0 && i %% 11 == 0 && i %% 12 == 0 && \n i %% 13 == 0 && i %% 14 == 0 && i %% 15 == 0 && i %% 16 == 0 &&\n i %% 17 == 0 && i %% 18 == 0 && i %% 19 == 0 && i %% 20 == 0) {\n print(i)\n break\n }\n i <- i + 1\n}\n\n\n\ni <- 2*3*5*7*11*13*17*19\n\n\n\n# O(n / (f(k))\nwhile (i <= 1000000000) {\n if (i %% 4 == 0 && i %% 6 == 0 && i %% 8 == 0 &&\n i %% 9 == 0 && i %% 10 == 0 && i %% 12 == 0 && \n i %% 14 == 0 && i %% 15 == 0 && i %% 16 == 0 &&\n i %% 18 == 0 && i %% 20 == 0) {\n print(i)\n break\n }\n i <- 2*3*5*7*11*13*17*19\n}\n", "meta": {"hexsha": "456d555a7e3d7d601267f851b89da41db156c58d", "size": 912, "ext": "r", "lang": "R", "max_stars_repo_path": "R/Problem5.r", "max_stars_repo_name": "ditekunov/projectEuler-research", "max_stars_repo_head_hexsha": "e8f84388045cdc1391d1e363c7b55ff4f85be708", "max_stars_repo_licenses": ["Unlicense"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2018-05-20T08:01:42.000Z", "max_stars_repo_stars_event_max_datetime": "2018-05-20T08:05:07.000Z", "max_issues_repo_path": "R/Problem5.r", "max_issues_repo_name": "ditekunov/ProjectEuler-asymptotics", "max_issues_repo_head_hexsha": "e8f84388045cdc1391d1e363c7b55ff4f85be708", "max_issues_repo_licenses": ["Unlicense"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/Problem5.r", "max_forks_repo_name": "ditekunov/ProjectEuler-asymptotics", "max_forks_repo_head_hexsha": "e8f84388045cdc1391d1e363c7b55ff4f85be708", "max_forks_repo_licenses": ["Unlicense"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 26.8235294118, "max_line_length": 112, "alphanum_fraction": 0.2379385965, "num_tokens": 374, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8354835371034369, "lm_q2_score": 0.6297746074044134, "lm_q1q2_score": 0.5261663165721676}} {"text": "#!/usr/local/bin/Rexec\n########### set arguments list ###############################\nargs<-commandArgs();\nargs_start=grep(\"--args\",args)+1;\nif (length(args_start)>0 && args_start < length(args) ) {\n args<-args[args_start:length(args)]\n} else {\n args=c()\n}\n\nonset<-function(\n width=4,\n height=6,\n eps=FALSE,\n dt=.001,\n time=10/dt,\n tau1=1,tau2=tau1*2/3,\n k=1,\n start1=ceiling(time/8),\n start2=ceiling(time*2/8)) {\n\n v = matrix(rep(0,time*2),time,2);\n u = matrix(rep(0,time*2),time,2);\n h = matrix(rep(0,time*2),time,2);\n h[start1,1]=1000;\n h[start2,2]=1000;\n xgap=time/5;\n xlim=c(-xgap,time+xgap)\n ylim=c(-.5,1.5)\n axlim=c(-xgap/2,time)\n aylim=c(-.3,1.3)\n\n lwd <- 1.5\n\n # simulation\n for(t in 2:time) {\n\n # pre-synaptic onset\n v[t,1] = v[t-1,1] + (dt/tau1)*( -v[t-1,1] +k*h[t,1] );\n u[t,1] = u[t-1,1] + (dt/tau1)*( -u[t-1,1] +v[t-1,1] );\n\n # post-synaptic onset\n v[t,2] = v[t-1,2] + (dt/tau2)*( -v[t-1,2] +k*h[t,2] );\n u[t,2] = u[t-1,2] + (dt/tau2)*( -u[t-1,2] +v[t-1,2] );\n\n }\n if(eps==TRUE) {\n postscript(\n 'onset.eps',\n onefile=FALSE,\n horizontal=FALSE,\n width=width,height=height,\n family=\"times\");\n }\n\n layout(c(4,1,1,1,2,2,2,3,3,3,4),11,1)\n par(mar=c(.01,.01,.01,.01))\n\n plot(\n h[,1]/1000,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=lwd,\n xlim=xlim,\n ylim=ylim,\n family=\"serif\")\n\n polygon(c(1:time,time), c(h[,1], h[,1][1]), col='#DDDDDD',border=NA)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=lwd,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(h(t))),cex=2)\n text(start1,-.3,expression(italic(t[i])),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n\n\n plot(\n 1e200,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=1,\n xlim=xlim,\n ylim=ylim)\n\n polygon(c(1:time,time), c(v[,1], v[,1][1]), col='#DDDDDD',border=NA)\n lines(v[,1], lwd=lwd)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=lwd,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(v(t))),cex=2)\n text(start1,-.3,expression(italic(t[i])),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n\n\n plot(\n 1e200,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=1,\n xlim=xlim,\n ylim=ylim)\n\n polygon(c(1:time,time), c(u[,1], u[,1][1]), col='#DDDDDD',border=NA)\n lines(u[,1], lwd=lwd)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=lwd,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(u(t))),cex=2)\n text(start1,-.3,expression(italic(t[i])),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n text(start1+ceiling(tau1/dt),-.3,expression(italic(tau+t[i])),cex=2,adj=c(.2,.5))\n arrows(start1+ceiling(tau1/dt),0,start1+ceiling(tau1/dt),\n u[start1+ceiling(tau1/dt),1],lwd=lwd,length=0,lty=3);\n\n if(eps==TRUE) {\n graphics.off();\n }\n\n}\n\nderivatives<-function(\n width=4,\n height=6,\n eps=FALSE,\n name=\"onset_tau.eps\",\n dt=.001,\n time=10/dt,\n tau1=1,tau2=tau1,\n k=1,\n start1=ceiling(dt)+1,\n start2=ceiling(time*2/8)) {\n\n v = matrix(rep(0,time*2),time,2);\n u = matrix(rep(0,time*2),time,2);\n h = matrix(rep(0,time*2),time,2);\n h[start1:length(h[,1]),1]=1;\n h[start2:length(h[,2]),2]=1;\n xgap=time/5;\n xlim=c(-xgap,time+xgap)\n ylim=c(-.5,1.3)\n axlim=c(-xgap/2,time)\n aylim=c(-.3,1.1)\n\n\n # simulation\n for(t in 2:time) {\n\n # pre-synaptic onset\n v[t,1] = v[t-1,1] + (dt/tau1)*( -v[t-1,1] +k*h[t,1] );\n u[t,1] = u[t-1,1] + (dt/tau1)*( -u[t-1,1] -v[t-1,1] + k*h[t,1] );\n\n # post-synaptic onset\n v[t,2] = v[t-1,2] + (dt/tau2)*( -v[t-1,2] +k*h[t,2] );\n u[t,2] = u[t-1,2] + (dt/tau2)*( -u[t-1,2] -v[t-1,2] + k*h[t,2] );\n\n }\n u1=u[,1];\n u2=u[,2];\n timetime=1:time;\n du1=c(0,(u1[2:time]-u1[1:(time-1)])/dt);\n du2=c(0,(u2[2:time]-u2[1:(time-1)])/dt);\n du1p=du1*(du1>0);\n du1n=abs(du1*(du1<0));\n du2p=du2*(du2>0);\n du2n=abs(du2*(du2<0));\n\n if(eps==TRUE) {\n postscript(\n name,\n onefile=FALSE,\n horizontal=FALSE,\n width=width,height=height,\n family=\"Times\");\n }\n\n layout(seq(1,4,1),4,1)\n par(mar=c(.01,.01,.01,.01))\n\n\n\n plot(\n 1e100,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=1,\n xlim=xlim,\n ylim=ylim,\n family=\"serif\")\n\n polygon(c(1:time,time), c(u1, u1[1]), col='#DDDDDD',border=NA)\n lines(u1,lwd=2);\n polygon(c(1:time,time), c(u2, u2[1]), col='#EEEEEE',border=NA)\n lines(u2,lwd=1);\n #lines(u1,lwd=0.5);\n arrows(4,-.1,start2-start1-4,-.1,code=3,angle=90,length=.02,lwd=.6)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(u(t))),cex=2)\n text(time*(1/16),-.3,expression(paste(Delta,\"t\")),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n text(timetime[u1==max(u1)][1],max(u1)+.2,expression(italic(u[1])),cex=2)\n text(timetime[u2==max(u2)][1],max(u2)+.2,expression(italic(u[2])),cex=2)\n\n plot(\n 1e200,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=3,\n xlim=xlim,\n ylim=ylim,\n family=\"serif\")\n\n\n polygon(c(1:time,time), c(du1, du1[1]), col='#DDDDDD',border=NA)\n lines(du1,lwd=2);\n polygon(c(1:time,time), c(du2, du2[1]), col='#EEEEEE',border=NA)\n lines(du2,lwd=1);\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(dot(u)(t))),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n text(timetime[du1==max(du1)]+1000,max(du1),expression(italic(dot(u)[1])),cex=2)\n text(timetime[du2==max(du2)]+1000,max(du2),expression(italic(dot(u)[2])),cex=2)\n taux = ceiling(tau1/dt);\n tau2x = (start2-start1)+ceiling(tau2/dt);\n\n plot(\n 1e200,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=3,\n xlim=xlim,\n ylim=ylim,\n family=\"serif\")\n\n\n polygon(c(1:time,time), c(du1p, du1p[1]), col='#DDDDDD',border=NA)\n lines(du1p,lwd=2);\n polygon(c(1:time,time), c(du2p, du2p[1]), col='#EEEEEE',border=NA)\n lines(du2p,lwd=1);\n points(c(taux,tau2x),c(0,0),pch=19)\n\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(dot(u)(t))),cex=2)\n text(taux,-.3,expression(italic(tau[1])),cex=2)\n text(tau2x,-.3,expression(italic(tau[2] + paste(Delta,'t'))),adj=c(0.1,.4),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n text(timetime[du1p==max(du1p)][1]+1400,max(du1p),expression(group(\"[\",italic(dot(u)[1]),\"]\")^{'+'}),cex=2)\n text(timetime[du2p==max(du2p)][1]+1400,max(du2p),expression(group(\"[\",italic(dot(u)[2]),\"]\")^{'+'}),cex=2)\n\n plot(\n 1e200,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=3,\n xlim=xlim,\n ylim=ylim,\n family=\"serif\")\n\n\n polygon(c(1:time,time), c(du1n, du1n[1]), col='#DDDDDD',border=NA)\n lines(du1n,lwd=2);\n polygon(c(1:time,time), c(du2n, du2n[1]), col='#EEEEEE',border=NA)\n lines(du2n,lwd=1);\n points(c(taux,tau2x),c(0,0),pch=19)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(dot(u)(t))),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n text(taux,-.3,expression(italic(tau[1])),cex=2)\n text(tau2x,-.3,expression(italic(tau[2] + paste(Delta,'t'))),adj=c(0.1,.4),cex=2)\n text(timetime[du1n==max(du1n)][1][1]-40,max(du1n)+.3,expression(group(\"[\",italic(dot(u)[1]),\"]\")^{'-'}),cex=2)\n text(timetime[du2n==max(du2n)][1][1],max(du2n)+.3,expression(group(\"[\",italic(dot(u)[2]),\"]\")^{'-'}),cex=2)\n\n if(eps==TRUE) {\n graphics.off();\n }\n\n\n}\n\nhsides<-function(\n width=4,\n height=6,\n ylim=c(-.3,1.5),\n eps=FALSE,\n name=\"hd\",\n dt=.01,\n time=10/dt,\n start1=ceiling(time/8),\n start2=ceiling(time*2/8),\n co1=1,\n co2=1)\n{\n\n\n # init variable s of simulation\n\n\n h = matrix(rep(0,time*2),time,2);\n h[start1:length(h[,1]),1]=1;\n h[start2:length(h[,2]),2]=1;\n\n\n # define constants for graph limits\n xgap=time/4;\n xlim=c(-xgap*2/3,time+xgap);\n aylim=ylim\n ylimrange= ylim[2]-ylim[1]\n ylim=ylim+c(-ylimrange/20, +ylimrange/5);\n axlim=c(-xgap/2,time);\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n\n\n # define constants for text positions\n maxlabx=1/4; maxlaby=8/18;\n maxlabx0=-1/12; maxlabx1=-1/12; maxlabx2=-1/12; maxlabx3=-1/12; maxlabx4=-1/12;\n maxlaby0=2/18; maxlaby1=7/18; maxlaby2=10/18; maxlaby3=13/18; maxlaby4=16/18;\n lpx=c(0,1,1,0); lpy=c(1,1,0,0); bl=.1*time;\n pgap = ylimrange/2; right=.12*time;\n\n\n timetime=1:time;\n\n hd=(h[2:dim(h)[1],]-h[1:(dim(h)[1]-1),]);\n\n if(eps==TRUE) {\n postscript(\n paste('onset_',name,'.eps',sep=''),\n onefile=FALSE,\n horizontal=FALSE,\n pointsize=6,\n width=width,height=height,family=\"Times\");\n }\n\n\n par(mar=c(.01,.01,.01,.01))\n plot(\n hd[,2],\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=1,\n col='#ffffff',\n xlim=xlim,\n ylim=ylim,\n family=\"serif\");\n\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n\n lines(hd[,1],lwd=2)\n lines(hd[,2],lwd=2)\n hd1=hd[,1]\n hd2=hd[,2]\n text(-xgap*3/6,aylim[2],expression(italic(s(t))),cex=2)\n\n text(timetime[hd1==max(hd1)][1], max(hd1)+.5*pgap,expression(italic(s[1])),cex=2)\n text(timetime[hd2==max(hd2)][1]+40*pgap, max(hd2)+.5*pgap,expression(italic(s[2])),cex=2)\n text(time+time*0.04,-.2*pgap,expression(italic(t)),cex=2)\n\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n\n if(eps==TRUE) {\n graphics.off();\n }\n}\n\ncomponentEvents<-function(\n width=4, # width of eps figure (inches)\n height=6, # height of eps figure (inches)\n ylim=c(-.3,1.5), # range of amplitude shown in the graph\n eps=FALSE, # print eps figure intead of screen\n name=\"wpp1\", # print eps figure intead of screen\n dt=.01, # integration step\n time=10/dt, # timesteps of the simulated trial\n tau1=1, # decay of the pre-synaptic neuron\n tau2=tau1, # decay of the post-synaptic neuron\n k=1, # amplification of input\n start1=ceiling(time/8), # start of the signal to the pre-synaptic neuron\n start2=ceiling(time*2/8), # start of the signal to the post-synaptic neuron\n co1=1, # type of the pre-synaptic event (can be\n # 0:onset,\n # 1:positive part of derivative of onset,\n # 2: absolute of negative derivative of onset)\n co2=1, # type of the post-synaptic event (can be\n # 0:onset,\n # 1:positive part of derivative of onset,\n # 2: absolute of negative derivative of onset)\n plot_prod=FALSE\n ) {\n\n\n # init variable s of simulation\n\n v = matrix(rep(0,time*2),time,2);\n u = matrix(rep(0,time*2),time,2);\n h = matrix(rep(0,time*2),time,2);\n h[start1:length(h[,1]),1]=1;\n h[start2:length(h[,2]),2]=1;\n\n\n # define constants for graph limits\n xgap=time/4;\n xlim=c(-xgap*2/3,time+xgap);\n aylim=ylim\n ylimrange= ylim[2]-ylim[1]\n ylim=ylim+c(-ylimrange/20, +ylimrange/20);\n axlim=c(-xgap/2,time);\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n\n\n # define constants for text positions\n maxlabx=1/4; maxlaby=8/18;\n maxlabx0=-1/12; maxlabx1=-1/12; maxlabx2=-1/12; maxlabx3=-1/12; maxlabx4=-1/12;\n maxlaby0=2/18; maxlaby1=7/18; maxlaby2=10/18; maxlaby3=13/18; maxlaby4=16/18;\n lpx=c(0,1,1,0); lpy=c(1,1,0,0); bl=.1*time;\n pgap = ylimrange/2; right=.12*time;\n\n\n # simulation\n for(t in 2:time) {\n\n # pre-synaptic onset\n v[t,1] = v[t-1,1] + (dt/tau1)*( -v[t-1,1] +k*h[t,1] );\n u[t,1] = u[t-1,1] + (dt/tau1)*( -u[t-1,1] -v[t-1,1] + k*h[t,1] );\n\n # post-synaptic onset\n v[t,2] = v[t-1,2] + (dt/tau2)*( -v[t-1,2] +k*h[t,2] );\n u[t,2] = u[t-1,2] + (dt/tau2)*( -u[t-1,2] -v[t-1,2] + k*h[t,2] );\n\n }\n\n # standardization of onsets\n u1=u[,1]; u1=u1/max(u1)\n u2=u[,2]; u2=u2/max(u2)\n\n # derivatives of onsets\n timetime=1:time;\n du1=c(0,(u1[2:time]-u1[1:(time-1)])/dt);du1=du1/max(du1)\n du2=c(0,(u2[2:time]-u2[1:(time-1)])/dt);du2=du2/max(du2)\n\n\n # positive and negative parts of derivatives\n du1p=du1*(du1>0);\n du1n=abs(du1*(du1<0));\n du2p=du2*(du2>0);\n du2n=abs(du2*(du2<0));\n dd=t(rbind(du1p,du1n,du2p,du2n));\n\n\n # decide if d1 and d2 events will be onsets\n # or positive/negative derivative\n if(co1==0) {\n d1=u1;\n du1=u1;\n d2=dd[,2+co2];\n } else if(co2==0) {\n d2=u2;\n du2=u2;\n d1=dd[,co1];\n } else {\n d2=dd[,2+co2];\n d1=dd[,co1];\n }\n\n # standardized combination on the two events\n ddd=d1*d2;\n if(max(ddd)!=0) {\n ddd=ddd/max(ddd)\n ddd=ddd*((max(d1)+max(d2))/2)\n }\n\n # decide if printing the graph on an eps file\n if(eps==TRUE) {\n postscript(\n paste('onset_',name,'.eps',sep=''),\n onefile=FALSE,\n horizontal=FALSE,\n width=width, height=height,\n pointsize=6, family=\"Times\");\n }\n\n #plot first event\n par(mar=c(.01,.01,.01,.01))\n plot(\n d1,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=1,\n col='#ffffff',\n xlim=xlim,\n ylim=ylim,\n family=\"serif\");\n\n\n #plot composition\n if(plot_prod) {\n polygon(c(1,1:time,time), c(0,ddd, 0), lwd=.01,col='#888888',border=NA)\n }\n #plot first event area\n polygon(c(1,1:time,time), c(0, d1,0), col='#444444',density=20,angle=-45,border=NA)\n #plot second event area\n polygon(c(1,1:time,time), c(0, d2,0), col='#444444',density=20,angle=45,border=NA)\n\n #plot first event\n lines(d1,lwd=2)\n #plot second event\n lines(d2,lwd=2)\n #plot axes\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n #plot axes'labels\n text(-xgap*3/6,aylim[2],expression(italic(u(t))),cex=2)\n text(time+time*0.04,-.1*pgap,expression(italic(t)),cex=2)\n\n # compose text to explain envent types in the legend\n if(co2==0) {\n if(co1==1) { first=\"group('[',dot(u)[1],']')^{'+'}\" }\n else if (co1==2) { first=\"group('[',dot(u)[1],']')^{'-'}\" }\n second=\"u[2]\"\n }else if(co1==0) {\n if(co2==1) { second=\"group('[',dot(u)[2],']')^{'+'}\" }\n else if (co2==2) { second=\"group('[',dot(u)[2],']')^{'-'}\" }\n first=\"u[1]\"\n } else{\n if(co1==1) { first=\"group('[',dot(u)[1],']')^{'+'}\" }\n else if (co1==2) { first=\"group('[',dot(u)[1],']')^{'-'}\" }\n if(co2==1) { second=\"group('[',dot(u)[2],']')^{'+'}\" }\n else if (co2==2) { second=\"group('[',dot(u)[2],']')^{'-'}\" }\n }\n txt_prod = eval(parse(text=paste('expression(',first,'*',second,')')));\n txt_1 = eval(parse(text=paste('expression(',first,')')));\n txt_2 = eval(parse(text=paste('expression(',second,')')));\n\n #legend\n if(plot_prod) {\n text(time-time*maxlabx0,\n aylim[2]-aylim[2]*maxlaby0,\n txt_prod,cex=7,adj=c(.75,.5));\n if(max(ddd)!=0 )\n {\n polygon(time-time*maxlabx0+right+ceiling(lpx*bl),\n (lpy*.1*pgap)+(aylim[2]-aylim[2]*maxlaby0-.05*pgap),\n col='#888888', border=NA);\n }\n }\n else{\n text(time-time*(1/2),\n aylim[2],\n \"component:\",cex=2,adj=c(.75,.5));\n\n text(time-time*maxlabx1,\n aylim[2],\n txt_prod,cex=2,adj=c(.75,.5));\n }\n if(co1==1) {first='+'} else if (co1==2) {first='-'}\n text(time - time * maxlabx1, aylim[2] - aylim[2] * maxlaby1, txt_1, cex =\n 2)\n\n polygon(\n time - time * maxlabx1 + right + ceiling(lpx * bl),\n (lpy * .2 * pgap) + (aylim[2] - aylim[2] * maxlaby1 - .05 *\n pgap)\n ,\n col = '#444444',\n density = 20,\n angle = -45,\n border = NA\n )\n\n polygon(\n time - time * maxlabx1 + right + ceiling(lpx * bl),\n (lpy * .2 * pgap) + (aylim[2] - aylim[2] * maxlaby1 - .05 *\n pgap)\n ,\n lwd = 2\n )\n\n\n if (co2 == 1) {\n first = '+'\n } else if (co2 == 2) {\n first = '-'\n }\n\n text(time - time * maxlabx2, aylim[2] - aylim[2] * maxlaby2, txt_2, cex =\n 2)\n\n polygon(\n time - time * maxlabx2 + right + ceiling(lpx * bl),\n (lpy * .2 * pgap) + (aylim[2] - aylim[2] * maxlaby2 - .05 *\n pgap)\n ,\n col = '#444444',\n density = 20,\n angle = 45,\n border = NA\n )\n\n polygon(\n time - time * maxlabx2 + right + ceiling(lpx * bl),\n (lpy * .2 * pgap) + (aylim[2] - aylim[2] * maxlaby2 - .05 *\n pgap)\n ,\n lwd = 2\n )\n\n\n # print the eps file\n if(eps==TRUE) {\n dev.off()\n }\n\n}\n\n\nallcomponents <-function() {\n\n\n labels1= c('wpp1','wpn1','wnp1','wnn1','wpp2','wpn2','wnp2','wnn2')\n labels2= c('wsp1','wsn1','wps1','wns1','wsp2','wsn2','wps2','wns2')\n\n tau1=1;\n tau2=1;\n start1=100;\n hh=1;\n l1 = 0; l2 = 0\n lab = 0\n for(start2 in c(50,150)) {\n hsides(eps=TRUE,dt=.01,start1=start1,start2=start2,\n name=paste(\"hd\",hh,sep=''),width=2.5,height=0.5);\n hh=hh+1;\n for(co1 in c(0,1,2)) {\n for(co2 in c(0,1,2)) {\n if ((co1+co2)> 0) {\n labels=labels1\n if ( (co1 == 0 || co2 == 0) ) {\n labels=labels2\n l2 = l2 + 1\n lab = l2\n } else if ( co1 > 0 && co2 > 0 ) {\n labels=labels1\n l1 = l1 + 1\n lab = l1\n }\n\n componentEvents(\n co1=co1,\n co2=co2,\n start1=start1,\n start2=start2,\n name=labels[lab],\n tau1=tau1,\n tau2=tau2,\n eps=TRUE,\n width=2.5,\n height=1.375);\n }\n }\n }\n }\n}\n\n\nonset_alpha<-function(\n width=4,\n height=6,\n eps=FALSE,\n dt=.001,\n time=10/dt,\n tau1=1,\n k=1,\n start1=ceiling(time/8)) {\n\n graphics.off();\n\n v = rep(0,time);\n u = rep(0,time);\n v = rep(0,time);\n h = rep(0,time);\n h[start1]=1;\n\n xgap=time/5;\n xlim=c(-xgap,time+xgap)\n ylim=c(-.5,1.5)\n axlim=c(-xgap/2,time)\n aylim=c(-.3,1.3)\n\n # simulation\n for(t in 2:time) {\n\n # pre-synaptic onset\n v[t] = v[t-1] + (dt/tau1)*( -v[t-1] +k*h[t] );\n u[t] = u[t-1] + (dt/tau1)*( -u[t-1] +v[t-1] );\n\n }\n if(eps==TRUE) {\n postscript(\n 'onset.eps',\n onefile=FALSE,\n horizontal=FALSE,\n width=width,height=height,\n family=\"times\");\n }\n\n layout(seq(1,3,1),3,1)\n par(mar=c(.01,.01,.01,.01))\n\n\n plot(\n h,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=3,\n xlim=xlim,\n ylim=ylim,\n family=\"serif\")\n\n polygon(c(1:time,time), c(h, h[1]), col='#DDDDDD',border=NA)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(h(t))),cex=2)\n text(start1,-.3,expression(italic(t[i])),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n text(time,-.3,expression(italic(t)),cex=2)\n\n ylim=c(-.0005,0.0015)\n aylim=c(-.0003,0.0013)\n plot(\n v,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=3,\n xlim=xlim,\n ylim=ylim)\n\n polygon(c(1:time,time), c(v, v[1]), col='#DDDDDD',border=NA)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(v(t))),cex=2)\n text(start1,aylim[1],expression(italic(t[i])),cex=2)\n text(time,aylim[1],expression(italic(t)),cex=2)\n\n\n plot(\n u,\n type='l',\n frame.plot=FALSE,\n axes=FALSE,\n xlab='',\n ylab='',\n lwd=3,\n xlim=xlim,\n ylim=ylim)\n\n polygon(c(1:time,time), c(u, u[1]), col='#DDDDDD',border=NA)\n d<-c()\n d<-rbind(d,c(rbind(axlim,aylim*0)))\n d<-rbind(d,c(rbind(axlim*0,aylim)))\n arrows(d[,1],d[,2],d[,3],d[,4],lwd=1,length=.08);\n text(-xgap*2/3,aylim[2],expression(italic(u(t))),cex=2)\n text(start1,aylim[1],expression(italic(t[i])),cex=2)\n text(time,aylim[1],expression(italic(t)),cex=2)\n text(start1+ceiling(tau1/dt),aylim[1],expression(italic(tau+t[i])),cex=2,adj=c(.2,.5))\n arrows(start1+ceiling(tau1/dt),0,start1+ceiling(tau1/dt),u[start1+ceiling(tau1/dt),1],lwd=1,length=0,lty=3);\n\n if(eps==TRUE) {\n graphics.off();\n }\n\n v\n}\n\n\n\ncase_study <- function() {\n ##### case study#\n x11()\n componentEvents(\n co1 = 1,\n co2 = 0,\n start1 = 20, start2 = 16,\n dt = 0.001,\n time = 80,\n tau1 = 0.029999733,\n tau2 = 0.006626032\n )\n axis(\n 1,\n outer = FALSE,\n lwd = 0,\n lwd.ticks = .2,\n pos = -.3,\n cex.axis = 1.5,\n at = c(0, 40, 80, 120, 160, 200),\n labels = format(c(0, 40, 80, 120, 160, 200))\n )\n\n x11()\n componentEvents(\n co1 = 1,\n co2 = 1,\n start1 = 20,\n start2 = 20,\n dt = 0.001,\n time = 80,\n tau1 = 0.029999733,\n tau2 = 0.006626032\n )\n axis(\n 1,\n outer = FALSE,\n lwd = 0, lwd.ticks = .2,\n pos = -.3,\n cex.axis = 1.5,\n at = c(0, 40, 80, 120, 160, 200),\n labels = format(c(0, 40, 80, 120, 160, 200))\n )\n}\n\n", "meta": {"hexsha": "e446a05c048f3534d055b34549d547bce6cd2639", "size": 24615, "ext": "r", "lang": "R", "max_stars_repo_path": "code/R/derivatives.r", "max_stars_repo_name": "GOAL-Robots/CNR_140618_GDHL", "max_stars_repo_head_hexsha": "c136714b1126a2d257ed8e49af7f51aabfc125b1", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "code/R/derivatives.r", "max_issues_repo_name": "GOAL-Robots/CNR_140618_GDHL", "max_issues_repo_head_hexsha": "c136714b1126a2d257ed8e49af7f51aabfc125b1", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "code/R/derivatives.r", "max_forks_repo_name": "GOAL-Robots/CNR_140618_GDHL", "max_forks_repo_head_hexsha": "c136714b1126a2d257ed8e49af7f51aabfc125b1", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 28.1314285714, "max_line_length": 114, "alphanum_fraction": 0.4728011375, "num_tokens": 8734, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7745833841649233, "lm_q2_score": 0.6791786861878392, "lm_q1q2_score": 0.526080525200063}} {"text": "#' Bass Model\n#'\n#' Fits the Bass Diffusion model. In particular, fits an observed curve of\n#' proportions of adopters to \\eqn{F(t)}, the proportion of adopters at time\n#' \\eqn{t}, finding the corresponding coeficients \\eqn{p}, Innovation rate,\n#' and \\eqn{q}, imitation rate.\n#'\n#' @param Time Integer vector with values greater than 0. The \\eqn{t} parameter.\n#' @param p Numeric scalar. Coefficient of innovation.\n#' @param q Numeric scalar. Coefficient of imitation.\n#' @param dat Either a diffnet object, or a numeric vector.\n#' Observed cumulative proportion of adopters.\n#' @param x An object of class \\code{diffnet_bass}.\n#' @param ... Further arguments passed to the method.\n#'\n#' @details The function fits the bass model with parameters \\eqn{[p, q]} for\n#' values \\eqn{t = 1, 2, \\dots, T}, in particular, it fits the following function:\n#'\n#' \\deqn{\n#' F(t) = \\frac{1 - \\exp{-(p+q)t}}{1 + \\frac{q}{p}\\exp{-(p+q)t}}\n#' }{\n#' F(t) = [1 - exp(-(p + q)*t)]/[1 + exp(-(p + q)*t)*(q/p)]\n#' }\n#'\n#' Which is implemented in the \\code{bass_F} function. The proportion of adopters\n#' at time \\eqn{t}, \\eqn{f(t)} is:\n#'\n#' \\deqn{\n#' f(t) = \\left\\{\\begin{array}{ll}\n#' F(t), & t = 1 \\\\\n#' F(t) - F(t-1), & t > 1\n#' \\end{array}\\right.\n#' }{\n#' f(t) = ifelse(t == 1, F(t), F(t) - F(t-1))\n#' }\n#'\n#' and it's implemented in the \\code{bass_f} function.\n#'\n#' For testing purposes only, the gradient of \\eqn{F} with respect to \\eqn{p}\n#' and \\eqn{q} is implemented in \\code{bass_dF}.\n#'\n#' The estimation is done using \\code{\\link[stats:nls]{nls}}.\n#'\n#'\n#' @return An object of class \\code{nls} and \\code{diffnet_bass}. For more\n#' details, see \\code{\\link[stats:nls]{nls}} in the \\pkg{stats} package.\n#'\n#' @examples\n#' # Fitting the model for the Brazilian Farmers Data --------------------------\n#' data(brfarmersDiffNet)\n#' ans <- fitbass(brfarmersDiffNet)\n#'\n#' # All the methods that work for the -nls- object work here\n#' ans\n#' summary(ans)\n#' coef(ans)\n#' vcov(ans)\n#'\n#' # And the plot method returns both, fitted and observed curve\n#' plot(ans)\n#'\n#' @references\n#' Bass's Basement Institute Institute. The Bass Model. (2010).\n#' Available at: \\url{http://www.bassbasement.org/BassModel/Default.aspx}. (Accessed: 29th March 2017)\n#' @name bass\n#' @author George G. Vega Yon\n#' @family statistics\nNULL\n\n#' @rdname bass\n#' @export\nfitbass <- function(dat, ...) UseMethod(\"fitbass\")\n\n#' @export\n#' @rdname bass\nfitbass.diffnet <- function(dat, ...) {\n .fitbass(cumulative_adopt_count(dat$cumadopt)[\"prop\",], ...)\n}\n\n#' @export\n#' @rdname bass\nfitbass.default <- function(dat, ...) {\n .fitbass(as.vector(dat), ...)\n}\n\n.fitbass <- function(dat, ...) {\n\n # Constants\n Time <- seq_along(dat)\n\n # # Optimization (Fit nonlinear regression)\n ans <- stats::nls(dat ~ bass_F(Time, p, q), start=list(p = dat[1], q = .5 ),...)\n\n structure(c(\n ans,\n list(\n q = stats::coef(ans)[2],\n p = stats::coef(ans)[1],\n Fvals = bass_F(Time, stats::coef(ans)[1], stats::coef(ans)[2]),\n fvals = bass_f(Time, stats::coef(ans)[1], stats::coef(ans)[2]),\n nper = length(Time),\n dat = dat,\n Time = Time\n )\n ), class = c(\"diffnet_bass\", class(ans)))\n\n}\n\n#' @rdname bass\n#' @param y Integer vector. Time (label).\n#' @param pch Passed to \\code{\\link[graphics:plot]{matplot}}.\n#' @param main Passed to \\code{\\link[graphics:plot]{matplot}}.\n#' @param xlab Character scalar. Label of the \\code{x} axis.\n#' @param ylab Character scalar. Label of the \\code{y} axis.\n#' @param type Passed to \\code{\\link[graphics:plot]{matplot}}.\n#' @param lty Passed to \\code{\\link[graphics:plot]{matplot}}.\n#' @param col Passed to \\code{\\link[graphics:plot]{matplot}}.\n#' @param bg Passed to \\code{\\link[graphics:plot]{matplot}}.\n#' @param include.legend Logical scalar. When \\code{TRUE}, draws a legend.\n#' @param add Passed to \\code{\\link[graphics:plot]{matplot}}.\n#' @export\nplot.diffnet_bass <- function(\n x, y=1:length(x$m$lhs()), add=FALSE,\n pch = c(21,24),\n main = \"Bass Diffusion Model\",\n ylab = \"Proportion of adopters\",\n xlab = \"Time\",\n type = c(\"b\", \"b\"),\n lty = c(2,1),\n col = c(\"black\",\"black\"),\n bg = c(\"lightblue\",\"gray\"),\n include.legend = TRUE,\n ...) {\n\n if (length(type) == 1) type <- rep(type,2)\n if (length(pch) == 1) pch <- rep(pch,2)\n if (length(lty) == 1) lty <- rep(lty,2)\n if (length(col) == 1) lty <- rep(col,2)\n if (length(bg) == 1) bg <- rep(bg,2)\n\n mat <- with(x, cbind(m$lhs(), m$fitted()))\n\n matplot(mat, xlab = xlab, ylab = ylab, main=main,\n ylim = c(0,1), type=type, lty=lty, bg=bg, col=col,\n pch=pch, add=add, ...)\n\n # Adding legend\n if (!add && include.legend)\n legend(\n \"topleft\", bty=\"n\",\n legend = c(\"Observed Cumulative Adopters\", \"Predicted Cumulative Adopters\"),\n col=col, pt.bg=bg, pch=pch\n )\n\n\n invisible(x)\n\n}\n\n#' @rdname bass\n#' @export\nbass_F <- function(Time, p, q) {\n (1 - exp(-(p + q)*Time))/(1 + (q/p)*exp(-(p+q)*Time))\n}\n\n#' @rdname bass\n#' @export\nbass_dF <- function(p, q, Time) {\n expv <- exp((p+q)*Time)\n\n rbind(\n expv * p^2*Time + q*(expv*(1 + p*Time) - 1)/\n (expv*p + q)^2,\n p * (1 + expv*(p*Time + q*Time - 1)) /\n (expv*p + q)^2\n )\n}\n\n#' @rdname bass\n#' @export\nbass_f <- function(Time, p, q) {\n ifelse(Time==1, bass_F(Time, p, q), bass_F(Time, p, q) - bass_F(Time-1, p, q))\n}\n\n", "meta": {"hexsha": "7315c9535c936f9a23444c83663e3fcab4919ed5", "size": 5367, "ext": "r", "lang": "R", "max_stars_repo_path": "R/bass.r", "max_stars_repo_name": "USCCANA/netdiffuseR", "max_stars_repo_head_hexsha": "4cda66a4381c2df3ee2d738f6c97ac0b2916a5c4", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 69, "max_stars_repo_stars_event_min_datetime": "2015-12-15T02:49:46.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-08T02:48:37.000Z", "max_issues_repo_path": "R/bass.r", "max_issues_repo_name": "USCCANA/netdiffuseR", "max_issues_repo_head_hexsha": "4cda66a4381c2df3ee2d738f6c97ac0b2916a5c4", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 30, "max_issues_repo_issues_event_min_datetime": "2015-12-17T03:43:07.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-21T18:50:22.000Z", "max_forks_repo_path": "R/bass.r", "max_forks_repo_name": "USCCANA/netdiffuseR", "max_forks_repo_head_hexsha": "4cda66a4381c2df3ee2d738f6c97ac0b2916a5c4", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 19, "max_forks_repo_forks_event_min_datetime": "2015-12-28T21:47:05.000Z", "max_forks_repo_forks_event_max_datetime": "2021-11-22T19:48:08.000Z", "avg_line_length": 29.0108108108, "max_line_length": 102, "alphanum_fraction": 0.601267002, "num_tokens": 1801, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7905303087996143, "lm_q2_score": 0.665410572017153, "lm_q1q2_score": 0.526027224975248}} {"text": "#' @title calcKristThrustFactor\n#'\n#' @description Calculate the thrust deduction factor (\\code{thrustFactor})\n#'(dimensionless) using the Kristensen method.\n#'\n#' @param shipType Ship type (vector of strings, see \\code{\\link{calcShipType}}),\n#' determined by Stat 5 code, \\itemize{\n#' \\item\"container.ship\"\n#' \\item\"bulk.carrier\"\n#' \\item\"tanker\"\n#' \\item\"general.cargo\"\n#' \\item\"vehicle.carrier\"\n#' \\item\"reefer\"\n#' \\item\"ro.ro\"\n#' \\item\"passenger\"\n#' \\item\"tug\"\n#' \\item\"misc\"\n#'}\n#'@param breadth Moulded breadth (vector of numericals, m)\n#'@param lwl Waterline Length (vector of numericals, m) (see\n#'\\code{\\link{calclwl}})\n#'@param Cbw Waterline block coefficient (vector of numericals, dimensionless)\n#' (see \\code{\\link{calcCbw}})\n#'@param propDiam Propeller diameter (vector of numericals, m) (see\n#' \\code{\\link{calcPropDia}})\n#'@param M Fineness/slenderness coefficient (vector of numericals,\n#' dimensionless) (see \\code{\\link{calcShipM}})\n#'@param nProp Number of propellers (vector of numericals, see\n#' \\code{\\link{calcPropNum}})\n#' @param tankerBulkCarrierShipTypes Ship types specified in input\n#'\\code{shipTypes} to be modeled as tankers and bulk carriers vessels\n#' (vector of strings)\n#'\n#' @details\n#'Thrust deduction factor is a component of hull efficiency as well as a\n#'component of propeller efficiency. It describes the increase in resistance on\n#'the hull from water getting sucked back towards the propeller.\n#'\n#' This method this requires ship types to be grouped. Use the\n#' \\code{tankerBulkCarrierShipTypes} grouping parameters to provide these ship\n#' type groupings. Any ship types not included in this grouping will be considered\n#' as miscellaneous vessels.\n#'\n#' @return \\code{thrustFactor} (vector of numericals, dimensionless)\n#'\n#' @references\n#'Kristensen, H. O. and Lutzen, M. 2013. \"Prediction of Resistance and Propulsion\n#'Power of Ships.\"\n#'\n#'\\href{https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}{Kristensen, H. O.\n#'\"Ship-Desmo-Tool.\" https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}\n#'\n#'@seealso \\itemize{\n#'\\item \\code{\\link{calclwl}}\n#'\\item \\code{\\link{calcCbw}}\n#'\\item \\code{\\link{calcPropDia}}\n#'\\item \\code{\\link{calcShipM}}\n#'\\item \\code{\\link{calcPropNum}}\n#'\\item \\code{\\link{calcShipType}}\n#'}\n#'\n#' @family Kristensen Calculations\n#'\n#' @examples\n#' calcKristThrustFactor(c(\"bulk.carrier\",\"container.ship\"),\n#' c(32.25,49),\n#' c(218.75,400),\n#' c(0.8,0.75),\n#' c(6.7,9.6),\n#' c(5.2,6.7),\n#' c(1,1))\n#'\n#' calcKristThrustFactor(\"bulk.carrier\", 32.25, 218.75, 0.8, 6.7,5.2,1)\n#'\n#' @export\n\ncalcKristThrustFactor <- function(shipType, breadth, lwl, Cbw, propDiam,M,nProp,\n tankerBulkCarrierShipTypes=c(\"tanker\",\"chemical.tanker\",\"liquified.gas.tanker\",\"oil.tanker\",\"other.tanker\",\"bulk.carrier\")\n){\n\n\n thrustFactor<-ifelse(nProp==1,\n ifelse(shipType%in%tankerBulkCarrierShipTypes,\n (#single prop case\n #t1\n #d\n (((0.625*breadth)/lwl)+0.08)+\n #e\n (0.165-((0.25*breadth)/lwl))/\n (#f\n (825-((8060*breadth)/lwl)+\n 20300*((breadth/lwl)^2))*\n (0.98-Cbw)^3+1)+\n #t2=0 since we assume F_a=0\n #t3\n (2*((propDiam/lwl)-0.04))\n )#thrust correction\n -0.26+0.04*M,\n (#single prop case for non bulk carriers non tankers\n #t1\n #d\n (((0.625*breadth)/lwl)+0.08)+\n #e\n (0.165-((0.25*breadth)/lwl))/\n (#f\n (825-((8060*breadth)/lwl)+\n 20300*((breadth/lwl)^2))*\n (0.98-Cbw)^3+1)+\n #t2=0 since we assume F_a=0\n #t3\n (2*((propDiam/lwl)-0.04))\n )),\n (#twin prop case\n 0.0665+0.62833*(1.133*(Cbw^2)-0.797*Cbw+0.215)\n )#end of twin prop case\n\n )#end elseif\n\n return(thrustFactor)\n\n\n}\n", "meta": {"hexsha": "73966a0b9c871fbf0379afa975747aa40718ec68", "size": 4588, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcKristThrustFactor.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcKristThrustFactor.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcKristThrustFactor.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 37.3008130081, "max_line_length": 156, "alphanum_fraction": 0.5318221447, "num_tokens": 1232, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8244619350028204, "lm_q2_score": 0.6370308082623217, "lm_q1q2_score": 0.5252076528363644}} {"text": "HIT <- 1 \nSTICK <- 2\nBJCard <- function()\n return(sample(10,1))\n\nStateInput <- function () {\n return ( c(sample(10, 1), sample(10, 1), 0))\n}\nStepFunc <- function (s, a) {\n if(s[3]==1)\n return(list(s, 0))\n \n NewState <- s\n BJReward <- 0\n \n if(a==1) { \n NewState[2] <- s[2] + BJCard() \n if (NewState[2]>21 || NewState[2]<1) { \n NewState[3] <- 1\n BJReward <- -1\n }\n } \n else { \n NewState[3] <- 1\n DealerWork <- FALSE\n DealerSum <- s[1]\n while(!DealerWork) { \n DealerSum <- DealerSum + BJCard()\n if (DealerSum>21) { \n DealerWork <- TRUE\n BJReward <- 1\n } else if (DealerSum >= 17) { \n DealerWork <- TRUE\n if(DealerSum==s[2])\n BJReward <- 0\n else \n BJReward <- 2*as.integer(DealerSum\"){\n currX <- currX+1\n }\n map[currX+xyMin,currY+xyMin] <- map[currX+xyMin,currY+xyMin] + 1\n }\n return(sum(map>0))\n}\n\nstopifnot(nVisited(input)==2572) # solution A\n\n\nsolve.B <- function(input){\n stopifnot(nchar(input)%%2==0) # make sure sanda and RoboSanta get the same amount of presents\n \n\n xyMin = 100 # increased until the code dont give errors anymore\n map <- matrix(0, nrow = 2*xyMin, ncol = 2*xyMin)\n currX = 0\n currY = 0\n roboX = 0\n roboY = 0\n map[currX+xyMin,currY+xyMin] <- map[currX+xyMin,currY+xyMin] + 1\n \n nInstructions = nchar(input)\n for (i in 1:(nInstructions/2)){\n zerosIdxI = i-1\n instrIdx = 2*zerosIdxI +1 \n roboInstrIdx = instrIdx +1 \n # print(c(i,zerosIdxI,instrIdx,roboInstrIdx))\n instr = substr(input,instrIdx,instrIdx)\n roboInstr = substr(input,roboInstrIdx,roboInstrIdx)\n # print(c(instr,roboInstr))\n if (instr == \"<\"){\n currX <- currX -1\n } else if (instr==\"^\"){\n currY <- currY+1\n }else if (instr==\"v\"){\n currY <- currY-1\n }else if (instr==\">\"){\n currX <- currX+1\n }\n if (roboInstr == \"<\"){\n roboX <- roboX -1\n } else if (roboInstr==\"^\"){\n roboY <- roboY+1\n }else if (roboInstr==\"v\"){\n roboY <- roboY-1\n }else if (roboInstr==\">\"){\n roboX <- roboX+1\n }\n map[currX+xyMin,currY+xyMin] <- map[currX+xyMin,currY+xyMin] + 1\n map[roboX+xyMin,roboY+xyMin] <- map[roboX+xyMin,roboY+xyMin] + 1\n }\n ans <- sum(map>0)\n # print(c(currX,currY,roboX,roboY,ans))\n return(ans)\n}\n\nstopifnot(solve.B(\"^v\")==3,solve.B(\"^>v<\")==3,solve.B(\"^v^v^v^v^v\")==11,\nsolve.B(input)==2631 # answer B\n)", "meta": {"hexsha": "3f3d41f1708173a0df8d5545d7b8b370d9455cc2", "size": 2446, "ext": "r", "lang": "R", "max_stars_repo_path": "day3/main.r", "max_stars_repo_name": "el-hult/adventofcode2015", "max_stars_repo_head_hexsha": "f84fd9c776310c33dde14c9fe2640310ccad0416", "max_stars_repo_licenses": ["CC0-1.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "day3/main.r", "max_issues_repo_name": "el-hult/adventofcode2015", "max_issues_repo_head_hexsha": "f84fd9c776310c33dde14c9fe2640310ccad0416", "max_issues_repo_licenses": ["CC0-1.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "day3/main.r", "max_forks_repo_name": "el-hult/adventofcode2015", "max_forks_repo_head_hexsha": "f84fd9c776310c33dde14c9fe2640310ccad0416", "max_forks_repo_licenses": ["CC0-1.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 29.4698795181, "max_line_length": 97, "alphanum_fraction": 0.5433360589, "num_tokens": 822, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.815232489352, "lm_q2_score": 0.6442250996557036, "lm_q1q2_score": 0.5251932316953596}} {"text": "\\name{Jeong}\n\\alias{Jeong}\n%- Also NEED an '\\alias' for EACH other topic documented here.\n\\title{\n Jeong's method for estimating the preferential attachment function\n}\n\\description{\n This function estimates the preferential attachment function by Jeong's method. \n}\n\\usage{\nJeong(net_object , \n net_stat = get_statistics(net_object) , \n T_0_start = 0 ,\n T_0_end = round(net_stat$T * 0.75) ,\n T_1_start = T_0_end + 1 ,\n T_1_end = net_stat$T ,\n interpolate = FALSE)\n}\n%- maybe also 'usage' for other objects documented here.\n\\arguments{\n \\item{net_object}{\n an object of class \\code{PAFit_net} that contains the network.\n }\n \\item{net_stat}{\n An object of class \\code{PAFit_data} which contains summerized statistics needed in estimation. This object is created by the function \\code{\\link{get_statistics}}. Default value is \\code{ get_statistics(net_object)}.\n }\n \\item{T_0_start}{Positive integer. The starting time-step of the \\code{T_0_interval}. Default value is \\code{0}.}\n \\item{T_0_end}{Positive integer. The ending time-step of \\code{T_0_interval}. Default value is \\code{round(net_stat$T * 0.75)}.}\n \n \\item{T_1_start}{Positive integer. The starting time-step of the \\code{T_1_interval}. Default value is \\code{T_0_end + 1}.}\n \\item{T_1_end}{Positive integer. The ending time-step of \\code{T_1_interval}. Default value is \\code{net_stat$T}.}\n \n \\item{interpolate}{\n Logical. If \\code{TRUE} then all the gaps in the estimated PA function are interpolated by linear interpolating in logarithm scale. Default value is \\code{FALSE}.\n }\n}\n\\value{\n Outputs an \\code{PA_result} object which contains the estimated attachment function. In particular, it contains the following field:\n \\itemize{\n \\item \\code{k} and \\code{A}: a degree vector and the estimated PA function.\n \n \\item \\code{center_k} and \\code{theta}: when we perform binning, these are the centers of the bins and the estimated PA values for those bins. \n \\item \\code{g}: the number of bins used.\n \\item \\code{alpha} and \\code{ci}: \\code{alpha} is the estimated attachment exponenet \\eqn{\\alpha} (when assume \\eqn{A_k = k^\\alpha}), while \\code{ci} is the confidence interval.\n \\item \\code{loglinear_fit}: this is the fitting result when we estimate \\eqn{\\alpha}. \n}\n}\n\\author{\n Thong Pham \\email{thongpham@thongpham.net}\n}\n\\references{\n 1. Jeong, H., \\enc{Néda}{Neda}, Z. & \\enc{Barabási}{Barabasi}, A. . Measuring preferential attachment in evolving networks. Europhysics Letters. 2003;61(61):567–572. doi: 10.1209/epl/i2003-00166-9 (\\url{https://iopscience.iop.org/article/10.1209/epl/i2003-00166-9}).\n}\n\\seealso{\n\n See \\code{\\link{get_statistics}} for how to create summerized statistics needed in this function.\n\nSee \\code{\\link{Newman}} and \\code{\\link{only_A_estimate}} for other methods to estimate the attachment function in isolation.\n\n}\n\n\\examples{\n library(\"PAFit\")\n net <- generate_net(N = 1000 , m = 1 , mode = 1 , alpha = 1 , s = 0)\n net_stats <- get_statistics(net)\n result <- Jeong(net, net_stats)\n # true function\n true_A <- result$center_k\n #plot the estimated attachment function\n plot(result , net_stats)\n lines(result$center_k, true_A, col = \"red\") # true line\n legend(\"topleft\" , legend = \"True function\" , col = \"red\" , lty = 1 , bty = \"n\")\n}\n\\concept{preferential attachment}\n\\concept{attachment function}\n", "meta": {"hexsha": "f35eb58aeccb829e6e2aeb411303f1f4a1a16fbc", "size": 3501, "ext": "rd", "lang": "R", "max_stars_repo_path": "man/Jeong.rd", "max_stars_repo_name": "eddelbuettel/PAFit", "max_stars_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "man/Jeong.rd", "max_issues_repo_name": "eddelbuettel/PAFit", "max_issues_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "man/Jeong.rd", "max_forks_repo_name": "eddelbuettel/PAFit", "max_forks_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 46.0657894737, "max_line_length": 268, "alphanum_fraction": 0.6918023422, "num_tokens": 981, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7122321964553657, "lm_q2_score": 0.7371581684030623, "lm_q1q2_score": 0.5250277814167275}} {"text": "#' @param X matrix. Rows are genes. Columns are samples. Row names are symbols.\n#' @param gene_sets list. Each element is a string vector with gene symbols.\n#' @param alpha numeric. Parameter for ssGSEA, the default is 0.25\n#' @param scale logical. If True, normalize the scores by number of genes in the gene sets.\n#' @param norm logical. If True, normalize the scores by the absolute difference between max and min values.\n#' @param single logical. If True, use ssGSEA algorithm, otherwise use GSEA.\n#'\n#' @return matrix containing enrichment scroes. Rows are gene sets, columns are samples.\n#'\n#' @examples\n#' # Create a fake matrix\n# m = 100\n# n = 100\n# set.seed(1)\n# X = matrix(rnorm(m*n), m, n)\n# # # Assign 'gene symbols' to row names\n# rownames(X) = 1:m\n# # Create 3 gene sets\n# gene_sets = list(a = sample(m, 5), b = sample(m, 5), c = sample(m, 5))\n# system.time(assign('a', GSVA::gsva(X, gene_sets, method = 'ssgsea')))\n# system.time(assign('b', ssgsea(X, gene_sets, scale = F, norm = T)))\n# identical(a, b)\nset.seed(1)\nssgsea = function(X, gene_sets, alpha = 0.25, scale = T, norm = F, single = T) {\n row_names = rownames(X)\n num_genes = nrow(X)\n gene_sets = lapply(gene_sets, function(genes) {which(row_names %in% genes)})\n\n # Ranks for genes\n R = matrixStats::colRanks(X, preserveShape = T, ties.method = 'average')\n\n # Calculate enrichment score (es) for each sample (column)\n es = apply(R, 2, function(R_col) {\n gene_ranks = order(R_col, decreasing = TRUE)\n\n # Calc es for each gene set\n es_sample = sapply(gene_sets, function(gene_set_idx) {\n # pos: match (within the gene set)\n # neg: non-match (outside the gene set)\n indicator_pos = gene_ranks %in% gene_set_idx\n indicator_neg = !indicator_pos\n\n rank_alpha = (R_col[gene_ranks] * indicator_pos) ^ alpha\n\n step_cdf_pos = cumsum(rank_alpha) / sum(rank_alpha)\n step_cdf_neg = cumsum(indicator_neg) / sum(indicator_neg)\n\n step_cdf_diff = step_cdf_pos - step_cdf_neg\n\n # Normalize by gene number\n if (scale) step_cdf_diff = step_cdf_diff / num_genes\n\n # Use ssGSEA or not\n if (single) {\n sum(step_cdf_diff)\n } else {\n step_cdf_diff[which.max(abs(step_cdf_diff))]\n }\n })\n unlist(es_sample)\n })\n\n if (length(gene_sets) == 1) es = matrix(es, nrow = 1)\n\n # Normalize by absolute diff between max and min\n if (norm) es = es / diff(range(es))\n\n # Prepare output\n rownames(es) = names(gene_sets)\n colnames(es) = colnames(X)\n return(es)\n}\n\n# setwd(\"~/github/GSEApy/tests/data\")\n# comparison with gseapy.ssgsea\nX2 = read.table(\"./data/testSet_rand1200.gct\", row.names = 1, header = T,\n comment='#', sep=\"\\t\", stringsAsFactors = F)\nX3 = as.matrix.data.frame(X2[,-1])\ngene_sets = fgsea::gmtPathways(\"./data/randomSets.gmt\")\n\n# test\nsystem.time(assign('a', GSVA::gsva(X3, gene_sets, method = 'ssgsea')))\nsystem.time(assign('b', ssgsea(X3, gene_sets, scale = F, norm = T)))\nsystem.time(assign('b2', ssgsea(X3, gene_sets, scale = T, norm = F)))\n\nidentical(a, b)\nprint(\"TEST with GSVA::gsva(method='ssgsea')\")\nprint(a)\n\nprint(\"\\n\\n\\n\")\nprint(\"TEST with ssgsea, NES\")\nprint(b)\n\nprint(\"\\n\\n\\n\")\nprint(\"TEST with ssgsea, Scaled ES\")\nprint(b2)\n", "meta": {"hexsha": "8abf646407c64fb8952913835b70ecbc87552e59", "size": 3222, "ext": "r", "lang": "R", "max_stars_repo_path": "tests/test.ssgsea.R.r", "max_stars_repo_name": "pirakd/GSEApy", "max_stars_repo_head_hexsha": "5fb660a77bba509ea88663ee6243df6d2aad6b69", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": 206, "max_stars_repo_stars_event_min_datetime": "2018-06-05T13:27:38.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-31T05:08:52.000Z", "max_issues_repo_path": "tests/test.ssgsea.R.r", "max_issues_repo_name": "pirakd/GSEApy", "max_issues_repo_head_hexsha": "5fb660a77bba509ea88663ee6243df6d2aad6b69", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": 92, "max_issues_repo_issues_event_min_datetime": "2018-06-01T13:52:59.000Z", "max_issues_repo_issues_event_max_datetime": "2022-01-21T17:44:14.000Z", "max_forks_repo_path": "tests/test.ssgsea.R.r", "max_forks_repo_name": "pirakd/GSEApy", "max_forks_repo_head_hexsha": "5fb660a77bba509ea88663ee6243df6d2aad6b69", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": 67, "max_forks_repo_forks_event_min_datetime": "2018-06-12T13:32:14.000Z", "max_forks_repo_forks_event_max_datetime": "2022-02-16T10:02:32.000Z", "avg_line_length": 33.2164948454, "max_line_length": 108, "alphanum_fraction": 0.6648044693, "num_tokens": 969, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104789086703224, "lm_q2_score": 0.6477982179521103, "lm_q1q2_score": 0.525026792724406}} {"text": "#' Is a set of chains sufficiently converged?\n#'\n#' Determines if a set of MCMC chains is sufficiently converged based on the proportion of coefficients for which rhat is < a threshold value.\n#' @param mcmc An object of class \\code{\\link[coda]{mcmc.list}}.\n#' @param rhatThresh Numeric, value of rhat below which a chain is considered to be converged (default = 1.1).\n#' @param minConv Numeric in the range [0, 1], proportion of rhat values to consider a set of chains \"sufficiently\" converged. Typically this is 0 (default).\n#' @param ignore Character or character vector, names of nodes to ignore when calculating rhat. These can include regex expressions (e.g., \\code{x[[]}). Default is \\code{NULL} (calculate rhat for all nodes).\n#' @return List with three elements:\n#' \\itemize{\n#' \t\t\\item Logical (if proportion of rhats is below the threshold).\n#' \t\t\\item Proportion of nodes that did not converge.\n#'\t\t\\item Vector of node names that did not converge.\n#'\t\t\\item Vector of node names with \\code{NA} for rhat value.\n#' @export\n\nrhatStats <- function(mcmc, rhatThresh = 1.1, minConv = 0, ignore = NULL) {\n\n\tnchains <- length(mcmc)\n\t\n\tif (!is.null(ignore)) {\n\t\tfor (chain in 1:nchains) {\n\t\t\tfor (ig in ignore) {\n\t\t\t\tmcmc[[chain]] <- mcmc[[chain]][ , !grepl(colnames(mcmc[[chain]]), pattern=ig)]\n\t\t\t}\n\t\t}\n\t}\n\t\n\trhat <- wiqid::simpleRhat(mcmc, nchains)\n\n\tpropUnconv <- sum(rhat >= rhatThresh, na.rm=TRUE) / sum(!is.na(rhat))\n\tsufficient <- (propUnconv < minConv)\n\tunconv <- names(rhat[rhat > rhatThresh])\n\tnas <- names(rhat)[is.na(rhat)]\n\n\treturns <- list(sufficient=sufficient, propUnconv=propUnconv, unconv=unconv, nas=nas, rhat=rhat)\n\treturns\n\n}\n", "meta": {"hexsha": "f2797e065bdad93195cac5707ae25859f7a1ce73", "size": 1651, "ext": "r", "lang": "R", "max_stars_repo_path": "code/rhatStats.r", "max_stars_repo_name": "adamlilith/tropicosMassModeling", "max_stars_repo_head_hexsha": "4772fcb431283224b90544ecdbdb5ee4513b8d86", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "code/rhatStats.r", "max_issues_repo_name": "adamlilith/tropicosMassModeling", "max_issues_repo_head_hexsha": "4772fcb431283224b90544ecdbdb5ee4513b8d86", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": 4, "max_issues_repo_issues_event_min_datetime": "2020-07-10T23:53:19.000Z", "max_issues_repo_issues_event_max_datetime": "2020-07-10T23:54:10.000Z", "max_forks_repo_path": "code/rhatStats.r", "max_forks_repo_name": "adamlilith/tropicosMassModeling", "max_forks_repo_head_hexsha": "4772fcb431283224b90544ecdbdb5ee4513b8d86", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 42.3333333333, "max_line_length": 207, "alphanum_fraction": 0.7013930951, "num_tokens": 479, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.7799929002541067, "lm_q2_score": 0.6723316860482762, "lm_q1q2_score": 0.5244139417335285}} {"text": "require(Matrix)\n\n##\n# sample reported deaths after rep.day (if Inf) predicit day\n# assuming Multionimal distribution\n#\n# samples - (int) how many samples should be done for each N (cheap)\n# deaths - (N x 1) true number of deaths each day\n# P - (N x N) matrix of probabilites (only upper triangular part relevant)\n# Reported - (N x N) matrix of reported deaths cumlative (only upper triangular relevant)\n# alpha - (N x 1) stepsizes\n# true.dat - (int) how many days after recording should one sample\n##\nsample.deaths <- function(samples, deaths, P, Reported, alpha, true.day = 0){\n\n N <- length(deaths)\n acc <- rep(0,N)\n for(i in 1:N){\n if(i > true.day){\n P_i = P[i,i:N]\n Reported_i = Reported[i,i:N]\n index = is.na(Reported_i)==F\n P_i = P_i[index]\n Reported_i = Reported_i[index] \n lik_i <- loglikDeathsGivenProb(deaths[i], P_i, Reported_i)\n for(j in 1:samples){\n death_star <- sample((deaths[i]-alpha[i]):(deaths[i]+alpha[i]), 1)\n lik_start <- loglikDeathsGivenProb(death_star, P_i, Reported_i)\n if(log(runif(1)) < lik_start-lik_i){\n lik_i = lik_start\n deaths[i] <- death_star\n acc[i] = acc[i] + 1\n }\n }\n }\n }\n return(list(deaths=deaths, acc = acc))\n}\n\n\n##\n# sample reported deaths after rep.day (if Inf) predicit day using Beta binomial dist\n#\n# samples - (int) how many samples should be done for each N (cheap)\n# deaths - (N x 1) true number of deaths each day\n# alpha - (N x N) matrix of beta binom parameter (only upper triangular part relevant)\n# beta - (N x N) matrix of beta binom parameter (only upper triangular part relevant)\n# Reported - (N x N) matrix of reported deaths cumlative (only upper triangular relevant)\n# alpha.MCMC - (N x 1) stepsizes\n# true.dat - (int) how many days after recording should one sample\n# prior - (function) should return log of prior density (use N and i)\n##\nsample.deathsBB <- function(samples, deaths, alpha, beta, Reported, alpha.MCMC, true.day = 0 , prior=NULL){\n \n N <- length(deaths)\n acc <- rep(0,N)\n for(i in 1:N){\n if(i > true.day){\n alpha_i = alpha[i,i:N]\n beta_i = beta[i,i:N]\n Reported_i = Reported[i,i:N]\n index = is.na(Reported_i)==F\n alpha_i = alpha_i[index]\n beta_i = beta_i[index]\n Reported_i = Reported_i[index] \n \n lik_i <- loglikDeathsGivenProbBB(deaths[i],alpha_i , beta_i, Reported_i) \n if(is.null(prior)==F)\n lik_i <- lik_i + prior(deaths[i], i)\n for(j in 1:samples){\n death_star <- sample((deaths[i]-alpha.MCMC[i]):(deaths[i]+alpha.MCMC[i]), 1)\n \n lik_star <- loglikDeathsGivenProbBB(death_star, alpha_i , beta_i, Reported_i) \n if(is.null(prior)==F)\n lik_star <- lik_star + prior(death_star, i)\n \n if(log(runif(1)) < lik_star-lik_i){\n lik_i = lik_star\n deaths[i] <- death_star\n acc[i] = acc[i] + 1\n }\n }\n }\n }\n return(list(deaths=deaths, acc = acc))\n}\n\n##\n# fill report using binimoal beta\n#\n#\n##\nfill.ReportBB <- function(deaths, Alpha, Beta, Reported, maxusage.day){\n N <- length(deaths)\n N_2 <- dim(Alpha)[2]\n for(i in 1:N){\n Alpha_i = Alpha[i,i:N_2]\n Beta_i = Beta[i,i:N_2]\n Reported_i = Reported[i,i:N_2]\n index = is.na(Reported_i)==T \n for(j in min(which(index)):length(Reported_i)){\n p <- rbeta(1, Alpha_i[j], Beta_i[j])\n \n if(i> maxusage.day){\n Reported_i[j] <- rbinom(1, size=deaths[i] - Reported_i[j-1], prob = p) + Reported_i[j-1]\n }else{\n Reported_i[j] <- Reported_i[j-1]\n }\n }\n Reported[i,i:N_2] = Reported_i\n }\n return(Reported)\n}\n\n##\n# fill report\n#\n#\n##\nfill.Report <- function(deaths, P, Reported){\n N <- length(deaths)\n N_2 <- dim(P)[2]\n for(i in 1:N){\n P_i = P[i,i:N_2]\n Reported_i = Reported[i,i:N_2]\n index = is.na(Reported_i)==T \n for(j in min(which(index)):length(Reported_i)){\n Reported_i[j] <- rbinom(1, size=deaths[i] - Reported_i[j-1], prob = P_i[j]) + Reported_i[j-1]\n }\n Reported[i,i:N_2] = Reported_i\n }\n return(Reported)\n}\n###\n#\n# posterior sampling of deaths given Prob vec\n#\n# P - (N x N) probability matrix over probability of detecting reminder\n# Reported - (N x N) matrix of reported deaths cumlative (only upper triangular relevant)\n# Predict.day - (int) which day to of reporting to predict\n# sim - (2 x 1) MCMC samples inner loop and outer loop\n# alpha - (N x 1) stepsizes\n#\n###\ndeath.givenProb <- function(P, Reported ,Predict.day = Inf,sim=c(2000,10), alpha = NULL){\n\n N <- dim(P)[1]\n if(is.null(alpha))\n alpha <- rep(4, N)\n\n deaths <- matrix(NA, nrow=sim[1], ncol = N)\n deaths_est <- apply(Reported, 1, max, na.rm=T)\n\n burnin = ceiling(0.3*sim[1])\n for(i in 1:(sim[1] + burnin -1)){\n\n res <- sample.deaths(sim[2],deaths_est, P, Reported, alpha,rep.day=Predict.day)\n deaths_est <- res$deaths\n if(i < burnin){\n alpha[res$acc/sim[2] > 0.3] <- alpha[res$acc/sim[2] > 0.3] +1\n alpha[res$acc/sim[2] < 0.3] <- alpha[res$acc/sim[2] < 0.3] -1\n alpha[alpha<1] <- 1\n }else{\n deaths[i - burnin +1,] = deaths_est\n }\n }\n return(deaths)\n}\n\n\n\n\n##\n# log liklihood of obseving report given death and prob\n# deaths - (int) true number of deaths\n# p - (n x 1) probability of report\n# report - (n x 1) reported deaths culmative each date\n##\nloglikDeathsGivenProb <- function(death, p, report){\n\n if(death < max(report,na.rm=T))\n return(-Inf)\n n <- length(p)\n if(is.na(report[1])){\n #we dont have data from day one\n\n ndeaths <- diff(report[is.na(report)==F])\n report_adj <- report[1:(n-1)]\n report_adj <- report_adj[is.na(report_adj)==F]\n }else{\n ndeaths <- c(report[1],diff(report))\n if(n>1){\n report_adj <- c(0, report[1:(n-1)])\n }else{\n report_adj <- 0\n }\n }\n\n ndeaths[ndeaths <0 ] = 0\n\n return(sum(dbinom(ndeaths, death - report_adj, prob = p[is.na(p)==F], log=T )))\n}\n\n##\n# build holiday covariates vector for holiday\n#\n##\n# holidays - (N x 1) true if day is holiday false else\n##\nbuildXholiday <- function(N,holidays){\n ##\n # base matrix\n ##\n index_base <- t(matrix(rep(holidays,N),ncol = N))\n\n index_base <- index_base[upper.tri(index_base,diag=T)]\n index_base <- 1* index_base\n #sparse matrix\n i_base <- 1:length(index_base)\n i_base <- i_base[index_base==1]\n return(sparseMatrix(j=rep(1,length(i_base)),i=i_base, dims=c(length(index_base), 1)))\n}\n\nbuildXall <- function(N){\n ##\n # base matrix\n ##\n index_base <- t(matrix(1:N^2,ncol = N, nrow=N))\n\n index_base <- index_base[upper.tri(index_base,diag=T)]\n #sparse matrix\n i_base <- 1:length(index_base)\n j_base <- 1:length(index_base)\n return(sparseMatrix(j=j_base,i=j_base))\n}\n\n##\n# build day effect matrix\n# nDayEffects - number of speical days effect (1- first day, 2- first + second day)\n# N - number of days\n# nDayEffects - how many day covariate effect to create\n##\nbuildXdayeffect <- function(N, nDayEffects = 1){\n nDayEffects <- min(N,nDayEffects)\n index_days <- toeplitz(c( (1:nDayEffects), rep(0,N -nDayEffects)))\n index_days <- index_days[upper.tri(index_days,diag=T)]\n j_ <- index_days[index_days>0]\n i_base <- 1:length(index_days)\n i_ <- i_base[index_days>0]\n return(sparseMatrix(i=i_,j=j_ , dims=c(length(index_days), nDayEffects) ) )\n}\n##\n# build day mixed effect matrix\n# nDayEffects - number of speical days effect (1- first day, 2- first + second day)\n# N - number of days\n# nDayEffects - how many day covariate effect to create\n##\nbuildXmixeddayeffect <- function(N, nDayEffects = 1){\n\n index = rep(0,N)\n index[nDayEffects] = 1\n index_days <- toeplitz(index)\n index_days <- index_days[upper.tri(index_days,diag=T)]\n j_ <- 1:sum(index_days[index_days>0])\n i_base <- 1:length(index_days)\n i_ <- i_base[index_days>0]\n return(sparseMatrix(i=i_,j=j_ , dims=c(length(index_days), sum(index_days[index_days>0]))) )\n}\n##\n# building an X matrix such that each day has a fixed effect\n#\n##\nbuildXday <- function(N){\n ##\n # base matrix\n ##\n index_base <- t(matrix(rep(1:N,N),ncol = N))\n index_base <- index_base[upper.tri(index_base,diag=T)]\n #sparse matrix\n j_base <- 1:length(index_base)\n #adding mean effect\n j_ <- c(index_base, rep(N+1,length(index_base)))\n i_ <- c(j_base, 1:length(index_base))\n #day effects\n if(nDayEffects > 0){\n index_days <- toeplitz(c( N+1 + (1:nDayEffects), rep(0,N -nDayEffects)))\n index_days <- index_days[upper.tri(index_days,diag=T)]\n j_ <- c(j_, index_days[index_days>0])\n i_ <- c(i_, j_base[index_days>0])\n }\n return(sparseMatrix(i=i_,j=j_))\n}\n##\n# transforms data,\n# we remove data if negative..\n#\n# deaths - (N x 1) how many has died (thruth)\n# reports - (N x N) reported cumlative deaths\n# maxusage.day - (int) only use data up to maxusage.days i.e.\n# reports[i,i + maxusage.day - 1]\n#\n##\nnewDeaths <-function(deaths, reports,maxusage.day = -1){\n newreport <- reports\n diff.report <- t(diff(t(reports)))\n newreport[upper.tri(newreport)] <- diff.report[upper.tri(diff.report,T)]\n newreport[newreport<0 & is.na(newreport)==F]=0 #fake\n death.rem <- diag(deaths)\n dr<-deaths-reports\n N <- length(deaths)\n dr <- dr[1:(N-1),1:(N-1)]\n death.rem[upper.tri(newreport)] <- dr[upper.tri(dr,T)]\n death.rem[lower.tri(death.rem)] <- NA\n diag(death.rem)[is.na(diag(reports))] <- NA\n if(maxusage.day>0){\n for(i in 1:length(death.rem)){\n if(i + maxusage.day - 1 < N ){\n death.rem[i,(i+maxusage.day):N] = NA\n newreport[i,(i+maxusage.day):N] = NA\n }\n }\n }\n #removing small bugs in reporting\n newreport[death.rem <0 & is.na(death.rem)==F] = 0\n death.rem[death.rem <0& is.na(death.rem)==F] = 0\n index <- (death.rem< newreport) & is.na(death.rem) ==F\n newreport[index] =death.rem[index] \n return(list(death.remain = death.rem, report.new = newreport))\n}\n\n\n##\n# likelihood observations given deaths and probabilites\n# model is:\n# newreport[upper.tri] \\sim Bin(deaths[upper.tri], logit(X\\beta))\n#\n#' @return obj - list\n#' $loglik - logliklihood\n#' $grad - gradient\n#' $hessian - Hessian of the likelihood\n##\nloglikProb <- function(beta, death.remain, report.new, X){\n\n N <- dim(report.new)[1]\n \n index <- upper.tri(death.remain,diag = T)\n y = report.new[index]\n n = death.remain[index]\n index = is.na(y)==F\n X <- X[index,]\n p = as.vector(1/(1+exp(-X%*%beta)))\n y <- y[index]\n n <- n[index]\n lik <- sum(dbinom(y,size = n, prob = as.vector(p), log = T))\n\n grad <- as.vector(t(X)%*%as.vector(y-n*p))\n\n Hessian <- -t(X)%*%diag(as.vector(n*p*(1-p)))%*%X\n Hessian <- diag(diag(Hessian))\n\n return(list(loglik = lik, grad = grad, Hessian = Hessian))\n}\nloglikPrior <- function(beta, n, sigma, res){\n n <- length(sigma)\n res$grad[1:n] <- res$grad[1:n]-beta[1:n]/(sigma^2)\n diag(res$Hessian) <- diag(res$Hessian) - c(1/sigma^2, rep(0,length(beta)-n))\n res$loglik <- res$loglik -sum(beta[1:n]^2/(2*sigma^2))\n return(res)\n}\n\nloglik <- function(beta,death.remain, report.new, X, sigma){\n res <- loglikProb(beta, death.remain, report.new, X)\n res <- loglikPrior(beta, n, sigma, res)\n return(res)\n}\n\n\n#####\n##\n# beta binomial part\n##\n#####\n\n\n\n##\n# likelihood observations given deaths and probabilites\n# model is:\n# newreport[upper.tri] \\sim BB(deaths[upper.tri], exp(X\\beta_1), exp(X\\beta_2))\n#\n#' @return obj - list\n#' $loglik - logliklihood\n#' $grad - gradient\n##\nloglikProbBB <- function(beta, death.remain, report.new, X, calcH=T){\n \n p <- dim(X)[2]\n beta_1 <- beta[1:p]\n beta_2 <- beta[(p+1):(2*p)]\n N <- dim(report.new)[1]\n \n index <- upper.tri(death.remain,diag = T)\n y = report.new[index]\n n = death.remain[index]\n index = is.na(y)==F\n X <- X[index,]\n alpha <- exp(X%*%beta_1)\n beta <- exp(X%*%beta_2) #unfortunate name\n if(min(1/(alpha+beta)) <0.001)\n return(list(loglik = -Inf, grad = 0, Hessian = 0))\n y <- y[index]\n n <- n[index]\n \n lik <- sum(dBB(y, size = n, alpha = alpha, beta = beta, log.p=T))\n if(is.na(lik))\n return(list(loglik = -Inf, grad = 0, Hessian = 0))\n \n grad_lik <- grad_dBB(y, size = n, alpha = alpha, beta = beta)\n grad_alpha <- as.vector(t(X)%*%(alpha*grad_lik$grad_alpha))\n grad_beta <- as.vector(t(X)%*%(beta*grad_lik$grad_beta))\n if(calcH){\n H_ <- Hessian_dBB(y, size = n, alpha = alpha, beta = beta)\n H_12 <- t(X)%*%diag(as.vector(alpha*beta*H_$grad_alpha_beta))%*%X\n H_11 <- t(X)%*%diag(as.vector(alpha^2*H_$grad_alpha_alpha))%*%X\n H_22 <- t(X)%*%diag(as.vector(beta^2*H_$grad_beta_beta))%*%X\n Hessian <- rbind(cbind(H_11, H_12) ,\n cbind(H_12, H_22) )\n Hessian <- diag(diag(Hessian) )\n diag(Hessian) <- diag(Hessian)\n }else{\n Hessian =NULL\n }\n grad = c(grad_alpha,grad_beta)\n if(sum(is.na(grad))>0)\n return(list(loglik = -Inf, grad = 0, Hessian = 0))\n return(list(loglik = lik, grad = grad, Hessian = Hessian))\n}\n##\n# PRobability of beta binomial\n##\ndBB<- function(x, size, alpha, beta, log.p = F){\n const = lgamma(size + 1) - lgamma(x + 1) -\n lgamma(size - x + 1)\n ld <- const + \n lgamma(x + alpha) + lgamma(size - x + beta) -\n lgamma(size + alpha + beta) + \n lgamma(alpha + beta) - \n lgamma(alpha) - lgamma(beta)\n if(log.p==F){\n return(exp(ld))\n }\n return(ld)\n}\n##\n# gradient of log of probability beta binomial\n# return list, \n# [[1]] grad alpha\n# [[2]] grad beta\n##\ngrad_dBB<- function(x, size, alpha, beta){\n grad_alpha <- digamma(x + alpha) -\n digamma(size + alpha + beta) + \n digamma(alpha + beta) -\n digamma(alpha)\n grad_beta <- digamma(size - x + beta) -\n digamma(size + alpha + beta) + \n digamma(alpha + beta) -\n digamma(beta)\n return(list(grad_alpha = grad_alpha,\n grad_beta = grad_beta ))\n}\n##\n# Hessian of log of probability beta binomial\n# return list, \n# [[1]] grad alpha\n# [[2]] grad beta\n##\nHessian_dBB<- function(x, size, alpha, beta){\n grad_alpha_alpha <- trigamma(x + alpha) -\n trigamma(size + alpha + beta) + \n trigamma(alpha + beta) -\n trigamma(alpha)\n grad_beta_beta <- trigamma(size - x + beta) -\n trigamma(size + alpha + beta) + \n trigamma(alpha + beta) -\n trigamma(beta)\n \n grad_alpha_beta <- -trigamma(size + alpha + beta) + \n trigamma(alpha + beta) \n return(list(grad_alpha_alpha = grad_alpha_alpha,\n grad_beta_beta = grad_beta_beta,\n grad_alpha_beta = grad_alpha_beta))\n}\n\n##\n# log liklihood of obseving report given death and prob\n# density is Beta binomial\n# deaths - (int) true number of deaths\n# alpha - (n x 1) bb parameter 1\n# beta - (n x 1) bb parameter 2\n# report - (n x 1) reported deaths culmative each date\n##\nloglikDeathsGivenProbBB <- function(death, alpha, beta, report){\n \n if(death < max(report,na.rm=T))\n return(-Inf)\n n <- length(alpha)\n if(is.na(report[1])){\n #we dont have data from day one\n \n ndeaths <- diff(report[is.na(report)==F])\n report_adj <- report[1:(n-1)]\n report_adj <- report_adj[is.na(report_adj)==F]\n }else{\n ndeaths <- c(report[1],diff(report))\n if(n>1){\n report_adj <- c(0, report[1:(n-1)])\n }else{\n report_adj <- 0\n }\n }\n \n ndeaths[ndeaths <0 ] = 0\n \n return(sum(dBB(x =ndeaths, \n size = death - report_adj, \n alpha = alpha[is.na(alpha)==F], \n beta = beta[is.na(alpha)==F],\n log.p=T)))\n}\n\n\n###\n#\n# posterior sampling of deaths given Prob vec\n#\n# alpha - (N x N) matrix of beta binom parameter (only upper triangular part relevant)\n# beta - (N x N) matrix of beta binom parameter (only upper triangular part relevant)\n# Reported - (N x N) matrix of reported deaths cumlative (only upper triangular relevant)\n# Predict.day - (int) which day to of reporting to predict\n# sim - (2 x 1) MCMC samples inner loop and outer loop\n# alpha - (N x 1) stepsizes\n# prior - (function) prior for N (needs use N and i)\n#\n###\ndeath.givenParamBB <- function(alpha, beta, Reported,Predict.day = Inf,sim=c(2000,10), alpha.MCMC = NULL, prior=NULL){\n \n N <- dim(alpha)[1]\n if(is.null(alpha.MCMC))\n alpha.MCMC <- rep(4, N)\n \n deaths <- matrix(NA, nrow=sim[1], ncol = N)\n deaths_est <- apply(Reported, 1, max, na.rm=T)\n \n burnin = ceiling(0.3*sim[1])\n for(i in 1:(sim[1] + burnin -1)){\n res <- sample.deathsBB(sim[2],\n deaths_est, \n alpha,\n beta, \n Reported, \n alpha.MCMC,\n rep.day=Predict.day,\n prior= prior)\n deaths_est <- res$deaths\n if(i < burnin){\n alpha.MCMC[res$acc/sim[2] > 0.3] <- alpha.MCMC[res$acc/sim[2] > 0.3] +1\n alpha.MCMC[res$acc/sim[2] < 0.3] <- alpha.MCMC[res$acc/sim[2] < 0.3] -1\n alpha.MCMC[alpha.MCMC<1] <- 1\n }else{\n deaths[i - burnin +1,] = deaths_est\n }\n }\n return(deaths)\n}\n\n##\n# buidling the X matrix\n#\n#\n# unique.prob - (int) how many of days shall we have unique probabilites\n##\nsetup_data <- function(N, Predict.day, dates_report, unique.prob=NULL){\n holidays.Sweden <- as.Date(c(\"2020-04-10\",\"2020-04-13\",\"2020-05-01\"))\n X <- buildXdayeffect(N,Predict.day)\n #holidays and weekends\n \n result$dates_report <- as.Date(dates_report)\n holidays <- weekdays(dates_report)%in%c(\"Sunday\",\"Saturday\") | c(dates_report)%in%c(holidays.Sweden)\n \n holidays.tommorow <- weekdays(dates_report + 1)%in%c(\"Sunday\",\"Saturday\") |\n (result$dates_report+1)%in%c(holidays.Sweden)\n Xhol <- buildXholiday(N,holidays)\n #Xhol.tommrow <- buildXholiday(N,holidays.tommorow)\n if(is.null(unique.prob)==F)\n X <- cbind(X[,1:unique.prob],rowSums(X[,1:unique.prob])==0)\n \n X <- cbind(X, Xhol)\n return(X)\n}\n\n\n", "meta": {"hexsha": "c05aded63ed75d2cb917a0a9978ac85c234c4157", "size": 17947, "ext": "r", "lang": "R", "max_stars_repo_path": "util.r", "max_stars_repo_name": "JonasWallin/CaptureRetainCovid", "max_stars_repo_head_hexsha": "d0d80efb973ebf89ca8521fb017dcb634006f2b7", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2020-04-29T10:10:03.000Z", "max_stars_repo_stars_event_max_datetime": "2020-04-29T10:10:03.000Z", "max_issues_repo_path": "util.r", "max_issues_repo_name": "JonasWallin/CaptureRetainCovid", "max_issues_repo_head_hexsha": "d0d80efb973ebf89ca8521fb017dcb634006f2b7", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "util.r", "max_forks_repo_name": "JonasWallin/CaptureRetainCovid", "max_forks_repo_head_hexsha": "d0d80efb973ebf89ca8521fb017dcb634006f2b7", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 29.7628524046, "max_line_length": 118, "alphanum_fraction": 0.607455285, "num_tokens": 5690, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8006920211198871, "lm_q2_score": 0.6548947290421276, "lm_q1q2_score": 0.524368984217502}} {"text": "# preprocessing modis/viirs active fire\n\n# preprocessing ================================================================\n\nread_csv(\"data/fire_nrt_V1_101893.csv\")%>%\n dplyr::mutate(acq_hour = as.numeric(substr(acq_time, start = 1, stop = 2)),\n acq_min = as.numeric(substr(acq_time, start = 3, stop = 4)),\n acq_datetime = ymd_hm(paste0(acq_date, \" \", acq_hour, \":\", acq_min)),\n acq_year = year(acq_datetime),\n acq_month = month(acq_datetime),\n acq_day = day(acq_datetime),\n solar_offset = longitude / 15,\n hemisphere = ifelse(latitude >= 0, yes = \"Northern hemisphere\", no = \"Southern hemisphere\"),\n acq_datetime_local = acq_datetime + as.duration(solar_offset * 60 * 60),\n local_doy = lubridate::yday(acq_datetime_local),\n local_hour_decmin = ((acq_hour) + (acq_min / 60) + solar_offset + 24) %% 24,\n local_solar_hour_decmin_round = round(local_hour_decmin),\n local_solar_hour_decmin_round0.5 = round(local_hour_decmin * 2) / 2,\n h = (local_hour_decmin - 12) * 15 * pi / 180,\n phi = latitude * pi / 180,\n delta = -asin(0.39779 * cos(pi / 180 * (0.98565 * (local_doy + 10) + 360 / pi * 0.0167 * sin(pi / 180 * (0.98565 * (local_doy - 2)))))),\n solar_elev_ang = (asin(sin(phi)*sin(delta) + cos(phi)*cos(delta)*cos(h))) * 180 / pi,\n daynight = ifelse(solar_elev_ang > 0, yes = \"day\", no = \"night\")) %>%\n write_csv(\"data/fire_nrt_australia.csv\")\n\n# preprocessing brazil =========================================================\n\nread_csv(\"data/fire_nrt_V1_104769.csv\")%>%\n dplyr::mutate(acq_hour = as.numeric(substr(acq_time, start = 1, stop = 2)),\n acq_min = as.numeric(substr(acq_time, start = 3, stop = 4)),\n acq_datetime = ymd_hm(paste0(acq_date, \" \", acq_hour, \":\", acq_min)),\n acq_year = year(acq_datetime),\n acq_month = month(acq_datetime),\n acq_day = day(acq_datetime),\n solar_offset = longitude / 15,\n hemisphere = ifelse(latitude >= 0, yes = \"Northern hemisphere\", no = \"Southern hemisphere\"),\n acq_datetime_local = acq_datetime + as.duration(solar_offset * 60 * 60),\n local_doy = lubridate::yday(acq_datetime_local),\n local_hour_decmin = ((acq_hour) + (acq_min / 60) + solar_offset + 24) %% 24,\n local_solar_hour_decmin_round = round(local_hour_decmin),\n local_solar_hour_decmin_round0.5 = round(local_hour_decmin * 2) / 2,\n h = (local_hour_decmin - 12) * 15 * pi / 180,\n phi = latitude * pi / 180,\n delta = -asin(0.39779 * cos(pi / 180 * (0.98565 * (local_doy + 10) + 360 / pi * 0.0167 * sin(pi / 180 * (0.98565 * (local_doy - 2)))))),\n solar_elev_ang = (asin(sin(phi)*sin(delta) + cos(phi)*cos(delta)*cos(h))) * 180 / pi,\n daynight = ifelse(solar_elev_ang > 0, yes = \"day\", no = \"night\")) %>%\n write_csv(\"data/fire_nrt_brazil.csv\")\n\n# preprocessing tubbs ==============\n\n\nread_csv(\"data/fire_archive_V1_104769.csv\")%>%\n dplyr::mutate(acq_hour = as.numeric(substr(acq_time, start = 1, stop = 2)),\n acq_min = as.numeric(substr(acq_time, start = 3, stop = 4)),\n acq_datetime = ymd_hm(paste0(acq_date, \" \", acq_hour, \":\", acq_min)),\n acq_year = year(acq_datetime),\n acq_month = month(acq_datetime),\n acq_day = day(acq_datetime),\n solar_offset = longitude / 15,\n hemisphere = ifelse(latitude >= 0, yes = \"Northern hemisphere\", no = \"Southern hemisphere\"),\n acq_datetime_local = acq_datetime + as.duration(solar_offset * 60 * 60),\n local_doy = lubridate::yday(acq_datetime_local),\n local_hour_decmin = ((acq_hour) + (acq_min / 60) + solar_offset + 24) %% 24,\n local_solar_hour_decmin_round = round(local_hour_decmin),\n local_solar_hour_decmin_round0.5 = round(local_hour_decmin * 2) / 2,\n h = (local_hour_decmin - 12) * 15 * pi / 180,\n phi = latitude * pi / 180,\n delta = -asin(0.39779 * cos(pi / 180 * (0.98565 * (local_doy + 10) + 360 / pi * 0.0167 * sin(pi / 180 * (0.98565 * (local_doy - 2)))))),\n solar_elev_ang = (asin(sin(phi)*sin(delta) + cos(phi)*cos(delta)*cos(h))) * 180 / pi,\n daynight = ifelse(solar_elev_ang > 0, yes = \"day\", no = \"night\")) %>%\n write_csv(\"data/fire_archive_brazil.csv\")\n", "meta": {"hexsha": "a93a4fe3f7c62341643cac22ef9cad336bbb7858", "size": 4676, "ext": "r", "lang": "R", "max_stars_repo_path": "r/preprocessing_modis_viirs_af.r", "max_stars_repo_name": "mbjoseph/tmin", "max_stars_repo_head_hexsha": "c41d781b00e434eb98f078d4ded3410d268ef2c6", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "r/preprocessing_modis_viirs_af.r", "max_issues_repo_name": "mbjoseph/tmin", "max_issues_repo_head_hexsha": "c41d781b00e434eb98f078d4ded3410d268ef2c6", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "r/preprocessing_modis_viirs_af.r", "max_forks_repo_name": "mbjoseph/tmin", "max_forks_repo_head_hexsha": "c41d781b00e434eb98f078d4ded3410d268ef2c6", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-03-17T05:21:15.000Z", "max_forks_repo_forks_event_max_datetime": "2021-03-17T05:21:15.000Z", "avg_line_length": 64.9444444444, "max_line_length": 152, "alphanum_fraction": 0.5538922156, "num_tokens": 1321, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8006920020959544, "lm_q2_score": 0.6548947155710233, "lm_q1q2_score": 0.5243689609726232}} {"text": "#' pair_dist\n#'\n#' vectorized pair-wise distance calculation\n#'\n#' @param A numeric matrix\n#' @param B numeric matrix\n#'\n.pair_dist <- function(A, B) {\n\n an <- rowSums(A^2)\n bn <- rowSums(B^2)\n\n m <- nrow(A)\n n <- nrow(B)\n\n tmp <- matrix(rep(an, n), nrow=m)\n tmp <- tmp + matrix(rep(bn, m), nrow=m, byrow=TRUE)\n\n sqrt( tmp - 2 * tcrossprod(A,B) )\n\n}\n", "meta": {"hexsha": "a80ef3aa2c13a92a675122a9fb0c6d87b1d391fe", "size": 357, "ext": "r", "lang": "R", "max_stars_repo_path": "R/internal.r", "max_stars_repo_name": "cjtexas/citygeocoder", "max_stars_repo_head_hexsha": "32b548a84def36f2ad6bad196a786f9e9dcc752b", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-01-15T22:34:30.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-01T23:14:48.000Z", "max_issues_repo_path": "R/internal.r", "max_issues_repo_name": "cjtexas/citygeocoder", "max_issues_repo_head_hexsha": "32b548a84def36f2ad6bad196a786f9e9dcc752b", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/internal.r", "max_forks_repo_name": "cjtexas/citygeocoder", "max_forks_repo_head_hexsha": "32b548a84def36f2ad6bad196a786f9e9dcc752b", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 16.2272727273, "max_line_length": 53, "alphanum_fraction": 0.5854341737, "num_tokens": 125, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.7879311856832191, "lm_q2_score": 0.6654105521116443, "lm_q1q2_score": 0.5242977252914534}} {"text": "gevFit <- function(data, aggregation = \"annual\", nYears, nReplicates = 1, nCovariates = 0, covariates = NULL, dataScaling = 1, locationModel = NULL, scaleModel = NULL, shapeModel = NULL, missingFlag = NULL, returnParams = FALSE, rvInterval = 20, newData = NULL, rvDifference = NULL, maxes = TRUE, optimMethod = \"Nelder-Mead\"){\n # data should be a 1-d array containing the maxes (or mins) by year (year x (optionally) month x (optionally) location), with the year index varying fastest and the location index varying slowest. For \"seasonal\" aggregation, the maxes should be provided by month. For seasonal analysis, data should be given as consecutive years, with NA for missing values, as the code needs to treat December as being with the following year. If there are NAs for any months of a given season, the seasonal value is taken to be NA.\n # aggregation should be one of \"annual\", \"seasonal\", or \"monthly\", indicating the stratification. If monthly or seasonal, separate results will be reported for each stratum (i.e., each month or season)\n # nYears is the number of years (more generally of blocks) of data provided\n # nReplicates is the number of replicate data sets; primarily for use with model output where you can run the model multiple times to get independent replicates\n # nCovariates indicates the number of covariates provided through 'covariates' (note that any subset of the covariates that are provided may be used in the location, scale, and shape modeling, as specified in locationModel, scaleModel, shapeModel)\n # covariates is a 1-d array of covariate values (year x covariate x (optionally) month or season), with the observation index varying fastest and month/season index varying slowest. This should be NULL if 'nCovariates' is 0.\n # dataScaling is a positive-valued scalar used to scale the data values for more robust optimization performance. When multiplied by the values, it should produce values with magnitude around 1\n # locationModel, scaleModel, and shapeModel are vectors indicating the indices of the covariates to be used in the location, scale and shape parameterization. The values are used to select columns from the 'covariates' array after they are transformed to multidimensional arrays with the second dimension indexing the covariates\n # missingFlag is a value to be interpreted as missing values (NA in R), intended for use in other languages calling this function\n # returnParams is a boolean indicating whether to return the fitted parameter values and their standard errors; WARNINGS: (1) Parameter values for models with covariates must be interpreted based on transforming each covariate by subtracting the mean of the yearly values (from 'covariatesByYear') and dividing by the difference of the max and min of the yearly values. This scales the covariates for better numerical performance in the optimization. (2) parameter values for models with covariates for the scale parameter must interpreted based on the log transformation of the scale parameter\n # rvInterval: the timespan for which return values should be calculated. For example a rvInterval of 20 years corresponds to the value of an event that occurs with probability 1/20 in any year and therefore occurs on average every 20 years\n # newData should be a 1-d array providing covariate values (observation x covariate) for which return values are desired. Values will be calculated for each stratum.\n # rvDifference should be a 1-d array of covariate values for two sets of covariates for which the difference in return values is desired (set x covariate), with the set index varying fastest; i.e. provide the first covariate for each set, then the second covariate for each set, etc. The difference is computed as the return value for the second set minus the return value for the first set. Values will be calculated for each stratum.\n # maxes should be TRUE when analyzing block maxima and FALSE when analyzing block minima\n # optimMethod allows you to use different methods as specified in 'optim'; by default this is the derivative-free Nelder-Mead method\n\n require(ismev) # won't be needed when start using gev.fit2\n \n if(!maxes){ # modeling minima is equivalent to modeling the negative of maxima. Location parameter values will be the negative of those on the original scale, but are corrected before returning parameter values to the user.\n data = -data\n }\n\n if(!is.null(missingFlag)){\n data[data == missingFlag] <- NA\n if(!is.null(covariates))\n covariates[covariates == missingFlag] <- NA\n }\n\n nStrata = 1\n if(aggregation == \"monthly\"){\n nStrata = 12\n }\n if(aggregation == \"seasonal\"){\n nStrata = 4\n }\n \n data <- data * dataScaling\n\n # check dimensionality of input arrays\n if((aggregation == \"annual\" && length(data) != nYears*nReplicates) ||\n (aggregation != \"annual\" && length(data) != nYears*12*nReplicates))\n stop(\"length of input data does not match number of years, covariates, and months (the latter is required for seasonal and monthly analyses\")\n if(!is.null(covariates) && !(length(covariates) %in% (nYears*nCovariates*c(1, nStrata))))\n stop(\"supplied 'covariate' values do not match number of years, covariates and strata\")\n if(!is.null(newData) && length(newData) %% nCovariates != 0)\n stop(\"length of newData should be a multiple of 'nCovariates'\")\n if(!is.null(rvDifference) && length(rvDifference) != 2*nCovariates)\n stop(\"length of rvDifference is not equal to two times the number of covariates\")\n \n # manipulate input arrays to have appropriate number of dimensions\n if(aggregation == \"annual\"){\n data <- array(data, c(nYears, nStrata, nReplicates))\n } else{\n data <- array(data, c(nYears, 12, nReplicates))\n }\n if(!is.null(covariates))\n covariates <- array(covariates, c(nYears, nCovariates, nStrata))\n if(!is.null(newData)){\n m = length(newData)/nCovariates # number of return values to compute\n newData <- array(newData, c(m, nCovariates))\n }\n if(!is.null(rvDifference))\n rvDifference <- array(rvDifference, c(2, nCovariates))\n \n if(is.null(rvInterval) && (!is.null(newData) || !is.null(rvDifference)))\n stop(\"'rvInterval' must be specified\")\n \n if(aggregation == 'seasonal'){\n seasons <- c('DJF', 'MAM', 'JJA', 'SON')\n data <- seasonalize(data, maxes = TRUE)\n }\n \n nParam <- 3 + length(locationModel) + length(scaleModel) + length(shapeModel)\n \n if(!is.null(locationModel) && !validateIndices(locationModel, nCovariates))\n stop(\"'locationModel' values do not provide legitimate indices of covariates\")\n \n if(!is.null(scaleModel) && !validateIndices(scaleModel, nCovariates))\n stop(\"'scaleModel' values do not provide legitimate indices of covariates\")\n \n if(!is.null(shapeModel) && !validateIndices(shapeModel, nCovariates))\n stop(\"'shapeModel' values do not provide legitimate indices of covariates\")\n\n if(nCovariates){\n for(p in 1:nCovariates){ # shift and scale to (-.5, .5) for better numeric properties in estimation\n if(!is.null(newData))\n newData[ , p] <- normalize(newData[, p], mean(covariates[ , p, ]), min(covariates[ , p, ]), max(covariates[ , p, ]))\n if(!is.null(rvDifference))\n rvDifference[ , p] <- normalize(rvDifference[, p], mean(covariates[ , p, ]), min(covariates[ , p, ]), max(covariates[ , p, ]))\n covariates[ , p, ] <- normalize(covariates[ , p, ])\n }\n }\n # do I need to save the original covariates or at least mean and divisor of normalization?\n\n mulink <- siglink <- shlink <- identity\n link = \"c(identity, identity, identity)\"\n if(!is.null(scaleModel)){\n siglink <- exp\n link = \"c(identity, exp, identity)\"\n }\n \n NAlist <- list(mle = rep(NA, nParam), se = rep(NA, nParam), cov = matrix(NA, nParam, nParam)) \n\n gev.fit.wrap <- function(xdat, ydat){\n fit <- try(gev.fit2(xdat, ydat = ydat, mul = locationModel, sigl = scaleModel, shl = shapeModel, mulink = mulink, siglink = siglink, shlink = shlink, show = FALSE))\n if(is(fit, 'try-error') || fit$conv || fit$flag)\n fit <- NAlist\n return(fit)\n }\n\n extract <- function(index, object, name)\n object[[index]][[name]]\n\n mle <- se <- array(NA, c(nParam, nStrata))\n covmat <- array(NA, c(nParam, nParam, nStrata))\n\n for(j in 1:nStrata){\n output <- gev.fit.wrap(c(data[ , j, ]), matrix(rep(c(t(covariates[ , , j])), nReplicates), ncol = nCovariates, byrow = TRUE))\n mle[ , j] <- output[[\"mle\"]]\n if(!maxes) # location parameters for minima are the negative of those computed based on negative of minima\n mle[1:(length(locationModel)+1), ] <- -mle[1:(length(locationModel)+1), ]\n se[ , j] <- output[[\"se\"]]\n covmat[ , , j] <- output[[\"cov\"]]\n }\n\n results <- list()\n numLocScaleParams = 2 + length(locationModel) + length(scaleModel)\n # rescale parameters so on scale of original data\n mle[1:numLocScaleParams, ] <- mle[1:numLocScaleParams, ] / dataScaling\n se[1:numLocScaleParams, ] <- se[1:numLocScaleParams, ] / dataScaling\n \n if(returnParams){\n if(aggregation == 'seasonal')\n attributes(mle)$dimnames[[2]] <- attributes(se)$dimnames[[2]] <- seasons\n results$mle <- mle[ , , drop = TRUE]\n results$se.mle <- se[ , , drop = TRUE]\n }\n \n # get return values for newData observations\n # perhaps make this more efficient with an apply, but it needs to pass in both mle and cov\n if(!is.null(newData)){\n rv <- array(0, c(m, nStrata, 2))\n for(i in 1:m)\n for(j in 1:nStrata){\n fit = list(mle = mle[ , j], cov = covmat[ , , j], model = list(locationModel, scaleModel, shapeModel), link = link) \n class(fit) = \"gev.fit\"\n rv[i, j, ] <- returnValue(fit, rvInterval, newData[i, ])\n }\n if(aggregation == \"seasonal\")\n attributes(rv)$dimnames[[2]] <- seasons\n results$returnValue <- rv[ , , 1]\n results$se.returnValue <- rv[ , , 2]\n }\n\n # get stationary return value\n if(is.null(newData) && !is.null(rvInterval) && is.null(covariates)){ \n rv <- array(0, c(nStrata, 2))\n for(j in 1:nStrata) {\n fit = list(mle = mle[ , j], cov = covmat[ , , j], model = list(locationModel, scaleModel, shapeModel), link = link)\n class(fit) = \"gev.fit\"\n rv[j, ] <- returnValue(fit, rvInterval, rvCovariates = NULL)\n }\n if(aggregation == \"seasonal\")\n attributes(rv)$dimnames[[1]] <- seasons\n results$returnValue <- rv[ , 1]\n results$se.returnValue <- rv[ , 2]\n }\n \n # get return value difference\n if(!is.null(rvDifference)){\n rvDiff <- array(0, c(nStrata, 2))\n for(j in 1:nStrata) {\n fit = list(mle = mle[ , j], cov = covmat[ , , j], model = list(locationModel, scaleModel, shapeModel), link = link)\n class(fit) = \"gev.fit\"\n rvDiff[j, ] <- returnValueDiff(fit, rvInterval, rvDifference)\n }\n if(aggregation == \"seasonal\")\n attributes(rvDiff)$dimnames[[1]] <- seasons\n results$returnValueDiff <- rvDiff[ , 1]\n results$se.returnValueDiff <- rvDiff[ , 2]\n }\n \n return(results)\n}\n", "meta": {"hexsha": "aad00757de495c82eea8bf35dc57ef0238d20449", "size": 10947, "ext": "r", "lang": "R", "max_stars_repo_path": "operators/ExtremeValueAnalysis/r_src/gevVisit.r", "max_stars_repo_name": "ahota/visit_ospray", "max_stars_repo_head_hexsha": "d80b2e18ff5654d04bfb56ae4d6f42e45f87c9b9", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "operators/ExtremeValueAnalysis/r_src/gevVisit.r", "max_issues_repo_name": "ahota/visit_ospray", "max_issues_repo_head_hexsha": "d80b2e18ff5654d04bfb56ae4d6f42e45f87c9b9", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "operators/ExtremeValueAnalysis/r_src/gevVisit.r", "max_forks_repo_name": "ahota/visit_ospray", "max_forks_repo_head_hexsha": "d80b2e18ff5654d04bfb56ae4d6f42e45f87c9b9", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 58.8548387097, "max_line_length": 596, "alphanum_fraction": 0.6990042934, "num_tokens": 2982, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.841825655188238, "lm_q2_score": 0.6224593312018545, "lm_q1q2_score": 0.5240022343170336}} {"text": "#########################################\r\n# Cost accumulation mapping in R\r\n#\r\n# Andy Nelson. Professor, Spatial Agriculture and Food Security\r\n# Department of Natural Resources (Room 4-144)\r\n# ITC - Faculty of Geo-Information Science and Earth Observation of the University of Twente\r\n# PO Box 217, 7500 AE Enschede, the Netherlands\r\n\r\n# Uses scripts or ideas by \r\n# Dan Weiss, Malaria Atlas Project, University of Oxford\r\n# Jacob van Etten, Bioversity International\r\n\r\n# Adaptations for Bhutan case by\r\n# Rolf de By, GIP/ITC - Faculty of Geo-Information Science and Earth Observation \r\n# University of Twente\r\n# Adaptations:\r\n# -1- Operating system independence (win/macOS/linux)\r\n# -2- Implementation on a 150m grid basis using national crs, as the data allows for this\r\n# -2- Earlier move to metric crs in the script\r\n# -3- Taken hard-coded data out and into external (csv) files. \r\n# Such as: landcover_class_speeds.csv\r\n# -4- Inclusion of a Tobler function for vehicle travel in sloped terrain\r\n# -5- Repair of a which.min() induced error that assigned a zone id to NaN territory\r\n# -6- Implementation of a tail end of the workflow, working with population census data\r\n# and housing information that allows fine-grain match of population with travel time\r\n# zones.\r\n########################################\r\n# call the required libraries (packages)\r\nif (!require(\"raster\"))\r\n install.packages(\"raster\")\r\nif (!require(\"gdistance\"))\r\n install.packages(\"gdistance\")\r\nif (!require(\"rgdal\"))\r\n install.packages(\"rgdal\")\r\n\r\nlibrary(raster)\r\nlibrary(gdistance)\r\nlibrary(rgdal)\r\n\r\n##########################################################################################\r\n# A function we will need to test which OS we have:\r\nget_os <- function(){\r\n sysinf <- Sys.info()\r\n if (!is.null(sysinf)){\r\n os <- sysinf['sysname']\r\n if (os == 'Darwin')\r\n os <- \"osx\"\r\n } else { ## mystery machine\r\n os <- .Platform$OS.type\r\n if (grepl(\"^darwin\", R.version$os))\r\n os <- \"osx\"\r\n if (grepl(\"linux-gnu\", R.version$os))\r\n os <- \"linux\"\r\n }\r\n tolower(os)\r\n}\r\n\r\n##########################################################################################\r\n# You may need A LOT of RAM for this, depends on size of region and number of targets.\r\n# Next command is only needed for Windows. Other platforms allow dynamic mem assignation.\r\nif (get_os()=='windows') memory.limit(999999)\r\nif (get_os()=='windows') dirsep <- \"\\\\\\\\\" else dirsep <- \"/\"\r\n\r\nfilepath <- function(fpath){\r\n gsub(\"/\",dirsep,fpath)\r\n }\r\n\r\n# Be conservative on raster's memory usage; terrain() has caused crashes if we do not set\r\n# chunksize. This value has not been optimized and may work if bigger. But all depends\r\n# on your hardware.\r\nrasterOptions(chunksize=1e+06)\r\n\r\n##########################################################################################\r\n# set the working directory\r\nsetwd(filepath(\"C:\\\\UT\\\\Work\\\\Workspace\\\\CSI\\\\accessibility_course_bhutan-master\"))\r\n\r\n##########################################################################################\r\n\r\n# Create a template metric raster that we will use for various cookie cuttings\r\ncountry_template <- raster()\r\n\r\n# set the dimensions of the Bhutan area\r\ndimensions <- extent(124000, 463000, 2953600, 3127600)\r\n\r\n# extent it into the necessary dimensions\r\ncountry_template <- setExtent(country_template, dimensions)\r\n\r\n# define the resolution of the raster cells ( 150 meters)\r\nres(country_template) <- 150\r\n\r\ncrs(country_template) <- '+init=EPSG:5266'\r\n\r\n# give a value for each column of the raster\r\ncountry_template[] <- 1:ncell(country_template)\r\ncountry_template\r\n\r\n##########################################################################################\r\n# Start adding various data sets to the environment\r\n# elevation raster\r\nbhutan_dem_srtm <- raster(filepath(\"inputs/bhutan_dem_srtm.tif\"))\r\nbhutan_dem_srtm\r\nplot(bhutan_dem_srtm, main=\"DTM\")\r\n\r\n# calculate the slope using the elevation in each point of the map\r\nbhutan_dem_srtm_slope <- terrain(bhutan_dem_srtm, opt='slope', unit='degrees')\r\n#plot(bhutan_dem_srtm_slope)\r\n\r\n# write the raster with slope\r\nwriteRaster(bhutan_dem_srtm_slope, filename=filepath(\"processing/bhutan_srtm_slope.tif\"),\r\n format=\"GTiff\", overwrite=TRUE)\r\n\r\n# landcover raster\r\nlandcover <- raster(filepath(\"inputs/landcover2010.tif\"))\r\nlandcover\r\nplot(landcover, main=\"land cover classes\")\r\n\r\n# target localities for accessibility; here we have chosen 23 hospitals\r\ntargets <- readOGR(filepath(\"inputs/bhu_facilities_point.shp\"))\r\n# transform targets also to the crs of the country raster template\r\ntargets <- spTransform(targets, crs(country_template))\r\ntargets\r\nplot(targets, main=\"location of 23 hospitals\")\r\n\r\n# reproject landcover and elevation rasters to national Bhutan metric system and resample to 150m \r\nlandcover_reprojected <- projectRaster(landcover, country_template, \r\n crs = '+init=EPSG:5266', res = 150)\r\n\r\nslope_reprojected <- projectRaster(bhutan_dem_srtm_slope, country_template, \r\n crs = '+init=EPSG:5266', res = 150)\r\n\r\nlandcover\r\nlandcover_reprojected\r\n\r\nbhutan_dem_srtm_slope\r\nslope_reprojected\r\n\r\n##########################################################################################\r\n# This all is for walking the terrain, depending on landcover and slope.\r\n# convert slope and landcover to speed rasters \r\n# Landcover - reclassify the landcover classes into walking speeds (km/h)\r\n# values >= 0 and < 1 become 1, values >= 1 and < 2 become 3, etc.\r\n# do not use speed = 0 km/hr for rather obvious reasons\r\n# See example data file, csv must be a 5-column (id,from,to,becomes,description) file,\r\n# from which we immediately drop the description.\r\nlcs = read.table(\"inputs/landcover_class_speeds.csv\", \r\n colClasses=c(\"integer\",\"integer\",\"integer\",\"numeric\",\"character\"), \r\n header=TRUE, sep=\",\", row.names=1)\r\nlcs$description <- NULL\r\n# view table for checking:\r\nlcs\r\n\r\n# apply the reclassification to get travel speeds per landcover and write to file\r\nlandcover_speed <- reclassify(landcover_reprojected,lcs,right=FALSE)\r\nwriteRaster(landcover_speed, filename=filepath(\"processing/landcover_speed.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\nplot(landcover_speed, main=\"speed in km per hour over land cover classes\")\r\n\r\n##########################################################################################\r\n# Step 5\r\n\r\n# slope for walking\r\n# Tobler's walking speed is given by W = 6e^-3.5|tan(slope)+0.05|\r\n# 6 km/h is maxspeed at a slight downward angle; 0.05 rad is that optimal angle for best horizontal speed;\r\n# -3.5 is a factor of decay of maximally attainable speed as slope angle changes.\r\nslp_walk <- 6 * exp(-0.4 * abs(tan(slope_reprojected*pi/180) + 0.05))\r\nslp_walk\r\nwriteRaster(slp_walk, filename=filepath(\"processing/sloped_walk.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\n\r\n# divide by base speed of 5km/h to get speed factor and write that to a file\r\nwriteRaster(slp_walk/6.0, filename=filepath(\"processing/sloped_factor_walking.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\n# Slope adjusted walking speeds over landcover, write to file\r\nterrain_walk_spd <- landcover_speed * slp_walk/6.0\r\nwriteRaster(terrain_walk_spd, filename=filepath(\"processing/terrain_walk_speed.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\nplot(terrain_walk_spd, main=\"slope adjusted walking speed\") \r\n\r\n##########################################################################################\r\n# Step 6\r\n\r\n# slope for car driving\r\n# Experimental Tobler's car speed is given by W = 50e^-2.4|tan(slope)+0.12|\r\n# Original -2.4 gave far too much decay.\r\nslp_car <- 50 * exp(-0.4 * abs(tan(slope_reprojected*pi/180) + 0.12))\r\nwriteRaster(slp_car, filename=filepath(\"processing/sloped_cardrive.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\nplot(slp_car, main=\"slope adjusted car speed\") \r\n\r\n# read the road network shapefile; in this version, we use OSM\r\n# The integer64 option is required or values will be read as strings and handled as\r\n# numbered classes.\r\nroad_shp <- readOGR(filepath(\"inputs/osm_roads.shp\"),integer64=\"allow.loss\")\r\n# transform to crs of the country raster template\r\nroad_shp <- spTransform(road_shp, crs(country_template))\r\nroad_shp\r\nplot(road_shp,main=\"roads\")\r\n\r\n# Our footpaths current version are not good.\r\nfootpath_shp <- readOGR(filepath(\"inputs/footpaths.shp\"))\r\n# transform to the crs of the country raster template\r\nfootpath_shp <- spTransform(footpath_shp, crs(country_template))\r\nfootpath_shp\r\nplot(footpath_shp,main=\"footpaths\")\r\n\r\n# create an empty raster for road speed\r\nroad_spd <- raster()\r\n# define its dimensions\r\nroad_spd <- setExtent(road_spd, dimensions)\r\n# and resolution\r\nres(road_spd) <- res(country_template)\r\n# and crs\r\ncrs(road_spd) <- crs(country_template)\r\n\r\n# same steps for footpath\r\nfootpath_spd <- raster()\r\nfootpath_spd <- setExtent(footpath_spd, dimensions)\r\nres(footpath_spd) <- res(country_template)\r\ncrs(footpath_spd) <- crs(country_template)\r\n\r\nroad_spd\r\nfootpath_spd\r\n\r\n# assign raster cell values using the maxspeed field for roads\r\n#road_spd <- rasterize(x=road_shp,y=road_spd,field=\"maxspeed\",fun=max)\r\n#writeRaster(road_spd, filename=filepath(\"processing/road_speed.tif\"), \r\n# format=\"GTiff\", overwrite=TRUE)\r\nroad_spd <- raster(filepath(\"processing/road_speed.tif\"))\r\nroad_spd\r\n\r\n# assign raster cell values using the maxspeed field for footpaths\r\nfootpath_spd <- rasterize(x=footpath_shp,y=footpath_spd, field=\"maxspeed\", fun=max)\r\nwriteRaster(footpath_spd, filename=filepath(\"processing/footpath_speed.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\n# unclear why this is needed, but above only gives cell values equal to 1, and should be 5\r\nfootpath_spd[footpath_spd==1] <- 5\r\nfootpath_spd\r\n\r\n# other modes of transporation (train, ship, ...) could go here:\r\n\r\n##########################################################################################\r\n\r\n# if slope needs to be in play for road and footpaths speeds, here is where that happens\r\n# we are dividing by the base speed for walking (5kmph) and road (50kmph)\r\nsloped_footpath_spd <- footpath_spd * slp_walk / 5.0\r\nsloped_road_spd <- road_spd * slp_car / 50.0\r\nwriteRaster(sloped_road_spd, filename=filepath(\"processing/sloped_road_speed.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\nplot(sloped_road_spd) \r\n\r\n##########################################################################################\r\n# Step 7\r\n\r\n# merging the various (road, footpath, ...) *network* speed rasters; ensure that you \r\n# prioritize properly\r\nroad_network_spd <- merge(sloped_road_spd, sloped_footpath_spd)\r\n# write a raster with the road network (all travel modes) speed \r\nwriteRaster(road_network_spd, filename=filepath(\"processing/road_network_speed.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\n\r\n# Merge the speed components and convert from travel speed to travel cost\r\n# road takes priority over rail, rail takes priority over ship travel, it over walking \r\n# the terrain ...\r\nmerged_spd <- merge(road_network_spd,terrain_walk_spd)\r\nwriteRaster(merged_spd, filename=filepath(\"processing/merged_speed.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\nplot(merged_spd, main=\"merged speed rasters in kmph\") \r\n\r\n##########################################################################################\r\n# Step 8\r\n\r\n# convert speed in km per hr to travel time in minutes per metre\r\n# THIS IS THE FRICTION SURFACE\r\nfriction <- 1.0 / (merged_spd * 1000 / 60.0 )\r\nwriteRaster(friction, filename=filepath(\"processing/friction.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\nplot(friction, main=\"friction layer reporting time in minutes to travel one metre\") \r\n\r\n##########################################################################################\r\n# Step 9\r\n\r\n# Make the graph for 8 directions using 1/mean(merge_mins) as the conductance value \r\n# between neighbouring cells \r\nT <- transition(friction, function(x) 1/mean(x), 8) \r\n# geo-corrected version of the graph to divide the conductance by the real distance\r\n# between pixels.\r\n# FALSE OBSERVATION: In Bhutan case, the next line is not needed as we are already having\r\n# metric rasters at 150m resolution.\r\n# CORRECT OBSERVATION: GC is also needed when directions = 8 or 16 (see gdistance documentation)\r\nT.GC <- geoCorrection(T,type=\"c\")\r\n\r\n# accumulated cost calculation to the nearest target using the geo-corrected graph and \r\n# the target points. This will need the use of T.GC instead if produced above.\r\naccess_mins <- accCost(T.GC, targets)\r\n# write the resulting raster showing time in minutes to the nearest target\r\nwriteRaster(access_mins, filename=filepath(\"outputs/access_mins.tif\"), \r\n format=\"GTiff\", overwrite=TRUE)\r\nplot(access_mins, main=\"travel time in minutes to nearest hospital\")\r\nplot(targets, add=TRUE)\r\n\r\n##########################################################################################\r\n# Step 10\r\n# Compute cost allocation to map which pixel is closest to which target\r\n# derived from https://stat.ethz.ch/pipermail/r-sig-geo/2011-July/012208.html\r\n# no function for this so...make a stach of accumulated cost rasters, one per target\r\n# and then use the minimum pixel value through the stack to identify which pixel\r\n# is closest to which target\r\n\r\n# set up a stack with 2 layers, use existing data to fill it for now\r\naccCost_stack <- stack(access_mins,access_mins)\r\n# run accumulative cost for each target (targets from shapefile (3 columns, [ID X Y])\r\nfor(i in 1:length(targets)) {\r\n # grab the X and Y coord\r\n t <- cbind(targets@coords[i,1],targets@coords[i,2])\r\n # add accumulated cost to the stack\r\n accCost_stack <- stack(accCost_stack, accCost(T, t))\r\n}\r\n# remove first two layers which have dud info\r\n# THIS STACK CAN ALSO BE SAVED AS A MULTIBAND RASTER WITH ONE TRAVEL TIME LAYER PER TARGET\r\naccCost_stack <- dropLayer(accCost_stack,1:2)\r\n\r\n# make a new raster based on the layer ID (pixel values will be from 1 to n layers)\r\n# Original code had just which.min(accCost_stack) *but this assigns the ID code of one target shed\r\n# to the (no data) area outside of study space too*. In current understanding, this is a flaw\r\n# of which.min (or possibly deriving from a raster stack feed with not quite robust data).\r\n# Hence, an extra step with a mask operator.\r\n# Original code line\r\naccCost_minID <- which.min(accCost_stack)\r\n\r\n# and next, mask out area outside of study; we use reprojected_landcover but this could be any\r\n# proper raster with appropriate nodata delineation.\r\naccCost_minID_masked <- mask(accCost_minID,reprojected_landcover)\r\n\r\n# make a new raster based on minimum value and compare to raster from section 4.5 (should be the same)\r\naccCost_min <- min(accCost_stack)\r\n\r\n# make a reclass table to assign target IDs to layer IDs.\r\n# layer ID 1 becomes target ID 1 etc...\r\nm <- c(1, 2, targets@data$gid[1])\r\nfor(i in 2:length(targets)) {\r\n m <- c(m,i, i+1, targets@data$gid[i])\r\n}\r\n\r\nrclcA <- matrix(m, ncol=3, byrow=TRUE)\r\n# apply the reclassification to get allocation zone IDs - THIS IS THE COST ALLOCATION ZONE MAP\r\naccCost_zones <- reclassify(accCost_minID_masked,rclcA,right=FALSE)\r\n\r\n# write cost allocation to file\r\nwriteRaster(accCost_zones, filename=\"outputs/access_alloc.tif\", format=\"GTiff\", overwrite=TRUE)\r\n# write minimum access to file and compare to output from 4.5 - should be identical\r\nwriteRaster(accCost_min, filename=\"outputs/access_minimum.tif\", format=\"GTiff\", overwrite=TRUE)\r\n\r\nplot(accCost_zones, main=\"catchments around each hospital\")\r\nplot(targets, add=TRUE)\r\n\r\n##########################################################################################\r\n# Population data\r\n# The code below is likely somewhat baroque and will make decent R coders chuckle.\r\n# Additional data inputs required are:\r\n# -a- a fairly complete set of building centroids: buildingcentroid.shp\r\n# -b- a topologically correct set of census tracts: censustract.shp\r\n# -c- a table of average nr of people per building, per censustract: \r\n# censustract_paxperbuilding.csv\r\n# (this one should really be superfluous and piggy-back on censustract.shp)\r\n# -d- a table that provides breakpoints for traveltime zonation: traveltime_zonebreaks.csv\r\n\r\n##########################################################################################\r\n\r\n# first get the building centroids\r\n# we represent buildings by centroids (more trivial assignation to raster cells)\r\nbuilding_shp <- readOGR(filepath(\"inputs/buildingcentroid.shp\"),integer64=\"allow.loss\")\r\n\r\n# towards a raster that holds building counts\r\n# create an empty raster for buildings\r\nbuildings <- raster()\r\nbuildings <- setExtent(buildings, dimensions)\r\nres(buildings) <- res(country_template)\r\ncrs(buildings) <- crs(country_template)\r\n\r\n# assign count of buildings to raster\r\nbuildings <- rasterize(x=building_shp,y=buildings,field=\"gid\",fun='count',background=0)\r\nwriteRaster(buildings, filename=filepath(\"outputs/building_counts.tif\"), format=\"GTiff\", \r\n overwrite=TRUE)\r\n \r\n# towards a raster that holds census tract id\r\n# create an empty raster \r\ncensustracts <- raster()\r\ncensustracts <- setExtent(censustracts, dimensions)\r\nres(censustracts) <- res(country_template)\r\ncrs(censustracts) <- crs(country_template)\r\n \r\n# obtain the census tracts:\r\ncensustracts_shp <- readOGR(filepath(\"inputs/censustract.shp\"),integer64=\"allow.loss\")\r\n# in below rasterize(), just using \" field='id' \" I could not make to work, so instead:\r\n# ids <- as.integer(as.matrix(censustracts_shp@data[[\"id\"]]))\r\n# Learned later that this problem is caused by reading the id field as a string not\r\n# an integer; such is now remedied by includion of the integer64 option above.\r\n# censustracts <- rasterize(x=censustracts_shp,y=censustracts,field=ids,fun='last')\r\ncensustracts <- rasterize(x=censustracts_shp,y=censustracts,field='id',fun='last')\r\nwriteRaster(censustracts, \r\n filename=filepath(\"processing/censustract_ids.tif\"), format=\"GTiff\", \r\n overwrite=TRUE) \r\n\r\n# paxperbuilding is census tract-specific; we have the correspondence in a csv file\r\n# below line allows enough memspace for our 1000+ census tracts\r\noptions(max.print = 9999)\r\n\r\nppb = read.table(\"inputs/censustract_paxperbuilding.csv\", \r\n colClasses=c(\"integer\",\"integer\",\"numeric\"), \r\n header=TRUE, sep=\",\", row.names=1)\r\n\r\n# reclassify censustracts to their ppb numeric value\r\n# towards a raster that holds building counts\r\n# create an empty raster for buildings\r\ncensustracts_housing <- raster()\r\ncensustracts_housing <- setExtent(censustracts_housing, dimensions)\r\nres(censustracts_housing) <- res(country_template)\r\ncrs(censustracts_housing) <- crs(country_template)\r\ncensustracts_housing <- reclassify(censustracts,ppb,right=FALSE)\r\nwriteRaster(censustracts_housing, \r\n filename=filepath(\"processing/censustract_housingfactor.tif\"), format=\"GTiff\", \r\n overwrite=TRUE) \r\n \r\n# towards a raster that holds populationcounts\r\n# create an empty raster \r\npopulation <- raster()\r\npopulation <- setExtent(population, dimensions)\r\nres(population) <- res(country_template)\r\ncrs(population) <- crs(country_template) \r\n\r\npopulation <- round(censustracts_housing*buildings,digits=0)\r\nwriteRaster(population, filename=filepath(\"outputs/population.tif\"), format=\"GTiff\", \r\n overwrite=TRUE)\r\nspplot(population,main='Population of Bhutan')\r\n \r\n# Verify overall correctness of population assignation to raster cells\r\n# For Bhutan, total population censused is 601,301.\r\n# Observe that the below shows some discrepancy with the above number.\r\n# This is caused by edged effects op popwogs, and above not being fully robust method.\r\n# We compute #houses and thus ppb per popwog in the database on the basis of\r\n# popwog vector polygons, however, above we have rasterized polygons for popwogs\r\n# and those are not geometrically identical. Since popwog boundaries often run along\r\n# topographic features, such boundaries may have substantial numbers of houses ...\r\n# See below for improvements\r\ncellStats(population,stat='sum')\r\n\r\n\r\n##########################################################################################\r\n# Determine statistics and mapplot\r\n# Read the breaks from traveltime_zones csv file:\r\ntt_zones = read.table(\"inputs/traveltime_zonebreaks.csv\", \r\n colClasses=c(\"integer\",\"integer\",\"character\"), \r\n header=TRUE, sep=\",\", row.names=1)\r\nbreaks <- tt_zones$minutes\r\ncolors <- tt_zones$color\r\nplotxnames <- 1:(length(breaks)-1)\r\n\r\n# determine traveltime zones spatially by breaks given\r\ntravtime_zonation <- cut(accCost_min,breaks)\r\n# save as raster and plot for pdf export\r\nwriteRaster(travtime_zonation, \r\n filename=filepath(\"outputs/travtime_zonation.tif\"), format=\"GTiff\", \r\n overwrite=TRUE)\r\nspplot(travtime_zonation,main='Travel time to hospitals zonation')\r\n\r\n# determine traveltime zonal statistics\r\nzonestats <- zonal(population,travtime_zonation,fun='sum')\r\n\r\nbarplot(t(zonestats)[2,],main='Population at travel time to reach hospital',\r\n names.arg=plotxnames, xlab='Travel time [x 30min]',\r\n ylab='Population', col=colors)\r\n\r\nspplot(accCost_min,main='Travel time to nearest hospital')\r\n\r\n##########################################################################################\r\n# Determine statistics and mapplot for one hospital shed, namely #23\r\n\r\npopzone <- raster()\r\npopzone <- setExtent(popzone, dimensions)\r\nres(popzone) <- res(country_template)\r\ncrs(popzone) <- crs(popzone)\r\n\r\npopzone <- population\r\npopzone <- mask(popzone,landcover_reprojected)\r\npopzone[accCost_zones!=18] <- NA\r\npopzone <- trim(popzone)\r\n# create ttz to have same extent as popzone\r\nttz <- crop(travtime_zonation,popzone)\r\npopzonestats <- zonal(popzone,ttz,fun='sum')\r\n\r\n# Obtain ed popdata above for one hospital shed; now plto the map and the barplot\r\n\r\nbarplot(t(popzonestats)[2,],main='Hospital shed #18 population at travel time to reach hospital',\r\n names.arg=plotxnames, xlab='Travel time [x 30min]',\r\n ylab='Population', col=colors)\r\n \r\nspplot(popzone, main='Population in hospital shed #18', xlab='Density for 150x150 m.')\r\n\r\n# As well as the at-risk areas\r\npoptravzone <- raster()\r\npoptravzone <- setExtent(poptravzone, dimensions)\r\nres(poptravzone) <- res(country_template)\r\ncrs(poptravzone) <- crs(country_template)\r\npoptravzone <- travtime_zonation\r\npoptravzone[accCost_zones!=18] <- NA\r\npoptravzone <- trim(poptravzone)\r\nspplot(poptravzone, main='Populations at risk in hospital shed #23')\r\n\r\n##########################################################################################\r\n# the barplots for all service areas:\r\n# should still be made independent of constants such as 17\r\n\r\nzero17 <- matrix(c(1,0,2,0,3,0,4,0,5,0,6,0,7,0,8,0,9,0,10,0,11,0,12,0,13,0,\r\n 14,0,15,0,16,0,17,0), byrow=TRUE, nrow=17, ncol=2)\r\nfor(i in 1:length(targets)) {\r\n popzone <- raster()\r\n popzone <- setExtent(popzone, dimensions)\r\n res(popzone) <- res(country_template)\r\n crs(popzone) <- crs(popzone)\r\n\r\n popzone <- population\r\n popzone <- mask(popzone,landcover_reprojected)\r\n popzone[accCost_zones!=i] <- NA\r\n popzone <- trim(popzone)\r\n # create ttz to have same extent as popzone\r\n ttz <- crop(travtime_zonation,popzone)\r\n # determine those stats per ttime zone\r\n popzonestats <- zonal(popzone,ttz,fun='sum')\r\n # pad results with zeroes to length 17 if needed\r\n if ( length(popzonestats[,1])>=17)\r\n { pzs <- popzonestats }\r\n else \r\n { pzs <- rbind(popzonestats,zero17[(length(popzonestats[,1])+1):17,]) }\r\n barplot(t(pzs)[2,],\r\n main=paste('Hospital shed #',i,'population at travel time to reach hospital'),\r\n names.arg=plotxnames, xlab='Travel time [x 30min]',\r\n ylab='Population', col=colors)\r\n}\r\n\r\n##########################################################################################\r\n## DONE\r\n##########################################################################################\r\n# Validation; Using travel times to the capital (proxy used is hospital there)\r\nthimphu_layer <- accCost_stack@layers[[22]]\r\nwriteRaster(thimphu_layer, \r\n filename=filepath(\"outputs/thimphu_travelcost.tif\"), format=\"GTiff\", \r\n overwrite=TRUE)\r\n \r\n##########################################################################################\r\n#\r\n# POPULATION DISCREPANCY ANALYSIS\r\n#\r\n\r\npopzonestats <- matrix(c(0,0),byrow=TRUE, nrow=1, ncol=2)\r\n\r\nfor(i in censustracts_shp@data[[\"id\"]]) { \r\n popzone <- raster()\r\n popzone <- setExtent(popzone, dimensions)\r\n res(popzone) <- res(country_template)\r\n crs(popzone) <- crs(popzone)\r\n\r\n popzone <- population\r\n popzone <- mask(popzone,landcover_reprojected)\r\n popzone[censustracts!=i] <- NA\r\n popzonestats <- rbind(popzonestats, c(i,cellStats(popzone,stat='sum')))\r\n}\r\n \r\nwrite.csv(popzonestats, \"outputs/popzonestats.csv\")\r\n\r\n##########################################################################################\r\n#\r\n# Same but now counting houses per census tract:\r\n#\r\n\r\npopzonestats <- matrix(c(0,0),byrow=TRUE, nrow=1, ncol=2)\r\nbuildings <- mask(buildings,censustracts)\r\n\r\nfor(i in censustracts_shp@data[[\"id\"]]) { \r\n popzone <- raster()\r\n popzone <- setExtent(popzone, dimensions)\r\n res(popzone) <- res(country_template)\r\n crs(popzone) <- crs(popzone)\r\n\r\n popzone <- buildings\r\n popzone <- mask(popzone,censustracts)\r\n popzone[censustracts!=i] <- NA\r\n popzonestats <- rbind(popzonestats, c(i,cellStats(popzone,stat='sum')))\r\n}\r\n \r\nwrite.csv(popzonestats, \"outputs/popzonebuildings.csv\")\r\n", "meta": {"hexsha": "762f4a1147a72f8954f7fefa0b34764789cc68a4", "size": 25752, "ext": "r", "lang": "R", "max_stars_repo_path": "BHN_R_macrdb.r", "max_stars_repo_name": "gip-itc-nl/accessibility_course_bhutan", "max_stars_repo_head_hexsha": "76e9958d33d4b730657c940abb21af5b945427b9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2020-03-03T22:25:24.000Z", "max_stars_repo_stars_event_max_datetime": "2020-03-15T22:10:00.000Z", "max_issues_repo_path": "BHN_R_macrdb.r", "max_issues_repo_name": "gip-itc-nl/accessibility_course_bhutan", "max_issues_repo_head_hexsha": "76e9958d33d4b730657c940abb21af5b945427b9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "BHN_R_macrdb.r", "max_forks_repo_name": "gip-itc-nl/accessibility_course_bhutan", "max_forks_repo_head_hexsha": "76e9958d33d4b730657c940abb21af5b945427b9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 3, "max_forks_repo_forks_event_min_datetime": "2020-03-03T21:40:23.000Z", "max_forks_repo_forks_event_max_datetime": "2020-09-01T23:57:02.000Z", "avg_line_length": 43.5736040609, "max_line_length": 107, "alphanum_fraction": 0.6651522212, "num_tokens": 6295, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.8128673087708699, "lm_q2_score": 0.6442251201477016, "lm_q1q2_score": 0.5236695396570525}} {"text": "\nnotchFilt <- function(inp, n)\n{\n for (j in 1:length(inp)) {\n out = n$a[1]*inp[j] + n$a[2]*n$xm[1] + n$a[3]*n$xm[2] + n$b[1]*n$ym[1] + n$b[2]*n$ym[2]\n n$xm[2] = n$xm[1]\n n$xm[1] = inp[j]\n n$ym[2] = n$ym[1]\n n$ym[1] = out\n inp[j] = out\n }\n inp\n}\n\ngetNotch <- function(f, b, sr=16000)\n{\n n <- list(a=c(0,0,0),b=c(0,0),xm=c(0,0),ym=c(0,0))\n tpf = 2.0 * pi * f / sr\n r = 1.0 - 3.0 * b/sr\n k = (1.0 - 2.0*r*cos(tpf) + r*r) / (2.0 - 2.0*cos(tpf))\n n$a[1] = k\n n$a[3] = k\n n$a[2] = -2.0 * k * cos(tpf)\n n$b[1] = 2.0 * r * cos(tpf)\n n$b[2] = -(r*r)\n n\n}\n\ngetNarrowBand <- function(f, b, sr=16000)\n{\n n <- list(a=c(0,0,0),b=c(0,0),xm=c(0,0),ym=c(0,0))\n tpf = 2.0 * pi * f / sr\n r = 1.0 - 3.0 * b/sr\n k = (1.0 - 2.0*r*cos(tpf) + r*r) / (2.0 - 2.0*cos(tpf))\n n$a[1] = 1.0 - k\n n$a[2] = 2.0 * (k - r) * cos(tpf)\n n$a[3] = r*r - k\n n$b[1] = 2.0 * r * cos(tpf)\n n$b[2] = -(r*r)\n n\n}\n", "meta": {"hexsha": "07a2007e790b2b284ed9fbacbba53bd18d041d70", "size": 910, "ext": "r", "lang": "R", "max_stars_repo_path": "R/notchFilt.r", "max_stars_repo_name": "NemoursResearch/FormantTracking", "max_stars_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2021-09-01T14:22:23.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-25T03:47:04.000Z", "max_issues_repo_path": "R/notchFilt.r", "max_issues_repo_name": "NemoursResearch/FormantTracking", "max_issues_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/notchFilt.r", "max_forks_repo_name": "NemoursResearch/FormantTracking", "max_forks_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-08-31T18:20:28.000Z", "max_forks_repo_forks_event_max_datetime": "2021-08-31T18:20:28.000Z", "avg_line_length": 21.6666666667, "max_line_length": 91, "alphanum_fraction": 0.421978022, "num_tokens": 527, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8670357494949105, "lm_q2_score": 0.6039318337259584, "lm_q1q2_score": 0.523630490098422}} {"text": "library(lubridate)\nlibrary(chron)\nlibrary(timeDate)\n\n\"sensor_repository_00800\"\n\nsensorData <- c(\"2459231.812\", \"2459231.833\", \"2459231.854\", \"2459231.875\", \"2459231.896\", \"2459231.917\")\nas.numeric(knownDate)\nknownDate <- sensorData[1]\nknownDateTime <- as.POSIXlt(as.numeric(knownDate)*86400,\n origin=structure(-210866760000,\n class=c(\"POSIXct\", \"POSIXt\"),\n tzone=\"Australia/Perth\"),\n tz=\"Australia/Perth\")\nknownDateTime #\"2021-01-17 15:29:16 AWST\"\n\n# 2459216 days between 4713 BC and 2021-01-17\n\n\n\n# 1721424 days between 4713 BC and 0001-01-01\n# 738,174 days between 0001-01-01 and 2021-01-17\n1721424+738174\n2459598 - 2459216\n2459216- 2459194\n\naug_birth <- ymd(\"0000-09-23\") - years(63 - 1)\n\nparse_bce_ymd <- function(str) {\n regex <- \"(\\\\d{4})(-\\\\d{2}-\\\\d{2})\"\n match <- stringr::str_match(str, regex)\n years_n <- readr::parse_number(match[, 2]) - 1 # Beware the -1 here\n right_side <- match[, 3]\n date <- ymd(paste0(\"0000-\",right_side)) - lubridate::years(years_n)\n return(date)\n}\n# Test the function.\naug_birth <- parse_bce_ymd(\"4713-01-01\")\naug_death <- ymd(\"2021-01-17\")\nage <- aug_death - aug_birth\nage\nas.numeric(age)\n\nlubridate::duration(age, \"seconds\")\n\nduration <- as.duration(age)\nduration$\n#> [1] \"2395353600s (~75.9 years)\"\n# Yay that's correct!\n\n\n\n# 1720693 days from 4713 BC to 0000\njulian(d = 17, x = 1, y = 2021, origin = c(month = 1, day = 1, year = 0000))\n# 738172 days since 0000\n1720693 + 738172\n# 2458865 days since 4713 BC\n# off by 366!\n2458865\n2459231.812\n", "meta": {"hexsha": "ca4617c67ea8317c437e726edd5f90012cc132e7", "size": 1654, "ext": "r", "lang": "R", "max_stars_repo_path": "deprecated/shiny/R/time_date_workings.r", "max_stars_repo_name": "gilesnknight/swan.canning.dashboard", "max_stars_repo_head_hexsha": "aa29b436b3c66ffd3651f039278654ecd7ebc384", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "deprecated/shiny/R/time_date_workings.r", "max_issues_repo_name": "gilesnknight/swan.canning.dashboard", "max_issues_repo_head_hexsha": "aa29b436b3c66ffd3651f039278654ecd7ebc384", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "deprecated/shiny/R/time_date_workings.r", "max_forks_repo_name": "gilesnknight/swan.canning.dashboard", "max_forks_repo_head_hexsha": "aa29b436b3c66ffd3651f039278654ecd7ebc384", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 27.1147540984, "max_line_length": 106, "alphanum_fraction": 0.6221281741, "num_tokens": 570, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8615381987656672, "lm_q2_score": 0.6076631698328916, "lm_q1q2_score": 0.5235250327940651}} {"text": "\n###############################################################################\n### Estimating Population-level density dependence and MSY reference points ###\n### by incorporating individual-level (post-recruit) density dependence #######\n############################### Shota Nishijima ###############################\n\n### set workind directory ---- \n\ngitdir = \"~/git/masaba_dd\"\n\nsetwd(gitdir)\n\nsavedir = \"res\"\n\nif (!file.exists(savedir)) dir.create(savedir)\n\nsavename = function(x) paste0(savedir,\"/\",x)\n\n### read packages ----\n\n# devtools::install_github(\"ichimomo/frasyr\", ref=\"dev\")\n\nlibrary(frasyr)\nlibrary(tidyverse)\nlibrary(MASS)\nlibrary(MuMIn)\noptions(na.action = \"na.fail\")\nlibrary(\"dplyr\", character.only = TRUE)\n\nlibrary(broom.mixed)\nlibrary(betareg)\nlibrary(lmtest)\n# install.packages(\"effects\")\n# library(effects)\n# install.packages(\"investr\")\n# library(investr)\n\n### read and handle data ----\n\n# vpa estimates from the stock assessment in Japan\nvpares = get(load(\"data/vpa_masaba_P2021.rda\"))\n\n\nwaa_dat = expand.grid(Age=as.numeric(rownames(vpares$naa)),\n Year=as.numeric(colnames(vpares$naa))) %>% \n mutate(Weight=as.numeric(unlist(vpares$input$dat$waa)),\n Maturity=as.numeric(unlist(vpares$input$dat$maa)))\n\nA = waa_dat$Age %>% max\n\nwaa_prev = sapply(1:nrow(waa_dat), function(i) {\n if (waa_dat$Age[i]==0 | waa_dat$Year[i]==min(waa_dat$Year)) {\n value <- NA\n } else {\n if (waa_dat$Age[i]%\n mutate(Weight_prev=waa_prev,\n Number_prev=num_prev/1000) %>%\n mutate(logN_prev=log(num_prev)) %>%\n mutate(YearClass=Year-Age) %>%\n mutate(interact_norm = Weight_prev*Number_prev,\n interact_log = Weight_prev*logN_prev) %>%\n filter(Year% mutate(pred=tmp$fit)\n\n## figure weight growth ----\n\nbase_size=12\npoint_size=1.5\npath_size=1.2\n\nwaa_dat2$Number_prev %>% summary()\nnewdata_wg = expand.grid(Weight_prev=seq(min(waa_dat2$Weight_prev)*1,max(waa_dat2$Weight_prev)*1,length=100),\n Number_prev=c(quantile(waa_dat2$Number_prev,probs=c(0.1,0.9)),mean(waa_dat2$Number_prev))) %>%\n as.data.frame()\n\nnewdata_wg = newdata_wg %>% mutate(Weight=predict(mod_w_growth,newdata=newdata_wg))\n\n(g_wg = ggplot(data=NULL,aes(x=Weight_prev,y=Weight))+\n geom_point(data=waa_dat2,aes(colour=Number_prev),size=point_size)+xlim(0,NA)+ylim(0,NA)+\n geom_path(data=newdata_wg,aes(colour=Number_prev,group=Number_prev),size=path_size)+\n scale_colour_gradient(low=\"deepskyblue\",high=\"sienna1\",name=\"Abundance\")+\n theme_bw(base_size=base_size)+\n xlab(\"Weight in year t-1\")+ylab(\"Weight in year t\")\n)\n\nggsave(g_wg,filename=savename(\"weight_growth.png\"),dpi=600,height=100,width=150,unit=\"mm\")\n\nsave(g_wg,file=savename(\"weight_growth_graph.rda\"))\n\n## initial weight modeling ----\n\nw0_dat = waa_dat %>% filter(Age==0 & Year>min(Year))\n\n# w0_dat$Year\n# waa_dat$Year\n\nplot(Weight~Number_prev,data=w0_dat,log=\"x\")\n\nfull_w0 = glm(Weight~Number_prev+log(Number_prev),data=w0_dat,family=Gamma(\"identity\"))\nsummary(full_w0)\n\ndredge_w0 = dredge(full_w0,subset=!(\"Number_prev\" && \"log(Number_prev)\"))\n\nhead(dredge_w0,100)\n\nsave(dredge_w0,file=savename(\"dredge_w0.rda\"))\n\nmodel_sel_w0 = model.sel(dredge_w0,beta=\"sd\")\nwrite.csv(model_sel_w0,file=savename(\"AICc_table_w0.csv\"))\n\nmod_w0 = get.models(dredge_w0,subset=1)[[1]]\nsummary(mod_w0)\n\nnewdata_w0 = expand.grid(Number_prev=exp(seq(log(min(w0_dat$Number_prev)),log(max(w0_dat$Number_prev)),length=200))) %>%\n as.data.frame()\n\ntmp = predict(mod_w0,newdata=newdata_w0,se.fit=TRUE)\n\nnewdata_w0 = newdata_w0 %>% mutate(Weight=tmp$fit,SE=tmp$se.fit) %>% \n mutate(Upper=Weight+1.96*SE,Lower=Weight-1.96*SE)\n\n(g_w0 = ggplot(data=NULL,aes(y=Weight,x=Number_prev))+\n geom_ribbon(data=newdata_w0,aes(ymax=Upper,ymin=Lower),alpha=0.4)+\n geom_point(data=w0_dat,size=point_size)+ylim(0,NA)+\n geom_path(data=newdata_w0,size=path_size)+\n # scale_colour_gradient(low=\"deepskyblue\",high=\"sienna1\",name=\"Abundance\")+\n theme_bw(base_size=base_size)+\n ylab(\"Weight at age 0\")+xlab(\"Abundance (billion)\")+\n scale_x_log10()\n)\n\nsave(g_w0,file=savename(\"weight_age0_graph.rda\"))\n\nggsave(g_w0,filename=savename(\"weight_age0.png\"),dpi=600,height=100,width=150,unit=\"mm\")\n\nsave.image(\".RData\")\n\n\nwaa_dat = waa_dat %>% \n mutate(Maturity_c = (Maturity*(n()-1)+0.5)/n())\nwaa_dat$Maturity_c\n\nvpares$input$dat$maa[c(1,5:7),]\nvpares$input$dat$maa[c(2:4),]\nwaa_dat3 = waa_dat %>% filter(Age>0 & Age<4)\n\nwaa_dat4 = waa_dat %>% filter(Maturity>0 & Maturity<1)\n\n\nplot(asin(waa_dat$Maturity)~waa_dat$Weight,col=waa_dat$Year)\n\n\nplot(waa_dat$Maturity~waa_dat$Weight,col=waa_dat$Year)\nplot(waa_dat$Maturity~waa_dat$Weight,col=waa_dat$Year)\n\nplot(Maturity~Weight,data=waa_dat4)\n\ndbeta(1,2,5,log=TRUE)\n?betareg\nmod10 = betareg(Maturity_c~Weight,data=waa_dat,link=\"logit\",type=\"BC\")\nsummary(mod10)\nAIC(mod10)\nwaa_dat4\n\nmod11 = betareg(Maturity_c~log(Weight),data=waa_dat,link=\"logit\",type=\"BC\")\nsummary(mod11)\n\nmod12 = betareg(Maturity_c~Age,data=waa_dat,link=\"logit\",type=\"BC\")\nsummary(mod12)\n\n\nAICc(mod10,mod11,mod12)\n\nwaa_dat5 = waa_dat %>% filter(Year>min(Year))\n\nmod13 = betareg(Maturity_c~Weight,data=waa_dat5,link=\"logit\",type=\"BC\")\nsummary(mod13)\nAIC(mod10)\nwaa_dat4\n\nmod14 = betareg(Maturity_c~log(Weight),data=waa_dat5,link=\"logit\",type=\"BC\")\nsummary(mod14)\n\nmod15 = betareg(Maturity_c~Age,data=waa_dat5,link=\"logit\",type=\"BC\")\nsummary(mod15)\n\nmod16 = betareg(Maturity_c~Age+Number_prev,data=waa_dat5,link=\"logit\",type=\"BC\")\nsummary(mod16)\n\nmod17 = betareg(Maturity_c~Age+logN_prev,data=waa_dat5,link=\"logit\",type=\"BC\")\nsummary(mod17)\n\n\nmod18 = betareg(Maturity_c~Age*Number_prev,data=waa_dat5,link=\"logit\",type=\"BC\")\nsummary(mod18)\n\nmod19 = betareg(Maturity_c~Age*logN_prev,data=waa_dat5,link=\"logit\",type=\"BC\")\nsummary(mod19)\n\n\nAICc(mod13,mod14,mod15,mod16,mod17,mod18,mod19)\n\n\nw0 = as.numeric(vpares$input$dat$waa[1,])\nw0 <- w0[-c(1,length(w0))]\n\nplot(w0)\n\nunique(waa_dat2$Number_prev) %>% length()\nlength(w0)\n\nplot(w0~unique(waa_dat2$Number_prev))\n\nplot(w0~unique(waa_dat2$Number_prev),log=\"x\")\n\n# 交互作用無し\n\nAICc(mod10,mod11)\n\nvpares$input$dat$maa\n\n\n\nsummary(mod10)\nAIC(mod10)\n\n\nwaa_dat\n\nhist(waa_dat2$logN_prev)\n\nobj_fun <- function(x) {\n w_inf = exp(x[1])\n k = exp(x[2])\n alpha = w_inf*(1-exp(-k))\n rho=exp(-k)\n w_pred = alpha+rho*waa_dat2$Weight_prev\n rss = sum((waa_dat2$Weight-w_pred)^2)\n return(rss)\n}\n\nplot(Weight~Weight_prev,data=waa_dat2)\n\nsummary(glm(Weight~Weight_prev,data=waa_dat2))\n\nobj_fun(x)\n\nopt = optim(x,obj_fun)\nopt$par[1] %>% exp\nopt$par[2] %>% exp\n\n# res_ysdata = get.SPR(vpares)\n# res_ysdata$ysdata\n# \n# dat = data.frame(Year =as.numeric(colnames(vpares$naa)),\n# R=as.numeric(vpares$naa[1,])/1000,\n# N = as.numeric(colSums(vpares$naa[-1,]))/1000,\n# B=as.numeric(colSums(vpares$baa[-1,]))/1000,\n# SSB=as.numeric(colSums(vpares$ssb[]))/1000,\n# catch=as.numeric(colSums(vpares$input$dat$caa[-1,]*vpares$input$dat$waa[-1,]))/1000)\n# \n# \n# dat = dat %>% bind_cols(res_ysdata$ysdata)\n# \n# dat = dat %>% filter(Year < max(Year)) %>% \n# mutate(F = -log(SPR/SPR0)) %>%\n# mutate(Weight_ypr = YPR/(1-exp(-F)))\n# \n# plot(dat$SPR0 ~ dat$N)\n# plot(dat$Weight_ypr ~ dat$N)\n# \n# plot(dat$SPR0 ~ dat$N,log=\"y\")\n# plot(dat$Weight_ypr ~ dat$N,log=\"y\")\n# \n# plot(dat$SPR0 ~ dat$N,log=\"xy\")\n# plot(dat$Weight_ypr ~ dat$N,log=\"xy\")\n# \n# summary(glm(log(dat$SPR0) ~ dat$N))\n# summary(glm(log(dat$SPR0) ~ log(dat$N)))\n# \n# summary(glm(log(dat$Weight_ypr) ~ dat$N))\n# summary(glm(log(dat$Weight_ypr) ~ log(dat$N)))\n# \n# \n# plot(dat$SPR0 ~ dat$SSB,log=\"y\")\n# plot(dat$Weight_ypr ~ dat$SSB, log=\"y\")\n# \n# plot(dat$SPR0 ~ dat$SSB,log=\"\")\n# plot(dat$Weight_ypr ~ dat$SSB, log=\"\")\n# \n# # exp(6.74)\n# summary(glm(log(dat$SPR0) ~ dat$SSB))\n# summary(glm(log(dat$SPR0) ~ log(dat$SSB)))\n# \n# summary(glm(log(dat$Weight_ypr) ~ dat$SSB))\n# summary(glm(log(dat$Weight_ypr) ~ log(dat$SSB)))\n# \n# \n# \n# \n# \n# # -log(0.5)\n# \n\n# dat = data.frame(Year =as.numeric(colnames(vpares$naa)),\n# R=as.numeric(vpares$naa[1,])/1000,\n# N = as.numeric(colSums(vpares$naa[,]))/1000,\n# B=as.numeric(colSums(vpares$baa[,]))/1000,\n# SSB=as.numeric(colSums(vpares$ssb[]))/1000,\n# w0=as.numeric(vpares$input$dat$waa[1,]),\n# catch=as.numeric(colSums(vpares$input$dat$caa[]*vpares$input$dat$waa[]))/1000,\n# catch_number=as.numeric(colSums(vpares$input$dat$caa[]))/1000) %>%\n# mutate(F=-log(1-(catch_number/N)*exp(-0.4/2))) %>%\n# mutate(S = exp(-F-0.4)) %>% filter(Year < max(Year))\n# \n# \n# nyear = nrow(dat)\n# \n# y = dat$B-dat$w0*dat$R\n# \n# dat$y <- c(y[-1],NA)\n# dat$log_y = log(dat$y)\n# \n# data=dat\n# \n# x = rep(0.,3)\n# \n# obj_fun = function(x,data=dat,dens_effect=TRUE,logN=TRUE,out=FALSE) {\n# nyear = nrow(dat)\n# N = data$N\n# B = data$B\n# y = data$B-data$w0*data$R\n# y = y[-1]\n# log_y= log(y)\n# log_k = rep(x[2],nyear)\n# if (isTRUE(dens_effect)) {\n# if (isTRUE(logN)) {\n# log_k = log_k + x[3]*log(N)\n# } else {\n# log_k = log_k + x[3]*N\n# }\n# }\n# k = exp(log_k)\n# w_inf = exp(x[1])\n# alpha = w_inf*(1-exp(-k))\n# rho = exp(-k)\n# data = data %>% \n# mutate(F=-log(1-(catch_number/N)*exp(-0.4/2))) %>%\n# mutate(S = exp(-F-0.4))\n# S = data$S\n# pred_y = S*alpha*N+S*rho*B\n# pred_y = pred_y[-1]\n# pred_logy = log(pred_y)\n# sigma = sqrt(sum((log_y-pred_logy)^2)/length(log_y))\n# nll = -sum(dnorm(log_y,pred_logy,sd=sigma,log=TRUE))\n# if (out==FALSE) {\n# return( nll )\n# } else {\n# return(list(y=y,pred_y=pred_y,w_inf=w_inf,k=k,alpha=alpha,rho=rho,loglik=-nll,npar=length(x))) \n# }\n# }\n# \n# opt = optim(par=c(log(1000),0),obj_fun,dens_effect=FALSE,out=FALSE)\n# opt\n# tmp = obj_fun(opt$par,dens_effect=FALSE,out=TRUE)\n# tmp\n\ndat = data.frame(Year =as.numeric(colnames(vpares$naa)),\n R=as.numeric(vpares$naa[1,])/1000,\n N = as.numeric(colSums(vpares$naa[-1,]))/1000,\n B=as.numeric(colSums(vpares$baa[-1,]))/1000,\n SSB=as.numeric(colSums(vpares$ssb[]))/1000,\n catch=as.numeric(colSums(vpares$input$dat$caa[-1,]*vpares$input$dat$waa[-1,]))/1000) %>%\n mutate(Weight = B/N,Maturity=SSB/B) %>%\n mutate(logW = log(Weight),\n logitM = log(Maturity/(1-Maturity))) %>%\n mutate(t = Year-min(Year)+1,logN=log(N)) %>%\n mutate(catch_number = as.numeric(colSums(vpares$input$dat$caa[-1,]))) %>%\n mutate(Weight_catch = 1000*catch/catch_number) %>%\n filter(Year < max(Year)) %>% #2019年まで\n mutate(F = -log(1-(catch/B)*exp(-0.4/2)))\n# \n# plot(dat$Weight_catch~dat$N)\n\n### Beverton-Holt SR relationship ----\n\nSRdata=get.SRdata(vpares,years=dat$Year)\nSRdata$SSB <- SRdata$SSB/1000\nSRdata$R <- SRdata$R/1000\nresHS = fit.SR(SRdata,SR=\"HS\",AR=0,out.AR=FALSE)\nresHS$pars\n\nresBH = fit.SR(SRdata,SR=\"BH\",AR=0,out.AR=FALSE)\nresBH$pars\n\n# predict(resBH$opt)\n\nc(resHS$AICc,resBH$AICc)\n\n# using nls() CIを求めるため\n\nbh_log = function(loga,logb,x) log(exp(loga)*x)-log(1+exp(logb)*x)\n\ninit = list(loga=resBH$opt$par[1],logb=resBH$opt$par[2])\n\nbh_res = nls(log(R/SSB)~log(exp(loga))-log(1+exp(logb)*SSB),start=init,data=SRdata)\n\npred_BH = resBH$pred %>% as.data.frame() \n\ntmp = predFit(bh_res,newdata=pred_BH,se.fit=TRUE, interval = \"confidence\", level= 0.95, adjust=\"none\",k=100)\ntmp$se.fit\ntmp$fit\nplot(tmp$fit[,3])\n\npred_BH = pred_BH %>% bind_cols(as.data.frame(tmp$fit)) %>% \n mutate(pred = exp(fit)*SSB,upper=exp(upr)*SSB,lower=exp(lwr)*SSB)\n\n(g_BH = ggplot(data=NULL,aes(x=SSB))+\n geom_ribbon(data=pred_BH,aes(ymax=upper,ymin=lower),alpha=0.4)+\n geom_path(data=pred_BH,aes(y=pred))+\n geom_point(data=SRdata,aes(y=R),size=2)+\n theme_bw(base_size=12)+\n xlab(\"SSB (thousand ton)\")+ylab(\"Recruits (billion)\") +\n ggtitle(\"(a)\")\n)\n\nggsave(g_BH,filename=\"BH_CI.png\",dpi=600,unit=\"mm\",height=100,width=150)\n\n### weight and maturity regression ---- \n\n\nw0 <- glm(logW~1,data=dat)\nw1 <- glm(logW~logN,data=dat)\n\nAICc(w0,w1)\nres_w = lrtest(w0,w1)\n\n\n# m0 <- glm(logitM~1,data=dat)\n# m1 <- glm(logitM~logN,data=dat)\n# \n# summary(m1)\nres_m = lrtest(m0,m1)\n\nm0 = betareg(Maturity~1,data=dat,link=\"logit\")\nsummary(m0)\n\nm1 = betareg(Maturity~logN,data=dat,link=\"logit\",type=\"BC\")\nsummary(m1)\n\nres_m = lrtest(m0,m1)\n\nplot(dat$Maturity~predict(m1))\n\n# N=100000\n# tmp = runif(N,0,1)\n# tmp_q = qnorm(tmp,0,1)\n# hist(tmp_q)\n# tmp_q %>% mean\n# tmp_q %>% sd()\n\nlogit = function(p) log(p/(1-p))\ninv_logit = function(x) 1/(1+exp(-x))\n\n# # only latest five years\n# w2 <- glm(logW~1,data=filter(dat,Year>max(Year)-5))\n# w2$coefficients %>% exp()\n# m2 <- glm(logitM~1,data=filter(dat,Year>max(Year)-5))\n# m2$coefficients %>% inv_logit()\n\ndat_wm = dat %>% \n pivot_longer(values_to=\"Value\",cols=c(\"Weight\",\"Maturity\"),\n names_to=\"Stat\") %>%\n mutate(Stat=factor(Stat,levels=c(\"Weight\",\"Maturity\")))\n\nqnorm(p=0.975)\n\npred_w = tibble(logN=seq(min(dat$logN),max(dat$logN),length=200))\ntmp_w = allEffects(w1,xlevels=list(logN=pred_w$logN))\npred_w = pred_w %>%\n mutate(logN=seq(min(dat$logN),max(dat$logN),length=200),\n pred = tmp_w[[1]]$fit[,1],SE=tmp_w[[1]]$se) %>%\n mutate(upper=tmp_w[[1]]$upper[,1],lower=tmp_w[[1]]$lower[,1]) %>%\n mutate(N=exp(logN),Value=exp(pred),\n Upper=exp(upper),Lower=exp(lower),\n Stat = \"Weight\")\n\npred_m = tibble(logN=seq(min(dat$logN),max(dat$logN),length=200))\n# tmp_m = predict(g1,newdata=pred_m,se=TRUE,type=\"link\")\ntmp_m = allEffects(m1,xlevels=list(logN=pred_m$logN))\n# tmp_m[[1]]$fit\n# tmp_m[[1]]$transformation$inverse(tmp_m[[1]]$fit)\n# tmp_m[[1]]$se\n# dat$Maturity\npred_m = tibble(logN=seq(min(dat$logN),max(dat$logN),length=200),\n pred = tmp_m[[1]]$fit[,1],\n SE=tmp_m[[1]]$se) %>%\n mutate(upper=tmp_m[[1]]$upper[,1],\n lower=tmp_m[[1]]$lower[,1]) %>%\n mutate(N=exp(logN),Value=inv_logit(pred),\n Upper=inv_logit(upper),Lower=inv_logit(lower),\n Stat = \"Maturity\")\n\npred_wm = bind_rows(pred_w,pred_m) %>%\n mutate(Stat=factor(Stat,levels=c(\"Weight\",\"Maturity\")))\n\nbase_size=12\npoint_size=1\npath_size=1\ng_wm = ggplot(data=dat_wm,aes(x=N,y=Value))+\n geom_ribbon(data=pred_wm,aes(ymin=Lower,ymax=Upper),alpha=0.4)+\n geom_path(data=pred_wm,size=path_size)+\n geom_point(size=point_size)+\n facet_wrap(vars(Stat),scales=\"free_y\")+\n scale_x_log10()+\n # scale_y_log10()+\n theme_bw(base_size=base_size)+\n # ylim(0,NA)+\n xlab(\"Fish number (billion)\")+ylab(\"\")\ng_wm\n\n### Figure 1 ----\n\nbase_size=12\npoint_size=1.5\npath_size=1\n\n# weight figure \ndat_wm$Stat %>% unique()\n\ng_w = ggplot(data=filter(dat_wm,Stat==\"Weight\"),aes(x=N,y=Value))+\n geom_ribbon(data=filter(pred_wm,Stat==\"Weight\"),aes(ymin=Lower,ymax=Upper),alpha=0.4)+\n geom_path(data=filter(pred_wm,Stat==\"Weight\"),size=path_size)+\n geom_point(size=point_size)+\n # facet_wrap(vars(Stat),scales=\"free_y\")+\n scale_x_log10()+\n # scale_y_log10()+\n theme_bw(base_size=base_size)+\n # ylim(0,NA)+\n xlab(\"Fish number (billion)\")+ylab(\"Body weight (g)\")+\n ggtitle(\"(b)\")\ng_w\n\n\ng_m = ggplot(data=filter(dat_wm,Stat==\"Maturity\"),aes(x=N,y=Value))+\n geom_ribbon(data=filter(pred_wm,Stat==\"Maturity\"),aes(ymin=Lower,ymax=Upper),alpha=0.4)+\n geom_path(data=filter(pred_wm,Stat==\"Maturity\"),size=path_size)+\n geom_point(size=point_size)+\n # facet_wrap(vars(Stat),scales=\"free_y\")+\n scale_x_log10()+\n # scale_y_log10()+\n theme_bw(base_size=base_size)+\n # ylim(0,NA)+\n xlab(\"Fish number (billion)\")+ylab(\"Maturity rate\")+\n ggtitle(\"(c)\")\ng_m\n\n(g_BH = ggplot(data=NULL,aes(x=SSB))+\n geom_ribbon(data=pred_BH,aes(ymax=upper,ymin=lower),alpha=0.4)+\n geom_path(data=pred_BH,aes(y=pred),size=path_size)+\n geom_point(data=SRdata,aes(y=R),size=point_size)+\n theme_bw(base_size=base_size)+\n xlab(\"SSB (thousand ton)\")+ylab(\"Recruits (billion)\") +\n ggtitle(\"(a)\")\n)\n\n\ng_fig1 = gridExtra::grid.arrange(g_BH,g_w,g_m,nrow=1)\n\n\nggsave(g_fig1,filename=\"Fig_SR-weight-maturity.png\",dpi=600,height=75,width=250,unit=\"mm\")\n\n\n\n\n### time-series figure ----\n\ncolnames(dat)\n\n# dat2 = bind_rows(data.frame(Year=dat$Year,Value=dat$R,Stat=\"Fish number\",Type=\"Age 0\"),\n# data.frame(Year=dat$Year,Value=dat$N,Stat=\"Fish number\",Type=\"Older\"),\n# data.frame(Year=dat$Year,Value=dat$B,Stat=\"Biomass\",Type=\"Total\"),\n# data.frame(Year=dat$Year,Value=dat$SSB,Stat=\"Biomass\",Type=\"Spawner\"),\n# data.frame(Year=dat$Year,Value=dat$catch,Stat=\"Biomass\",Type=\"Catch\"),\n# data.frame(Year=dat$Year,Value=dat$Weight,Stat=\"Weight\",Type=\"Weight\"),\n# data.frame(Year=dat$Year,Value=dat$Maturity,Stat=\"Maturity\",Type=\"Maturity\"),\n# data.frame(Year=dat$Year,Value=dat$F,Stat=\"F\",Type=\"F\")\n# )\n\n\ntemp = frasyr::get.SPR(vpares)\nysdata = temp$ysdata[-nrow(temp$ysdata),]\n\ndat = dat %>% bind_cols(ysdata)\n\nplot(dat$SPR0/(dat$Weight*dat$Maturity))\n\nplot(dat$SPR0~as.numeric(dat$Weight*dat$Maturity),log=\"xy\")\nplot(dat$SPR0/(dat$Weight*dat$Maturity)~dat$N,log=\"xy\")\n\nsummary(glm(log(dat$SPR0/(dat$Weight*dat$Maturity))~dat$N))\nsummary(glm(log(dat$SPR0/(dat$Weight*dat$Maturity))~log(dat$N)))\n\nc1 = glm(log(dat$SPR0)~1+offset(log(dat$Weight*dat$Maturity)))\nlibrary(effects)\nplot(c1)\n\nexp(mean(log(dat$SPR0/(dat$Weight*dat$Maturity))))\n\nexp(-0.4)/(1-exp(-0.4))\n\nplot((dat$SPR0/(dat$Weight*dat$Maturity))[-1]~dat$F[-50],log=\"y\")\n\nmodel = glm(log((dat$SPR0/(dat$Weight*dat$Maturity))[-1])~dat$F[-50])\nsummary(model)\n\nmodel = glm(log((dat$SPR0/(dat$Weight*dat$Maturity))[-1])~dat$F[-50])\nsummary(model)\n\nplot(dat$Weight,log(dat$Maturity/(1-dat$Maturity)),type=\"b\")\nplot(dat$Weight,dat$Maturity,type=\"b\")\n\n\nmodel=glm(log(dat$Maturity/(1-dat$Maturity))~dat$Weight)\nsummary(model)\n\ndat2 = bind_rows(data.frame(Year=dat$Year,Value=dat$R,Stat=\"Fish number\",Type=\"Age 0\"),\n data.frame(Year=dat$Year,Value=dat$N,Stat=\"Fish number\",Type=\"Age 1+\"),\n data.frame(Year=dat$Year,Value=dat$B,Stat=\"Biomass\",Type=\"Age 1+\"),\n data.frame(Year=dat$Year,Value=dat$SSB,Stat=\"Biomass\",Type=\"Spawner\"),\n data.frame(Year=dat$Year,Value=dat$catch,Stat=\"Biomass\",Type=\"Catch\"),\n data.frame(Year=dat$Year,Value=dat$Weight,Stat=\"Weight\",Type=\"Weight\"),\n data.frame(Year=dat$Year,Value=dat$Maturity,Stat=\"Maturity\",Type=\"Maturity\"),\n data.frame(Year=dat$Year,Value=dat$F,Stat=\"F\",Type=\"F\")\n)\n\nhead(dat2)\n\n# SPR0\n\n\nplot(spr0~dat$N,log=\"\")\nplot(spr0~dat$N,log=\"x\")\nplot(spr0~dat$N,log=\"xy\")\n\nspr0_res= glm(log(spr0)~dat$logN)\nsummary(spr0_res)\n\nspr0_res2= glm(log(spr0)~dat$N)\nsummary(spr0_res2)\n\nAICc(spr0_res,spr0_res2)\n\nspr0_2 = 1:nrow(dat) %>% map_dbl(function(i) {\n temp = frasyr::calc_steepness(SR=\"BH\",rec_pars=resBH$pars,M=0.4,\n waa=c(0,dat$Weight[i]), maa=c(0,dat$Maturity[i]),\n plus_group = FALSE)\n temp[1,\"SPR0\"]\n})\n\nplot(spr0_2~dat$N,log=\"x\")\nplot(spr0_2~dat$N,log=\"xy\")\n\n\nmatplot(cbind(spr0,spr0_2))\n\nplot(spr0)\n\npath_size=0.8\nbase_size=10\n\n# theme()\n\ng1 = ggplot(data=filter(dat2,Stat==\"Fish number\"),aes(x=Year,y=Value))+\n geom_path(aes(colour=Type,linetype=Type),size=path_size)+\n theme_bw(base_size=base_size)+\n scale_colour_brewer(palette=\"Set1\",name=\"\",guide=guide_legend(nrow=1))+\n scale_linetype_discrete(name=\"\",guide=guide_legend(nrow=1))+\n ylim(0,NA)+labs(colour=NULL,linetype=NULL)+\n ylab(\"Number (billion)\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1),\n legend.margin = margin(0, 0, 0, 0),\n legend.spacing.x = unit(0, \"pt\"),\n legend.spacing.y = unit(-2, \"pt\"))\ng1\n\ng2 = ggplot(data=filter(dat2,Stat==\"Biomass\") %>% \n mutate(Type=factor(Type,levels=c(\"Age 1+\",\"Spawner\",\"Catch\"))),aes(x=Year,y=Value))+\n geom_path(aes(colour=Type,linetype=Type),size=path_size)+\n theme_bw(base_size=base_size)+\n scale_colour_brewer(palette=\"Set1\",name=\"\",guide=guide_legend(nrow=1))+\n scale_linetype_discrete(name=\"\",guide=guide_legend(nrow=1))+\n ylim(0,NA)+labs(colour=NULL,linetype=NULL)+\n ylab(\"Biomass (1000 ton)\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1),\n legend.margin = margin(0, 0, 0, 0),\n legend.spacing.x = unit(0, \"pt\"),\n legend.spacing.y = unit(-2, \"pt\"))\ng2\n\ng3 = ggplot(data=filter(dat2,Stat==\"Weight\"),aes(x=Year,y=Value))+\n # geom_path(aes(colour=Type,linetype=Type),size=path_size)+\n geom_path(size=path_size)+\n theme_bw(base_size=base_size)+\n # scale_colour_brewer(palette=\"Set1\",name=\"\")+\n # scale_linetype_discrete(name=\"\")+\n ylim(0,NA)+\n ylab(\"Body weight (g)\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1))\ng3\n\ng4 = ggplot(data=filter(dat2,Stat==\"Maturity\"),aes(x=Year,y=Value))+\n # geom_path(aes(colour=Type,linetype=Type),size=path_size)+\n geom_path(size=path_size)+\n theme_bw(base_size=base_size)+\n # scale_colour_brewer(palette=\"Set1\",name=\"\")+\n # scale_linetype_discrete(name=\"\")+\n ylim(0,NA)+\n ylab(\"Maturation rate\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1))\ng4\n\ng5 = ggplot(data=filter(dat2,Stat==\"F\"),aes(x=Year,y=Value))+\n # geom_path(aes(colour=Type,linetype=Type),size=path_size)+\n geom_path(size=path_size)+\n theme_bw(base_size=base_size)+\n # scale_colour_brewer(palette=\"Set1\",name=\"\")+\n # scale_linetype_discrete(name=\"\")+\n ylim(0,NA)+\n ylab(\"F\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1))\ng5\n\ng_time = gridExtra::grid.arrange(g1,g2,g3,g4,g5,nrow=2)\n\nggsave(g_time,filename=\"graph_timeseries.png\",dpi=600,\n unit=\"mm\",height=120,width=240)\n\n\n# pred_dat$SSB %>% range\npred_dat = data.frame(logN = seq(min(dat$logN)-1,max(dat$logN)+0.1,length=1000)) %>%\n mutate(N = exp(logN),\n Weight=exp(predict(w1,newdata=.)),\n Maturity=inv_logit(predict(m1,newdata=.))) %>%\n mutate(B = Weight*N) %>%\n mutate(SSB = Maturity*B,Density=\"Dependent\")\n\npred_dat2 = data.frame(logN = seq(min(dat$logN)-1,max(dat$logN)+0.1,length=1000)) %>%\n mutate(N = exp(logN),\n Weight=exp(predict(w0,newdata=.)),\n Maturity=inv_logit(predict(m0,newdata=.))) %>%\n mutate(B = Weight*N) %>%\n mutate(SSB = Maturity*B,Density=\"Independent\")\n\n# pred_dat3 = data.frame(logN = seq(min(dat$logN)-1,max(dat$logN)+0.1,length=1000)) %>%\n# mutate(N = exp(logN),\n# Weight=exp(predict(w2,newdata=.)),\n# Maturity=inv_logit(predict(m2,newdata=.))) %>%\n# mutate(B = Weight*N) %>%\n# mutate(SSB = Maturity*B,Type=\"DI_latest5\")\n\n# pred_dat = bind_rows(pred_dat,pred_dat2,pred_dat3)\n\npred_dat = bind_rows(pred_dat,pred_dat2)\n\nrec_pars_BH = as.list(resBH$pars)\nrec_pars_HS = as.list(resHS$pars)\n\n\n# i=1\n# calc_steepness(SR=\"HS\",rec_pars=rec_pars_HS,plus_group = TRUE,\n# waa = c(0,pred_dat$Weight[i]),\n# maa = c(0,pred_dat$Maturity[i]),\n# M = rep(0.4,2))\n# \n# calc_steepness(SR=\"BH\",rec_pars=rec_pars_BH,plus_group = TRUE,\n# waa = c(0,pred_dat$Weight[i]),\n# maa = c(0,pred_dat$Maturity[i]),\n# M = rep(0.4,2))\n\npred_dat2 = 1:nrow(pred_dat) %>% map_dfr(., function(i) {\n bind_cols(pred_dat[i,],calc_steepness(SR=\"BH\",rec_pars=rec_pars_BH,plus_group = TRUE,\n waa = c(0,pred_dat$Weight[i]),\n maa = c(0,pred_dat$Maturity[i]),\n M = rep(0.4,2))\n )\n})\n\n# pred_dat2$Type %>% unique()\n\n# pred_dat2 %>% filter(Type==\"DI_latest5\")\n\npred_dat2 = pred_dat2 %>% \n mutate(R = SSB/SPR0) %>%\n mutate(R_SR = SSB %>% map_dbl(., function(x) SRF_BH(x,a=rec_pars_BH$a,b=rec_pars_BH$b))) %>%\n mutate(SSB_excess = R_SR*SPR0 ) %>%\n mutate(SPS = SSB_excess/SSB)\n\n\n# plot(R ~ SSB,data=pred_dat2)\n\n# data_line =select(pred_dat2,R,SSB) %>% mutate(Type=\"Replacement\") %>%\n# # full_join(resBH$pred %>% mutate(Type=\"Beverton-Holt\")) %>%\n# # full_join(resHS$pred %>% mutate(Type=\"Hockey-Stick\")) %>%\n# mutate(Type = factor(Type,levels=c(\"Replacement\",\"Beverton-Holt\",\"Hockey-Stick\")))\n\ng1_BH = ggplot(data=NULL,aes(x=SSB,y=R))+\n geom_path(data=resBH$pred,size=path_size,colour=\"black\",linetype=\"longdash\")+\n geom_point(data=as.data.frame(SRdata))+\n geom_path(data=filter(pred_dat2,SSBb,b*a,x*a)\n if (SR==\"BH\") SRF <- function(x,a,b) a*x/(1+b*x)\n if (SR==\"RI\") SRF <- function(x,a,b) a*x*exp(-b*x)\n \n R_pred1 = SRF(x=SSB,a=a,b=b)\n R_pred2 = SSB/SPR\n if (out==TRUE) {\n (R_pred1-R_pred2)^2\n } else {\n Yield = n*w_pred*exp(-0.4/2)*(1-exp(-f))\n tibble(N=n,Weight=w_pred,Maturity=m_pred,B=n*w_pred,SSB=SSB,F=f,SPR=SPR,R=R_pred1,Yield=Yield,U=Yield/B)\n }\n}\n\nrange(dat$N)\nx = 0.00001\n# f = seq(0.0001,100,length=1000)\nf = c(exp(seq(log(0.001),log(10),length=901)))\n# x = f[2]\ndd_f = f %>% map_dfr(., function(x) {\n\n # tmp_n = exp(seq(log(0.1),log(100),length=31))\n # tmp_obj = tmp_n %>% map_dbl(., function(i) obj_fun(n=i,F=x,\n # model_w=w1,model_m=m1,resSR=resBH,out=TRUE))\n # int = tmp_n[c(max(1,which.min(tmp_obj)-1),min(length(tmp_obj),which.min(tmp_obj)+1))]\n int = c(0.1,1000)\n opt = optimize(obj_fun, interval = int,F=x,\n model_w=w1,model_m=m1,resSR=resBH,out=TRUE)\n\n tbl = obj_fun(opt$minimum,F=x,model_w=w1,model_m=m1,resSR=resBH,out=FALSE)\n tbl\n})\n\n# x = 0.3\n\nopt_msy = function(x,out=TRUE,model_w=w1,model_m=m1,resSR=resBH) {\n int = c(0.1,1000)\n opt = optimize(obj_fun, interval = int,F=x,\n model_w=model_w,model_m=model_m,resSR=resSR,out=TRUE)\n tbl = obj_fun(opt$minimum,F=x,model_w=model_w,model_m=model_m,resSR=resSR,out=FALSE)\n if (out==TRUE) {\n -pull(tbl,Yield)[1]\n } else {\n tbl\n }\n}\n\nmsy_dd_opt = optimize(opt_msy,interval=c(0.1,2),model_w=w1,model_m=m1,resSR=resBH)\nmsy_dd = opt_msy(msy_dd_opt$minimum,model_w=w1,model_m=m1,resSR=resBH,out=FALSE)\n\nopt_msy(10,model_w=w0,model_m=m0,resSR=resBH)\n\nmsy_di_opt = optimize(opt_msy,interval=c(0.1,1),model_w=w0,model_m=m0,resSR=resBH)\nmsy_di = opt_msy(msy_di_opt$minimum,model_w=w0,model_m=m0,resSR=resBH,out=FALSE)\n\nmsy_data = full_join(\n msy_dd %>% mutate(Density=\"Dependent\"),\n msy_di %>% mutate(Density=\"Independent\")\n)\n\nwrite.csv(msy_data,file=\"msy_data.csv\")\n\nplot(Yield~SSB,data=dd_f,type=\"l\")\nplot(Yield/SSB~SSB,data=dd_f,type=\"l\",log=\"\")\n\nplot(Yield~F,data=dd_f,type=\"l\")\ndd_f$Yield\nplot(catch/SSB~SSB,data=dat,type=\"p\",log=\"\")\n\n# c(tbl$R,tbl$SSB/tbl$SPR)\n\nf2 = c(exp(seq(log(0.001),log(10),length=901)))\n# x=0.64\n# f2 = c(seq(0.62,0.64,length=11))\n\ndi_f = f2 %>% map_dfr(., function(x) {\n \n # tmp_n = exp(seq(log(0.1),log(100),length=31))\n # tmp_obj = tmp_n %>% map_dbl(., function(i) obj_fun(n=i,F=x,\n # model_w=w1,model_m=m1,resSR=resBH,out=TRUE))\n # int = tmp_n[c(max(1,which.min(tmp_obj)-1),min(length(tmp_obj),which.min(tmp_obj)+1))]\n int = c(0.1,1000)\n opt = optimize(obj_fun, interval = int,F=x,\n model_w=w0,model_m=m0,resSR=resBH,out=TRUE)\n \n tbl = obj_fun(opt$minimum,F=x,model_w=w0,model_m=m0,resSR=resBH,out=FALSE)\n tbl\n})\n\ndi_f\n\nplot(Yield~SSB,data=di_f,type=\"l\")\nplot(Yield/SSB~SSB,data=di_f,type=\"l\",log=\"\")\n\nplot(Yield~F,data=dd_f,type=\"l\",log=\"\")\nplot(Yield~F,data=di_f,type=\"l\",log=\"\")\n\nplot(SSB~F,data=di_f,type=\"l\",log=\"\")\n\nplot(SSB~F,data=dd_f,type=\"l\",log=\"\")\n\nplot(catch/SSB~SSB,data=dat,type=\"p\",log=\"\")\n\nd_f = full_join(\n dd_f %>% mutate(Density=\"Dependent\"),\n di_f %>% mutate(Density=\"Independent\")) %>%\n mutate(YPS = Yield/SSB)\n\ndi_f\n\nwrite.csv(d_f, file=\"Sustainable_yield.csv\",row.names=FALSE)\n\n\ng_yc = ggplot(filter(d_f,SSB<=max(dd_f$SSB) & SSB>=min(dd_f$SSB)),aes(x=SSB,y=Yield))+\n geom_path(size=path_size,aes(colour=Density,linetype=Density))+\n # geom_point(data=dat,aes(x=SSB,y=catch))+\n theme_bw(base_size=base_size)+\n scale_color_brewer(palette=\"Set1\")+\n ylab(\"Sustainable yield (1000 ton)\")+\n xlab(\"Spawning stock biomass (1000 ton)\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1),\n legend.margin = margin(0, 0, 0, 0),\n legend.spacing.x = unit(0, \"pt\"),\n legend.spacing.y = unit(0, \"pt\"))\n\ng_yc\n\ng_surplus = ggplot(filter(d_f,SSB<=max(dd_f$SSB) & SSB>=min(dd_f$SSB)),aes(x=SSB,y=YPS))+\n geom_path(size=path_size,aes(colour=Density,linetype=Density))+\n # geom_point(data=dat,aes(x=SSB,y=catch))+\n theme_bw(base_size=base_size)+\n scale_color_brewer(palette=\"Set1\")+\n ylab(\"Sustainable yield relative to SSB\")+\n xlab(\"Spawning stock biomass (1000 ton)\")+\n theme(legend.position=c(0.99,0.99),legend.justification=c(1,1),\n legend.margin = margin(0, 0, 0, 0),\n legend.spacing.x = unit(0, \"pt\"),\n legend.spacing.y = unit(0, \"pt\"))\n\ng_surplus\n\nhead(d_f)\n\nd_YS = d_f %>% \n mutate(MSY = if_else(Density==\"Dependent\",msy_dd$Yield,msy_di$Yield),\n SSB0 = if_else(Density==\"Dependent\",max(dd_f$SSB),max(di_f$SSB))\n ) %>%\n mutate(\"SY/MSY\" = Yield/MSY,\"SSB/SSB0\"=SSB/SSB0) %>%\n pivot_longer(cols=c(\"SY/MSY\",\"SSB/SSB0\"),names_to=\"Reference\",values_to=\"Value\") %>%\n mutate(Reference = factor(Reference,levels=c(\"SY/MSY\",\"SSB/SSB0\")))\n\n# d_YS$`SY/MSY` %>% range\n# d_YS$`SSB/SSB0` %>% range\n\ng_F = ggplot(filter(d_YS,F<1.5),aes(x=F,y=Value))+\n geom_path(size=path_size,aes(colour=Density,linetype=Reference))+\n # geom_point(data=dat,aes(x=SSB,y=catch))+\n theme_bw(base_size=base_size)+\n scale_color_brewer(palette=\"Set1\")+\n scale_linetype_discrete(name=\"Type\")+\n ylab(\"Relative value\")+\n xlab(\"Fishing mortality coefficient\")+\n theme(legend.position=c(0.99,0.99),legend.justification=c(1,1),\n legend.margin = margin(0, 0, 0, 0),\n legend.spacing.x = unit(0, \"pt\"),\n legend.spacing.y = unit(0, \"pt\"),\n legend.key.size=unit(0.4,\"cm\"))+\n guides(colour=guide_legend(nrow=1,order=1,title.position=\"top\"),\n linetype=guide_legend(nrow=1,order=2,title.position=\"top\"))+\n ylim(0,1.25)\n\ng_F\n\n\n\ng1_BH = ggplot(data=NULL,aes(x=SSB,y=R))+\n geom_path(data=resBH$pred,size=path_size,colour=\"black\",linetype=\"longdash\")+\n geom_point(data=as.data.frame(SRdata))+\n geom_path(data=filter(pred_dat2,SSB%\n mutate(Category = \"Annual value\")\n\n\ndat3$Stat %>% unique()\ndat3$Type %>% unique()\n\nmsy_longer = msy_data %>% filter(Density==\"Dependent\") %>%\n pivot_longer(names_to = \"stat\", values_to=\"Value\",cols=-Density) %>%\n mutate(Stat=case_when(stat==\"N\" |stat==\"R\" ~ \"Fish number\",\n stat==\"B\" |stat==\"SSB\" | stat==\"Yield\" ~ \"Biomass\",\n TRUE ~ stat)) %>%\n mutate(Type=case_when(stat==\"R\" ~ \"Age 0\",\n stat==\"N\" ~ \"Older\",\n stat==\"B\" ~ \"Total\",\n stat==\"SSB\" ~ \"Spawner\",\n stat==\"Yield\" ~ \"Catch\",\n TRUE ~ stat)) %>%\n mutate(Category = \"MSY level\")\n\ndat3 = full_join(dat3,msy_longer)\n\n\ng1 = ggplot(data=filter(dat2,Stat==\"Fish number\"),aes(x=Year,y=Value))+\n geom_path(aes(colour=Type),size=path_size)+\n theme_bw(base_size=base_size)+\n scale_colour_brewer(palette=\"Set1\",name=\"\",guide=guide_legend(nrow=1))+\n # scale_linetype_discrete(name=\"\",guide=guide_legend(nrow=1))+\n ylim(0,NA)+labs(colour=NULL,linetype=NULL)+\n ylab(\"Number (billion)\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1),\n legend.margin = margin(0, 0, 0, 0),\n legend.spacing.x = unit(0, \"pt\"),\n legend.spacing.y = unit(-2, \"pt\"))+\n geom_hline(data=filter(msy_longer,Stat==\"Fish number\"),aes(yintercept=Value,colour=Type),\n size=0.5,linetype=\"dotted\")\ng1\n\ng2 = ggplot(data=filter(dat2,Stat==\"Biomass\") %>% mutate(Type=factor(Type,levels=c(\"Total\",\"Spawner\",\"Catch\"))),aes(x=Year,y=Value))+\n geom_path(aes(colour=Type),size=path_size)+\n theme_bw(base_size=base_size)+\n scale_colour_brewer(palette=\"Set1\",name=\"\",guide=guide_legend(nrow=1))+\n # scale_linetype_discrete(name=\"\",guide=guide_legend(nrow=1))+\n ylim(0,NA)+labs(colour=NULL,linetype=NULL)+\n ylab(\"Biomass (1000 ton)\")+\n theme(legend.position=c(0.01,0.99),legend.justification=c(0,1),\n legend.margin = margin(0, 0, 0, 0),\n legend.spacing.x = unit(0, \"pt\"),\n legend.spacing.y = unit(-2, \"pt\"))+\n geom_hline(data=filter(msy_longer,Stat==\"Biomass\"),aes(yintercept=Value,colour=Type),\n size=0.5,linetype=\"dotted\")\n\ng2\n\ng3 = ggplot(data=filter(dat2,Stat==\"Weight\"),aes(x=Year,y=Value))+\n # geom_path(aes(colour=Type,linetype=Type),size=path_size)+\n geom_path(size=path_size,aes(colour=Type))+\n theme_bw(base_size=base_size)+\n scale_colour_brewer(palette=\"Set1\",name=\"\")+\n # scale_linetype_discrete(name=\"\")+\n ylim(0,NA)+\n ylab(\"Body weight (g)\")+\n theme(legend.position=\"\",legend.justification=c(0,1))+\n geom_hline(data=filter(msy_longer,Stat==\"Weight\"),aes(yintercept=Value,colour=Type),\n size=0.5,linetype=\"dotted\")\n\ng3\n\ng4 = ggplot(data=filter(dat2,Stat==\"Maturity\"),aes(x=Year,y=Value))+\n geom_path(aes(colour=Type),size=path_size)+\n # geom_path(size=path_size)+\n theme_bw(base_size=base_size)+\n scale_colour_brewer(palette=\"Set1\",name=\"\")+\n # scale_linetype_discrete(name=\"\")+\n ylim(0,NA)+\n ylab(\"Maturation rate\")+\n theme(legend.position=\"\",legend.justification=c(0,1))+\n geom_hline(data=filter(msy_longer,Stat==\"Maturity\"),aes(yintercept=Value,colour=Type),\n size=0.5,linetype=\"dotted\")\n\ng4\n\ng5 = ggplot(data=filter(dat2,Stat==\"F\"),aes(x=Year,y=Value))+\n geom_path(aes(colour=Type),size=path_size)+\n # geom_path(size=path_size)+\n theme_bw(base_size=base_size)+\n scale_colour_brewer(palette=\"Set1\",name=\"\")+\n # scale_linetype_discrete(name=\"\")+\n ylim(0,NA)+\n ylab(\"F\")+\n theme(legend.position=\"\",legend.justification=c(0,1))+\n geom_hline(data=filter(msy_longer,Stat==\"F\"),aes(yintercept=Value,colour=Type),\n size=0.5,linetype=\"dotted\")\n\ng5\n\ng_time = gridExtra::grid.arrange(g1+ggtitle(\"(a)\"),\n g2+ggtitle(\"(b)\"),\n g3+ggtitle(\"(c)\"),\n g4+ggtitle(\"(d)\"),\n g5+ggtitle(\"(e)\"),nrow=2)\n\nggsave(g_time,filename=\"graph_timeseries_MSY.png\",dpi=600,\n unit=\"mm\",height=120,width=240)\n\n\n### FishLife ----\n\nlibrary(FishLife)\n\n( Predictions = Plot_taxa(Search_species(Genus=\"Scomber\",Species=\"japonicus\")$match_taxonomy) )\nPredictions[[1]]$Mean_pred[\"ln_MASPS\"] %>% exp\n\n# あんまり合わない\n\n\n\n\n\n# pred_dat = data.frame(N = seq(exp(min(dat$logN)-1),exp(max(dat$logN)+1),length=1000)) %>%\n# mutate(logN = log(N)) %>%\n# mutate(Weight=exp(predict(m3,newdata=.)),\n# Maturity=inv_logit(predict(l3,newdata=.))) %>%\n# mutate(B = Weight*N) %>%\n# mutate(SSB = Maturity*B)\n\nplot(Weight~N,data=pred_dat)\nplot(Maturity~N,data=pred_dat)\n\nplot(Weight~SSB,data=pred_dat)\nplot(Maturity~SSB,data=pred_dat)\n\nplot(pred_dat$SSB~pred_dat$N)\n\nplot(dat$SSB~dat$N)\n\n\nrange(dat$N)\nrange(dat$SSB)\nas.numeric(logLik(m1))\n\nhead(dat)\n\nplot(dat)\n\nsummary(glm(logW ~ N,data=dat))\nsummary(glm(logW ~ log(N),data=dat))\n\nmodel1 = glm(logW ~ log(N),data=dat)\n\nplot(model1)\n\nsummary(glm(logitM ~ N,data=dat))\nsummary(glm(logitM ~ log(N),data=dat))\n\nmodel2 = glm(logitM ~ log(N),data=dat)\nplot(model2)\n\nsummary(glm(logitM ~ B,data=dat))\nsummary(glm(logitM ~ log(B),data=dat))\n\nlogit = function(p) log(p/(1-p))\ninv_logit = function(y) 1/(1+exp(-y))\n\nmu = -1\nsigma = 1\nobj = function(p,mu,sigma) {\n tmp = integrate(function(x) inv_logit(x)*dnorm(x,mean=logit(p),sd=sigma),lower=-Inf,upper=+Inf)\n (tmp$value - inv_logit(mu))^2\n}\n\n\nopt = optimize(obj,c(0.000001,0.999999),mu=mu,sigma=sigma)\n\n\nN = 10000\ny = rnorm(N,-1,sd=1)\ninv_logit(-1)\np = inv_logit(y)\n\nlibrary(nlme)\n\nmodelW0 = glm(log(Weight)~log(N),data=dat)\nmodelW1 = glm(log(Weight)~N,data=dat)\nAICc(modelW0,modelW1)\n\n\nmodelMat0 = glm(logit(Maturity)~log(N),data=dat)\nsummary(modelMat0)\n\nmodelMat1 = glm(logit(Maturity)~N,data=dat)\n\n# AICc(modelMat0,modelMat1)\n\nplot(modelMat0$residuals)\nstats::ar(modelMat0$residuals,order.max = 1)\n\narima_res = arima(dat[,\"logW\"],order=c(0,0,0),\n method=\"ML\",xreg=dat[,\"logN\"])\narima_res$coef\nAICc(arima_res)\n\narima_res = arima(dat[,\"logW\"],order=c(1,0,0),\n method=\"ML\",xreg=dat[,\"logN\"])\narima_res$coef\nAICc(arima_res)\n\n\n\nAICc(m)\ndat$t\n\n# install.packages(\"brms\")\nlibrary(brms)\nd <- simCor1(phi=0.8,sdgrp=2,sdres=1,seed=101)\n\n\n\nmodelW0_N = glm(log(Weight)~N,data=dat)\nmodelW_RE_N = lme(log(Weight)~N,random=~1|YEAR,data=dat)\nmodelW_AR_N = lme(log(Weight)~N,random=~1|YEAR,data=dat,correlation=corAR1())\n\n\nsummary(modelW_AR)\nAIC(modelW0,modelW_RE,modelW_AR,modelW0_N,modelW_RE_N,modelW_AR_N)\n\nlogLik(modelW0)\nlogLik(modelW_AR)\n\nplot(modelW0$residuals,type=\"l\")\n\nsummary(modelW_AR)\n\nmodelW0\n\nlibrary(MuMIn)\n\nmodelW_RE2 = glmmTMB(log(Weight)~log(N)+(1|YEAR),data=dat)\nsummary(modelW_RE2)\n\nAICtab(modelW_RE2)\nAICc(modelW_RE2)\n\nmodelW = lme(log(Weight)~log(N),random=~1|YEAR,data=dat,correlation=corAR1())\nsummary(modelW)\nAIC(modelW)\n\n\n?simCor1\n?glmmTMB\n\nlog(1)\n\n\nsummary(p)\n", "meta": {"hexsha": "cd93696d62b8e002b965336b1ac78acfd2960168", "size": 43613, "ext": "r", "lang": "R", "max_stars_repo_path": "masaba_dd.r", "max_stars_repo_name": "ShotaNishijima/masaba_dd", "max_stars_repo_head_hexsha": "bbb4618e4b0054a146bfe67ffff1c28727dbacef", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "masaba_dd.r", "max_issues_repo_name": "ShotaNishijima/masaba_dd", "max_issues_repo_head_hexsha": "bbb4618e4b0054a146bfe67ffff1c28727dbacef", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2021-12-13T04:35:58.000Z", "max_issues_repo_issues_event_max_datetime": "2021-12-13T05:16:09.000Z", "max_forks_repo_path": "masaba_dd.r", "max_forks_repo_name": "ShotaNishijima/masaba_dd", "max_forks_repo_head_hexsha": "bbb4618e4b0054a146bfe67ffff1c28727dbacef", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 30.2029085873, "max_line_length": 190, "alphanum_fraction": 0.6526265104, "num_tokens": 14994, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8080672135527632, "lm_q2_score": 0.6477982315512488, "lm_q1q2_score": 0.5234645119140253}} {"text": "# basic plotting functions for assessing output\n\nrequire('fields')\n\ntemp.colors=function(n=25){\n m <- floor(n/2)\n blues <- hsv(h=.65, s=seq(1,0,length=m+1)[1:m])\n reds <- hsv(h=0, s=seq(1,0,length=m+1)[1:m])\n c(blues,if(n%%2!=0) \"#FFFFFF\", reds[m:1])\n }\n\n\nf.clr=function(vec){\n return(log(vec)-mean(log(vec)))\n}\n\n# rIts: space, taxon, time, iteration?\n# rMean: space, taxon, time (mean across iterations? )\n# p: taxon number?\nspatialEOFbyTaxon=function(p,rMean,rIts,transform='asin'){\n # this returns EOFs for a specific taxon where the patterns are with respect to the locations (so true eofs) and there is a score for each time for each of the nT patterns (where nT is the number of time points)\n # since we work with a single taxon, the clr transformation doesn't make sense as the transformation is based on having all P taxa, and if one applied it and then took only the transformed data for the taxon of interet, one will have different data values even if the abundance of beech, say, were 24% in two different time periods because the abundunace of the other taxa changed between the two times\n # scoreMean is the mean of the scores on the patterns (eofs) for each time point, averaging over the scores for each iteration; each row is one pattern and each column a time point\n # scoreUpper is the 97.5th percentile and scoreLower the 2.5th percentile of the scores for a given time and pattern; each row is one pattern and each column a time point\n # scoreIts is the entire set of scores by iteration: patterns X times X iterations\n # scoresOnMeans are the scores calculated on rMean rather than by iteration - note that the EOFs are computed based on rMean so that the patterns are the same for every iteration; otherwise the meaning of the scores would change between iterations; each row is one pattern and each column a time point\n # loadings are the weights on the locations for each pattern - each column is a pattern and each row is a grid cell\n # eigenvals are the variances associated with each pattern with higher variance meaning more variability explained by that pattern\n nT=dim(rIts)[3]\n S=dim(rIts)[1]\n nIts=dim(rIts)[4]\n\n calcScores=function(data){\n return((lm(data~mysvd$u-1))$coef)\n }\n\n \n data=asin(sqrt(rMean[,p,]))\n\n locMeans=rowMeans(data)\n data=data-locMeans \n\n mysvd=svd(data)\n\n workItsMat=matrix(c(asin(sqrt(rIts[,p,,]))),S,nT*nIts)-locMeans\n\n tmp=lm(workItsMat~mysvd$u-1)$coef\n scoreIts=array(tmp,c(nT,nT,nIts)) # there are only 31 patterns because only 31 time points, so this is the score on each pattern for each time point for each iteration\n scoreMean=apply(scoreIts,c(1,2),mean) # mean score on each pattern for each time point\n scoreUpper=apply(scoreIts,c(1,2),quantile,.975) # quantiles of scores on each pattern for each time point\n scoreLower=apply(scoreIts,c(1,2),quantile,.025)\n \n scoresOnMeans=diag(mysvd$d)%*%t(mysvd$v)\n \n return(list(scoreMean=scoreMean,scoreUpper=scoreUpper,scoreLower=scoreLower,scoreIts=scoreIts,loadings=mysvd$u,eigenvals=mysvd$d,scoresOnMeans=scoresOnMeans))\n}\n\nspatialEOFallTaxa=function(rMean,rIts,transform='clr'){ # could also do transform='asin' or transform='none'\n # this returns spatial EOFs where they are calculated based on all taxa\n # the patterns are with respect to the locations (so true eofs) and there is a score for taxon for each time for each of the S=256 patterns\n # one can either invoke the clr transformation that deals with having proportions or one could invoke 'asin' for the arcsin sqrt transformation. I assume that the EOFs based on the covariance are desired, but to change this, just change the line below that does the svd (mysvd=svd(....\n # scoreMean is the mean of the scores on the patterns (eofs) for each taxon and time point, averaging over the scores for each iteration; each row is one pattern and each column a time point\n # scoreUpper is the 97.5th percentile and scoreLower the 2.5th percentile of the scores for a given taxon and time; each row is one pattern\n # scoreIts is the entire set of scores by iteration: patterns X taxa X times X iterations\n # scoresOnMeans are the scores calculated on rMean rather than by iteration - note that the EOFs are computed based on rMean so that the patterns are the same for every iteration; otherwise the meaning of the scores would change between iterations; each row is one pattern\n # loadings are the weights on the locations for each pattern - each column is a pattern and each row is a grid cell\n # eigenvals are the variances associated with each pattern with higher variance meaning more variability explained by that pattern\n nT = dim(rIts)[3]\n S = dim(rIts)[1]\n P = dim(rIts)[2]\n nIts = dim(rIts)[4]\n\n calcScores=function(data){\n return((lm(data~mysvd$u-1))$coef)\n }\n \n print('Got dimensions!')\n \n if(transform=='asin'){\n data=asin(sqrt(rMean))\n workItsMat=matrix(c(asin(sqrt(rIts))),S,P*nT*nIts)\n }\n \n if(transform=='clr'){\n tmp=apply(rMean,c(1,3),f.clr)\n data=array(NA,c(S,P,nT))\n for(p in 1:P){\n data[,p,]=tmp[p,,]\n }\n workIts=apply(rIts,c(1,3,4),f.clr)\n tmp=array(NA,c(S,P,nT,nIts))\n for(p in 1:P){\n tmp[,p,,]=workIts[p,,,]\n }\n workItsMat=matrix(c(tmp),S,P*nT*nIts)\n }\n \n print('Transformed!')\n \n tmp=data\n data=matrix(c(data),S,P*nT)\n\n locMeans=rowMeans(data)\n data=data-locMeans \n\n mysvd=svd(data)\n\n workItsMat=workItsMat-locMeans\n\n tmp = lm(workItsMat~mysvd$u-1)$coef\n scoreIts = array(tmp,c(S,P,nT,nIts)) # there are 256 patterns, and this is the score on each pattern for each taxon and each time point for each iteration\n scoreMean = apply(scoreIts,c(1,2,3),mean) # mean score on each pattern for each taxon and time point\n scoreUpper = apply(scoreIts,c(1,2,3),quantile,.975) # quantiles of scores on each pattern for each taxon and time point\n scoreLower = apply(scoreIts,c(1,2,3),quantile,.025)\n \n scoresOnMeans=array(diag(mysvd$d)%*%t(mysvd$v),c(S,P,nT))\n \n return(list(scoreMean=scoreMean,scoreUpper=scoreUpper,scoreLower=scoreLower,scoreIts=scoreIts,loadings=mysvd$u,eigenvals=mysvd$d,scoresOnMeans=scoresOnMeans))\n}\n\n\ntaxaEOFs=function(rMean,rIts){\n # this returns EOFs where the patterns are with respect to the taxa and there is a score for each grid cell at each time for each of the 9 patterns (the 10th is meaningless because of sum to one constraint\n # scoreMean is the mean of the scores on the patterns (eofs) for each grid cell by time, averaging over the scores for each iteration\n # scoreUpper is the 97.5th percentile and scoreLower the 2.5th percentile of the scores for a given cell and time and pattern\n # scoreIts is the entire set of scores by iteration\n # scoresOnMeans are the scores calculated on rMean rather than by iteration - note that the EOFs are computed based on rMean so that the patterns are the same for every iteration; otherwise the meaning of the scores would change between iterations\n # loadings are the weights on the taxa for each pattern - each column is a pattern and each row is a taxon\n # eigenvals are the variances associated with each pattern with higher variance meaning more variability explained by that pattern\n nT=dim(rIts)[3]\n S=dim(rIts)[1]\n nIts=dim(rIts)[4]\n P=dim(rIts)[2]\n mymat=matrix(0,nr=S*nT,nc=P)\n for(p in 1:P){\n mymat[,p]=c(rMean[,p,])\n }\n \n# f.clr=function(vec){\n# return(log(vec)-mean(log(vec)))\n# }\n \n\n data=apply(mymat,1,f.clr) # transformation to account for proportional data\n taxaMeans=rowMeans(data)\n data=data-taxaMeans # this puts rMean output into a P=10 by n= S*nT matrix; n is the number of 'observations' or 'replicates'\n\n mysvd=svd(data)\n\n calcScores=function(data){\n return((lm(data~mysvd$u-1))$coef)\n }\n\n workIts=array(NA,c(P,S,nT,nIts))\n for(p in 1:P){\n workIts[p,,,]=rIts[,p,,]\n }\n workItsMat=matrix(c(workIts),P,S*nT*nIts)\n workItsMat=apply(workItsMat,2,f.clr)-taxaMeans\n\n tmp=lm(workItsMat~mysvd$u-1)$coef\n\n scoreIts=array(tmp,c(P,S,nT,nIts))\n scoreMean=apply(scoreIts,c(1,2,3),mean)\n scoreUpper=apply(scoreIts,c(1,2,3),quantile,.975)\n scoreLower=apply(scoreIts,c(1,2,3),quantile,.025)\n \n scoresOnMeans=array(diag(mysvd$d)%*%t(mysvd$v),c(P,S,nT)) # scores on the rMean values at each location and time\n\n tmp1=tmp2=tmp3=tmp4=array(NA,c(S,P,nT))\n tmp5=array(NA,c(S,P,nT,nIts))\n for(p in 1:P){\n tmp1[,p,]=scoreMean[p,,]\n tmp2[,p,]=scoreLower[p,,]\n tmp3[,p,]=scoreUpper[p,,]\n tmp4[,p,]=scoresOnMeans[p,,]\n tmp5[,p,,]=scoreIts[p,,,]\n }\n \n return(list(scoreMean=tmp1,scoreUpper=tmp3,scoreLower=tmp2,scoreIts=tmp5,loadings=mysvd$u,eigenvals=mysvd$d,scoresOnMeans=tmp4))\n}\n\n\naddPondLocs=function(tNew,tOld=NULL){\n# plots black circles for ponds with pollen data and green open circles for ponds without\n# use tOld argument if you want to plot information for two time points\n points(pondLocs[pondsUsed[[tNew]],],pch=16,cex=.8)\n points(pondLocs[pondsNotUsed[[tNew]],],col='green')\n if(!is.null(tOld)){\n points(pondLocs[pondsUsed[[tOld]],],pch=16,cex=.8)\n points(pondLocs[pondsNotUsed[[tOld]],],col='green')\n }\n}\n\n\ntaxaSignif=function(p1,p2,rIts,locs,restrict=FALSE,inputRestricted=FALSE,ylab='',xlab=''){\n# compares significance of differences in composition between two predictions\n# red colors indicate more of the taxon in the older time period; blue in the newer\n nComparisons=length(rIts[1,1,])\n less=rIts[,p1,]rItsOld[,p1,] & rItsNew[,p2,]>rItsOld[,p2,]\n bothMore=matrix(apply(bothMore,1,sum),sqrt(S),sqrt(S))\n firstLess=rItsNew[,p1,]rItsOld[,p2,]\n firstLess=matrix(apply(firstLess,1,sum),sqrt(S),sqrt(S))\n secondLess=rItsNew[,p1,]>rItsOld[,p1,] & rItsNew[,p2,]=lowCut | moreProp>=lowCut\n highSignif=lessProp>=highCut | moreProp>=highCut\n\n diff=meanNew-meanOld\n diff[!lowSignif]=0\n diff=matrix(thresh(diff,maxDiff,-maxDiff),sqrt(S),sqrt(S))\n \n xs=sort(unique(locs[,1]))\n ys=sort(unique(locs[,2]))[rows]\n if(legend){\n image.plot(xs,ys,diff[,rows],zlim=c(-maxDiff,maxDiff),xlab=xlab,ylab=ylab,xaxt='n',yaxt='n',col=temp.colors(nColors*2+1))\n } else{\n image(xs,ys,diff[,rows],zlim=c(-maxDiff,maxDiff),xlab=xlab,ylab=ylab,xaxt='n',yaxt='n',col=temp.colors(nColors*2+1))\n }\n xincr=diff(range(xs))/(2*length(xs))\n yincr=diff(range(ys))/(2*length(ys))\n xrg=range(xs)+c(-1,1)*xincr\n yrg=range(ys)+c(-1,1)*yincr\n\n incr=(yrg[2]-xrg[1]-yrg[1]+xrg[2])/nIncr\n for(i in 1:nIncr){\n abline(a=yrg[1]-xrg[2]+incr*(i-1),b=1,col='white',lwd=2,lty=1)\n }\n box()\n diff[!highSignif]=NA\n if(legend){\n largeplot=image.plot.cp(xs,ys,diff[,rows],zlim=c(-maxDiff,maxDiff),xlab=xlab,ylab=ylab,xaxt='n',yaxt='n',col=temp.colors(nColors*2+1),add=T)\n par(plt=largeplot)\n } else{\n image(xs,ys,diff[,rows],zlim=c(-maxDiff,maxDiff),xlab=xlab,ylab=ylab,xaxt='n',yaxt='n',col=temp.colors(nColors*2+1),add=T)\n }\n lines(ma$x,ma$y,lwd=1)\n lines(ct$x,ct$y,lwd=1)\n lines(ri$x,ri$y,lwd=1)\n lines(ny$x,ny$y,lwd=1)\n lines(nh$x,nh$y,lwd=1)\n box()\n}\n\nimage.plot.cp=function (..., add = FALSE, nlevel = 64, legend.shrink = 0.9, \n legend.width = 0.05, graphics.reset = FALSE, horizontal = FALSE, \n offset = 2 * legend.width, bigplot = NULL, smallplot = NULL, \n legend.only = FALSE, col = tim.colors(nlevel)) \n{\n old.par <- par(no.readonly = TRUE)\n info <- image.plot.info(...)\n if (add) \n big.plot <- old.par$plt\n if (legend.only) \n graphics.reset <- TRUE\n temp <- image.plot.plt(add = add, legend.shrink = legend.shrink, \n legend.width = legend.width, horizontal = horizontal, \n offset = offset, bigplot = bigplot, smallplot = smallplot)\n smallplot <- temp$smallplot\n bigplot <- temp$bigplot\n if (!legend.only) {\n if (!add) {\n par(plt = bigplot)\n }\n image(..., add = add, col = col)\n big.par <- par(no.readonly = TRUE)\n }\n if ((smallplot[2] < smallplot[1]) | (smallplot[4] < smallplot[3])) {\n par(old.par)\n stop(\"plot region too small to add legend\\n\")\n }\n temp <- list(...)\n iy <- seq(info$zlim[1], info$zlim[2], , nlevel)\n iz <- matrix(iy, nrow = 1, ncol = length(iy))\n ix <- 1\n if (!horizontal) {\n par(new = TRUE, pty = \"m\", plt = smallplot, err = -1)\n image(ix, iy, iz, xaxt = \"n\", yaxt = \"n\", xlab = \"\", \n ylab = \"\", col = col)\n axis(4, mgp = c(3, 1, 0), las = 2)\n box()\n }\n else {\n par(new = TRUE, pty = \"m\", plt = smallplot, err = -1)\n image(iy, ix, t(iz), yaxt = \"n\", xlab = \"\", ylab = \"\", \n col = col)\n box()\n }\n mfg.save <- par()$mfg\n if (graphics.reset | add) {\n par(old.par)\n par(mfg = mfg.save, new = FALSE)\n return(bigplot)\n #invisible()\n }\n else {\n par(big.par)\n par(plt = big.par$plt, xpd = FALSE)\n par(mfg = mfg.save, new = FALSE)\n return(bigplot)\n #invisible()\n }\n}\n\n\ntimeSignif=function(p,rItsNew,rItsOld,locs,restrict=FALSE,inputRestricted=FALSE,ylab='',xlab=''){\n# compares significance of differences in composition between two predictions\n# red colors indicate more of the taxon in the newer time period; blue in the older\n nComparisons=length(rItsNew[1,1,])\n less=rItsNew[,p,]up]=up\n }\n if(!is.null(lo)){\n vec[vec0){\n minVal=min(proportions[i,])\n if(minVal==0){\n warning(\"Some proportions are zero; proceeding with jittering.\")\n eps=1e-10*max(proportions[i,])\n proportions[i,]=proportions[i,]+eps\n }\n pieAdd(as.vector(proportions[i,]),as.vector(centers[i,]),radius=radius[i],...)\n } else{\n points(centers[i,],pch='X')\n }\n }\n lines(ma$x,ma$y,lwd=1)\n lines(ct$x,ct$y,lwd=1)\n lines(ri$x,ri$y,lwd=1)\n lines(ny$x,ny$y,lwd=1)\n lines(nh$x,nh$y,lwd=1)\n}\n\n\npieAdd= function (x, center, labels = names(x), edges = 200, radius = 0.8, density = NULL, angle = 45, col = NULL, border = NULL, lty = NULL, ...) # modified from the pie() function in R\n{\n if (!is.numeric(x) || any(is.na(x) | x <= 0)) \n stop(\"pie: `x' values must be positive.\")\n if (is.null(labels)) \n labels <- rep(\"\",length(x))\n x <- c(0, cumsum(x)/sum(x))\n dx <- diff(x)\n\n pin <- par(\"pin\")\n nx <- length(dx)\n if (is.null(col)) \n col <- if (is.null(density)) \n c(\"white\", \"black\",\"lightblue\", \"red\",\"darkblue\",\"yellow\",\n \"purple\",\"orange\",\"lightgreen\",\"darkgreen\")\n else par(\"fg\")\n col <- rep(col, length.out = nx)\n border <- rep(border, length.out = nx)\n lty <- rep(lty, length.out = nx)\n angle <- rep(angle, length.out = nx)\n density <- rep(density, length.out = nx)\n for (i in 1:nx) {\n n <- max(2, floor(edges * dx[i]))\n t2p <- 2 * pi * seq(x[i], x[i + 1], length = n)\n xc <- c(cos(t2p), 0) * radius + center[1]\n yc <- c(sin(t2p), 0) * radius + center[2]\n polygon(xc, yc, density = density[i], angle = angle[i], \n border = border[i], col = col[i], lty = lty[i],...)\n t2p <- 2 * pi * mean(x[i + 0:1])\n xc <- cos(t2p) * radius + center[1]\n yc <- sin(t2p) * radius + center[2]\n if (!is.na(lab <- labels[i]) && lab != \"\") {\n lines(c(1, 1.05) * xc, c(1, 1.05) * yc)\n text(1.1 * xc, 1.1 * yc, lab, xpd = TRUE, adj = ifelse(xc < \n 0, 1, 0), ...)\n }\n }\n invisible(NULL)\n}\n\n\n# template code for using above functions to analyze output\n\nif(FALSE){\n\n# piemap of predictions for third time point (500 in 300-3000 predictions), restricting to approximate region with ponds\npieMap(rMean[,,3],gridLocs,restrict=TRUE)\n# piemap of aggregated pollen for third time point\npieMap(cMat[,,3],pondLocs)\n\n# surface maps for third time point\nsurfMap(1,rMean[,,3],gridLocs,max=0.6,restrict=TRUE)\n\npar(mfrow=c(3,3),mar=c(2.1,2.1,3.1,1.1)) \n for(p in 1:(P-1)){\n surfMap(p,rMean[,,3],gridLocs,max=0.6,restrict=TRUE)\n title(taxa[p])\n }\n\n# feature significance map\n\nfeature(2,less[[3]],gridLocs,restrict=TRUE,nComparisons=nComparisons) # 2nd taxon, 3rd time period\n# ignore the 'graphical parameter' warning messages\n\n\n\n# maps of significant differences between two time periods (the third and fourth time periods) for 2nd taxon\n\n# pure significance\ntimeSignif(2,rIts[,,3,],rIts[,,4,],gridLocs,restrict=TRUE)\n# actual mean differences with only significant ones shown\ntimeDiffs(p=1,rIts[,,3,],rIts[,,4,],gridLocs,restrict=T,nColors=4,maxDiff=.225)\n\n# putting all taxa on a page\n par(mfrow=c(3,3),mar=c(2.1,2.1,3.1,1.1)) \n for(p in 1:(P-1)){\n timeSignif(p,rIts[,,3,],rIts[,,4,],gridLocs,restrict=TRUE)\n title(taxa[p])\n }\n\n# map of significant sychrony between two taxa (pine and oak) between two time periods (the third and fourth time periods)\ntimeSignifSynch(8,7,rIts[,,3,],rIts[,,4,],gridLocs,restrict=TRUE)\n\n\n\n\n# time series plots of coefficients\nwhichts=1:23 # use 300-2500 time interval (if the output is for 300-3000 time period)\ntimePlotDecomp(times[whichts],b1sMean[whichts,],b1sLow[whichts,],b1sUp[whichts,],b1s[whichts,,])\n\n# maps of significant differences between two taxa (2nd and 7th here) for 3rd time period\ntaxaSignif(2,7,rIts[,,3,],gridLocs,restrict=TRUE)\n\n\n# vegetation + pollen diagam for 300-2500 time interval for cell 139 (Snake Pond I think)\ndat=read.table('PollenTimeSeries.csv',sep=',',header=T)\ni=20\nsubdat=dat[dat$sitenumber==i,]\nage=-subdat$'cal.age'\ntmp=subdat[,5:14]\ntmp=tmp/apply(tmp,1,sum)\nwhichTimeIndices=1:23\nvegPlusPollenDiagramDecomp(139,whichTimeIndices,seq(300,2500,by=100),age,tmp)\n\n# example calculation of taxa eofs and uncertainty\n\nmyTaxaEOFs=taxaEOFs(rMean,rItsSub)\n# some plotting ideas\n# map of scores on first EOF for 12th time point\nsurfMap(1,myTaxaEOFs$scoreMean[,,12],gridLocs,restrict=T)\n# time series plot of first EOF at 80th grid point for all time points, with uncertainty\ntsplot(myTaxaEOFs$scoreMean[80,1,],ylim=c(-12,12),xlab='time',ylab='scores')\nlines(myTaxaEOFs$scoreUpper[80,1,],lty=2)\nlines(myTaxaEOFs$scoreLower[80,1,],lty=2)\n\n# example of spatial EOFs specific to a taxon\n\nbeechEOF=spatialEOFbyTaxon(1,rMean,rItsSub)\n# plot the first eof\nsurfMap(1,beechEOF$loadings,gridLocs,max=0.2,min=-0.2,restrict=T) # here '1' is the pattern desired not the taxon\n# plot time series and uncertainty of scores for first eof\ntsplot(beechEOF$scoreMean[1,],ylim=c(-4,4),xlab='time',ylab='scores')\nlines(beechEOF$scoreUpper[1,],lty=2)\nlines(beechEOF$scoreLower[1,],lty=2)\n\n# spatial EOFs for all taxa\nallEOF=spatialEOFallTaxa(rMean,rItsSub,transform='clr')\n# plot the first eof\nsurfMap(1,allEOF$loadings,gridLocs,max=0.2,min=-0.2,restrict=T) # here '1' is the pattern desired not the taxon\n# plot time series and uncertainty of scores for first eof, second taxon (birch)\ntsplot(allEOF$scoreMean[1,2,],ylim=c(-60,60),xlab='time',ylab='scores')\nlines(allEOF$scoreUpper[1,2,],lty=2)\nlines(allEOF$scoreLower[1,2,],lty=2)\n# plot time series of posterior mean scores for first eof, all taxa\nfor(p in 1:P){\n if(p==1){\n tsplot(allEOF$scoreMean[1,p,],ylim=c(-40,40),xlab='time',ylab='scores')\n } else{\n lines(allEOF$scoreMean[1,p,],col=p)\n }\n}\n\n}\n", "meta": {"hexsha": "b356def7dac2c4af8b3701990551b6049903db3e", "size": 34531, "ext": "r", "lang": "R", "max_stars_repo_path": "r/utils/util.r", "max_stars_repo_name": "PalEON-Project/STEPPS-prediction-Ecology", "max_stars_repo_head_hexsha": "3d4115e79b2f5278c232441cfd83f613a2f6bbbd", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "r/utils/util.r", "max_issues_repo_name": "PalEON-Project/STEPPS-prediction-Ecology", "max_issues_repo_head_hexsha": "3d4115e79b2f5278c232441cfd83f613a2f6bbbd", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "r/utils/util.r", "max_forks_repo_name": "PalEON-Project/STEPPS-prediction-Ecology", "max_forks_repo_head_hexsha": "3d4115e79b2f5278c232441cfd83f613a2f6bbbd", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 37.2101293103, "max_line_length": 404, "alphanum_fraction": 0.6606237873, "num_tokens": 12019, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8080672227971211, "lm_q2_score": 0.647798211152541, "lm_q1q2_score": 0.5234645014189768}} {"text": "library(Diagnostics)\n\n\n\nlearn <- function(cases) {\n # order: P(Pn), P(Te |Pn), P(VTB), P(TB| VTB), P(Sm), P(LC |Sm), P(BR |Sm), P(XR |Pn, VTB, LC), P(Dy |LC, BR)\n pn = table(cases$Pn) + 1\n pn = pn / sum(pn)\n \n #special handling for temperature\n te__pn = generate_te_model(cases)\n \n vtb = table(cases$VTB) + 1\n vtb = vtb / sum(vtb)\n \n tb__vtb = table(cases$VTB, cases$TB) + 1\n for (i in 1:2) {\n tb__vtb[i, ] = tb__vtb[i, ] / sum(tb__vtb[i, ])\n }\n \n sm = table(cases$Sm) + 1\n sm = sm / sum(sm)\n \n lc__sm = table(cases$Sm, cases$LC) + 1\n for (i in 1:2) {\n lc__sm[i, ] = lc__sm[i, ] / sum(lc__sm[i, ])\n }\n \n br__sm = table(cases$Sm, cases$Br) + 1\n for (i in 1:2) {\n br__sm[i, ] = br__sm[i, ] / sum(br__sm[i, ])\n }\n \n xr__pn_tb_lc = table(cases$Pn, cases$TB, cases$LC, cases$XR) + 1\n for (i in 1:2) {\n for (j in 1:2) {\n for (k in 1:2) {\n xr__pn_tb_lc[i, j, k, ] = xr__pn_tb_lc[i, j, k, ] / sum(xr__pn_tb_lc[i, j, k, ])\n }\n }\n }\n \n dy__lc_br = table(cases$LC, cases$Br, cases$Dy) + 1\n for (i in 1:2) {\n for (j in 1:2) {\n dy__lc_br[i, j, ] = dy__lc_br[i, j, ] / sum(dy__lc_br[i, j, ])\n }\n }\n \n model = list()\n model$pn = pn\n model$te__pn = te__pn\n model$vtb = vtb\n model$tb__vtb = tb__vtb\n model$sm = sm\n model$lc__sm = lc__sm\n model$br__sm = br__sm\n model$xr__pn_tb_lc = xr__pn_tb_lc\n model$dy__lc_br = dy__lc_br\n \n return(model)\n}\n\ngenerate_te_model <- function(cases) {\n cases0 = list()\n cases1 = list()\n \n for (i in 1:length(cases$Te)) {\n if (cases$Pn[i] == 0) {\n cases0 = append(cases0, cases$Te[i])\n } else{\n cases1 = append(cases1, cases$Te[i])\n }\n }\n cases0 = as.vector(unlist(cases0))\n cases1 = as.vector(unlist(cases1))\n \n l0 = list()\n l0$mean = mean(cases0)\n l0$sd = sd(cases0)\n l1 = list()\n l1$mean = mean(cases1)\n l1$sd = sd(cases1)\n \n return(list(l0, l1)) # access with var_name[[1]]$sd, var_name[[2]]$sd, etc\n}\n\ndiagnose <- function(model, cases) {\n samples_number = 1500\n burn_period = 0.1\n \n for (i in 1:length(cases[,1])){\n case = cases[i,]\n samples = generate_samples(cases[i,], samples_number, burn_period, model)\n case = make_predictions(case, samples)\n cases[i,] = case\n }\n return((c(cases$Pn, cases$TB, cases$LC, cases$Br)))\n}\n\nevaluate_var_prob <- function(samples, var_name) {\n samples_count = length(samples[,1])\n var_column = samples[var_name]\n var_true_count = sum(var_column)\n true_prob = var_true_count / samples_count\n return(true_prob)\n}\n\nmake_predictions <- function(case, samples) {\n \n case[\"Pn\"] = evaluate_var_prob(samples, \"Pn\")\n case[\"TB\"] = evaluate_var_prob(samples, \"TB\")\n case[\"LC\"] = evaluate_var_prob(samples, \"LC\")\n case[\"Br\"] = evaluate_var_prob(samples, \"Br\")\n \n return(case)\n}\n\ninvert <- function(val) {\n if (val == 0) {\n return(1)\n } else {\n return(0)\n }\n}\n\ncalc_conditional_probs <- function(sample, model) {\n Pn = sample$Pn\n Te = sample$Te\n VTB = sample$VTB\n TB = sample$TB\n Sm = sample$Sm\n LC = sample$LC\n Br = sample$Br\n XR = sample$XR\n Dy = sample$Dy\n return(\n model$pn[Pn + 1] *\n dnorm(\n mean = model$te__pn[[Pn + 1]]$mean,\n sd = model$te__pn[[Pn + 1]]$sd,\n x = Te\n ) *\n model$vtb[VTB + 1] *\n model$tb__vtb[VTB + 1, TB + 1] *\n model$sm[Sm + 1] *\n model$lc__sm[Sm + 1, LC + 1] *\n model$br__sm[Sm + 1, Br + 1] *\n model$xr__pn_tb_lc[Pn + 1, TB + 1, LC + 1, XR + 1] *\n model$dy__lc_br[LC + 1, Br + 1, Dy + 1]\n )\n}\n\npropose_value_for_unknown <- function(sample, unknown_var, model) {\n old_sample = sample\n new_sample = sample\n new_sample[unknown_var] = invert(new_sample[unknown_var])\n p_old = calc_conditional_probs(old_sample, model)\n p_new = calc_conditional_probs(new_sample, model)\n \n if (p_new > p_old) {\n return(new_sample)\n } else {\n threshold = p_new / p_old\n random = runif(1, 0, 1)\n if (random < threshold) {\n return(new_sample)\n } else {\n return(old_sample)\n }\n }\n \n}\n\ngenerate_one_sample <- function(samples, number, model) {\n sample = samples[number,]\n if (number == 1) { #first sample, assign unknown variables to random values\n sample$Pn = sample(0:1, 1)\n sample$TB = sample(0:1, 1)\n sample$LC = sample(0:1, 1)\n sample$Br = sample(0:1, 1)\n } else {\n prev_sample = samples[number - 1,]\n sample$Pn = prev_sample$Pn\n sample$TB = prev_sample$TB\n sample$LC = prev_sample$LC\n sample$Br = prev_sample$Br\n }\n \n sample = propose_value_for_unknown(sample, \"Pn\", model)\n sample = propose_value_for_unknown(sample, \"TB\", model)\n sample = propose_value_for_unknown(sample, \"LC\", model)\n sample = propose_value_for_unknown(sample, \"Br\", model)\n \n return(sample)\n}\n\ngenerate_samples <- function(case, samples_number, burn_period, model) {\n \n samples = hist\n samples = samples[1:samples_number,] # samples_number cannot be more than hist size, maybe fix it later\n samples$Te = case$Te\n samples$VTB = case$VTB\n samples$Sm = case$Sm\n samples$XR = case$XR\n samples$Dy = case$Dy\n \n for (i in 1:samples_number) {\n samples[i,] = generate_one_sample(samples, i, model)\n }\n \n samples_burned = samples_number * burn_period\n samples = samples[samples_burned:length(samples[,1]),]\n \n return(samples)\n}\n#run_count = 10\n#runs = vector(length = run_count)\nrunDiagnostics(learn, diagnose, verbose = 2)\n#for (i in 1:run_count) {\n# runs[i] = runDiagnostics(learn, diagnose, verbose = 2)\n#}\n# print(\"Result: \")\n# cat(\"mean: \", mean(runs), \"\\n\")\n# cat(\"sd: \", sd(runs), \"\\n\")\n# cat(\"worse than 0.01345: \", length(runs[runs > 0.01345]), \"\\n\")\n# cat(\"worse than 0.013375: \", length(runs[runs > 0.013375]), \"\\n\")", "meta": {"hexsha": "b7c5ee8cf0eb76e586f9797d39318440f19c5878", "size": 5679, "ext": "r", "lang": "R", "max_stars_repo_path": "assignment_3.r", "max_stars_repo_name": "13hannes11/UU_ai_course", "max_stars_repo_head_hexsha": "701d9282374c472526d7cb54ef55ff182f93e96b", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "assignment_3.r", "max_issues_repo_name": "13hannes11/UU_ai_course", "max_issues_repo_head_hexsha": "701d9282374c472526d7cb54ef55ff182f93e96b", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "assignment_3.r", "max_forks_repo_name": "13hannes11/UU_ai_course", "max_forks_repo_head_hexsha": "701d9282374c472526d7cb54ef55ff182f93e96b", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 25.0176211454, "max_line_length": 111, "alphanum_fraction": 0.6108469801, "num_tokens": 2054, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7905303186696747, "lm_q2_score": 0.6619228825191871, "lm_q1q2_score": 0.5232701072526427}} {"text": "#' Plot annual SPR targets\r\n#' \r\n#' Computes and plots SPR targets using selectivity, M, WAA, and maturity by year.\r\n#' @param asap name of the variable that read in the asap.rdat file\r\n#' @param pspr user defined value(s) of spr, e.g. 0.4 or c(0.4, 0.43, 0.46); if empty (default), calculated at seq(0.2, 0.5, by=0.1) \r\n#' @param save.plots save individual plots\r\n#' @param od output directory for plots and csv files \r\n#' @param plotf type of plot to save\r\n#' @export\r\n\r\nPlotAnnualSPRtargets <- function(asap, pspr=c(), save.plots,od,plotf){\r\n \r\n if(length(pspr)>0) spr.targ.values <- pspr\r\n if(length(pspr)==0) spr.targ.values <- seq(0.2, 0.5, by=0.1)\r\n n.spr <- length(spr.targ.values)\r\n nages<- asap$parms$nages\r\n nyears <- asap$parms$nyears\r\n years <- seq(asap$parms$styr,asap$parms$endyr) \r\n fec.age <- asap$WAA.mats$WAA.ssb\r\n mat.age <- asap$maturity\r\n wgt.age <- asap$WAA.mats$WAA.catch.all\r\n M.age <- asap$M.age\r\n sel.age <- asap$F.age/apply(asap$F.age,1,max)\r\n \r\n spawn.time <- asap$options$frac.yr.spawn\r\n \r\n spr0.vals<- asap$SR.annual.parms$s.per.r.vec\r\n F.start <-0.11 # starting guess for optimization routine to find F_SPR%\r\n \r\n f.spr.vals <- matrix(NA, nyears, n.spr)\r\n ypr.spr.vals <- matrix(NA, nyears, n.spr)\r\n conv.vals <- matrix(NA, nyears, n.spr)\r\n \r\n for (i in 1:n.spr) {\r\n for (j in 1:nyears) {\r\n t.spr <- spr.targ.values[i]\r\n \r\n spr.f <- function(F.start) {\r\n abs(s.per.recr(nages=nages, fec.age=fec.age[j,], mat.age=mat.age[j,], M.age= M.age[j,], F.mult=F.start, sel.age=sel.age[j,], spawn.time=spawn.time)/spr0.vals[j] - t.spr )\r\n }\r\n yyy <- nlminb(start=F.start, objective=spr.f, lower=0, upper=3)\r\n f.spr.vals[j,i] <- yyy$par\r\n ypr.spr.vals[j,i] <- ypr(nages, wgt.age=wgt.age[j,], M.age=M.age[j,], F.mult=f.spr.vals[j,i], sel.age=sel.age[j,] )\r\n conv.vals[j,i] <- ifelse(yyy$convergence==0, TRUE, FALSE)\r\n } # end j-loop over nyears\r\n } #end i-loop over SPR values\r\n \r\n # if(length(pspr)==0) {\r\n par(mfrow=c(1,1), mar=c(4,4,2,4) )\r\n lty.seq=c(1,2,4,6)\r\n plot(years, f.spr.vals[,1], type='n', xlab=\"Year\", ylab=\"Full F (%SPR)\", lwd=2,\r\n col=\"blue3\", ylim=c(0,1.2*max(f.spr.vals)) )\r\n for (i in 1:n.spr) {\r\n lines(years, f.spr.vals[,i], lwd=2, col=i, lty=lty.seq[i] ) \r\n }\r\n if(length(pspr)==0) legend('top', legend=c(\"F20%\", \"F30%\", \"F40%\", \"F50%\"), col=seq(1,4), lty=lty.seq, \r\n horiz=T,lwd=rep(2,4), cex=0.9)\r\n if(length(pspr)>0) legend('top', legend=c(paste0(\"F\", round(100*pspr,0), \"%\")), col=seq(1,4), lty=lty.seq, \r\n horiz=T,lwd=rep(2,4), cex=0.9)\r\n \r\n title (main=\"Annual F(%SPR) Reference Points\", outer=T, line=-1 ) \r\n \r\n if (save.plots) savePlot(paste0(od, \"Annual.FSPR.\", plotf), type=plotf)\r\n \r\n \r\n plot(years, ypr.spr.vals[,1], type='n', xlab=\"Year\", ylab=\"YPR (%SPR)\", lwd=2,\r\n col=\"blue3\", ylim=c(0,1.2*max(ypr.spr.vals)) )\r\n for (i in 1:n.spr) {\r\n lines(years, ypr.spr.vals[,i], lwd=2, col=i, lty=lty.seq[i] ) \r\n }\r\n if(length(pspr)==0) legend('top', legend=c(\"YPR20%\", \"YPR30%\", \"YPR40%\", \"YPR50%\"), col=seq(1,4), lty=lty.seq, \r\n horiz=T,lwd=rep(2,4), cex=0.9)\r\n if(length(pspr)>0) legend('top', legend=c(paste0(\"YPR\", round(100*pspr,0), \"%\")), col=seq(1,4), lty=lty.seq, \r\n horiz=T,lwd=rep(2,4), cex=0.9)\r\n \r\n title (main=\"Annual YPR(%SPR) Reference Points\", outer=T, line=-1 ) \r\n \r\n if (save.plots) savePlot(paste0(od, \"Annual.YPR.\", plotf), type=plotf)\r\n \r\n if(length(pspr)==0) {\r\n f.hist.20 <- hist(f.spr.vals[,1], plot=F)\r\n f.hist.30 <- hist(f.spr.vals[,2], plot=F)\r\n f.hist.40 <- hist(f.spr.vals[,3], plot=F)\r\n f.hist.50 <- hist(f.spr.vals[,4], plot=F)\r\n \r\n par(mfrow=c(2,2), mar=c(4,4,2,2), oma=c(0,0,2,0) )\r\n \r\n plot(f.hist.20$mids, f.hist.20$counts, xlab=\"Full F20%\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(f.hist.20$counts)), col='blue4' )\r\n lines(f.hist.20$mids, f.hist.20$counts, lwd=2, col='blue4')\r\n plot(f.hist.30$mids, f.hist.30$counts, xlab=\"Full F30%\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(f.hist.30$counts)), col='blue4' )\r\n lines(f.hist.30$mids, f.hist.30$counts, lwd=2, col='blue4')\r\n plot(f.hist.40$mids, f.hist.40$counts, xlab=\"Full F40%\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(f.hist.40$counts)), col='blue4' )\r\n lines(f.hist.40$mids, f.hist.40$counts, lwd=2, col='blue4')\r\n plot(f.hist.50$mids, f.hist.50$counts, xlab=\"Full F50%\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(f.hist.50$counts)), col='blue4' )\r\n lines(f.hist.50$mids, f.hist.50$counts, lwd=2, col='blue4')\r\n \r\n title (main=\"Annual F (%SPR) Reference Points\", outer=T, line=-1 ) \r\n \r\n if (save.plots) savePlot(paste0(od, \"Annual.F_SPR.4panel.hist.\", plotf), type=plotf) \r\n \r\n \r\n ypr.hist.20 <- hist(ypr.spr.vals[,1], plot=F)\r\n ypr.hist.30 <- hist(ypr.spr.vals[,2], plot=F)\r\n ypr.hist.40 <- hist(ypr.spr.vals[,3], plot=F)\r\n ypr.hist.50 <- hist(ypr.spr.vals[,4], plot=F)\r\n \r\n par(mfrow=c(2,2), mar=c(4,4,2,2), oma=c(0,0,2,0) )\r\n \r\n plot(ypr.hist.20$mids, ypr.hist.20$counts, xlab=\"YPR (F20%)\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(ypr.hist.20$counts)), col='blue4' )\r\n lines(ypr.hist.20$mids, ypr.hist.20$counts, lwd=2, col='blue4')\r\n plot(ypr.hist.30$mids, ypr.hist.30$counts, xlab=\"YPR (F30%)\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(ypr.hist.30$counts)), col='blue4' )\r\n lines(ypr.hist.30$mids, ypr.hist.30$counts, lwd=2, col='blue4')\r\n plot(ypr.hist.40$mids, ypr.hist.40$counts, xlab=\"YPR (F40%)\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(ypr.hist.40$counts)), col='blue4' )\r\n lines(ypr.hist.40$mids, ypr.hist.40$counts, lwd=2, col='blue4')\r\n plot(ypr.hist.50$mids, ypr.hist.50$counts, xlab=\"YPR (F50%)\", ylab=\"Frequency\", type='h', lwd=2, \r\n ylim=c(0, max(ypr.hist.50$counts)), col='blue4' )\r\n lines(ypr.hist.50$mids, ypr.hist.50$counts, lwd=2, col='blue4')\r\n \r\n title (main=\"Annual YPR (%SPR) Reference Points\", outer=T, line=-1 ) \r\n \r\n if (save.plots) savePlot(paste0(od, \"Annual.YPR_SPR.4panel.hist.\", plotf), type=plotf) \r\n par(mfrow=c(1,1), mar=c(5.1,4.1,4.1,2.1), oma=c(0,0,0,0))\r\n } # end test if pspr has length 0\r\n \r\n spr.list <- list(f.spr.vals=f.spr.vals, ypr.spr.vals=ypr.spr.vals, conv.vals=conv.vals)\r\n return(spr.list)\r\n} # end function\r\n", "meta": {"hexsha": "5e128bcf030858359aaec3db7cf85bd6520a5ee3", "size": 6388, "ext": "r", "lang": "R", "max_stars_repo_path": "R/plot_annual_SPR_targets.r", "max_stars_repo_name": "liz-brooks/ASAPplots", "max_stars_repo_head_hexsha": "f42263d80f28b9de5d1abd2f87676d26bb30d4ec", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/plot_annual_SPR_targets.r", "max_issues_repo_name": "liz-brooks/ASAPplots", "max_issues_repo_head_hexsha": "f42263d80f28b9de5d1abd2f87676d26bb30d4ec", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/plot_annual_SPR_targets.r", "max_forks_repo_name": "liz-brooks/ASAPplots", "max_forks_repo_head_hexsha": "f42263d80f28b9de5d1abd2f87676d26bb30d4ec", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 47.3185185185, "max_line_length": 179, "alphanum_fraction": 0.5925172198, "num_tokens": 2494, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8558511469672594, "lm_q2_score": 0.6113819732941511, "lm_q1q2_score": 0.5232519630789055}} {"text": "###############################################################################\r\n# \"Infection Rate Models for COVID-19: \r\n# Model Risk and Public Health News Sentiment Exposure Adjustments\"\r\n# \r\n# Ioannis Chalkiadakis, Kevin Hongxuan Yan, Gareth W. Peters, Pavel V. Shevchenko\r\n#\r\n# Kevin Hongxuan Yan\r\n# March 2021\r\n###############################################################################\r\n\r\nfor (k in 1:7){\r\n\r\n\r\nD_GM=c(\"Jan/27\",\"Mar/17\",\"May/05\",\"Jun/25\",\"Jul/17\")\r\nD_IT=c(\"Jan/31\",\"Mar/13\",\"May/09\",\"Jun/29\",\"Jul/17\")\r\nD_JP=c(\"Jan/22\",\"Mar/12\",\"Apr/30\",\"Jun/19\",\"Jul/17\")\r\nD_SP=c(\"Feb/01\",\"Mar/20\",\"May/10\",\"Jun/30\",\"Jul/17\")\r\nD_UK=c(\"Jan/31\",\"Mar/21\",\"May/09\",\"Jun/29\",\"Jul/17\")\r\nD_US=c(\"Jan/22\",\"Mar/12\",\"Apr/30\",\"Jun/19\",\"Jul/17\")\r\nD_AU=c(\"Jan/26\",\"Mar/16\",\"May/04\",\"Jun/24\",\"Jul/17\")\r\n\r\nDate=c(D_GM,D_IT,D_JP,D_SP,D_UK,D_US,D_AU)\r\n\r\n\r\nname=c(\"Germany\",\"Italy\",\"Japan\",\"Spain\",\"U.K.\",\"U.S.\",\"Australia\")\r\nabn=c(\"GM\",\"IT\",\"JP\",\"SP\",\"UK\",\"US\",\"AU\")\r\n\r\n\r\n\r\nsetwd(\"./covid19modelrisk/output/M4\")\r\n\r\nload(paste(\"M2_\",abn[k],\".RData\",sep=\"\"))\r\n\r\n\r\nEpars_m <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(6+N):(5+2*N)] , 2 , mean )\r\nEpars_5 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(6+N):(5+2*N)] , 2 , quantile, probs=0.05 )\r\nEpars_9 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(6+N):(5+2*N)] , 2 , quantile, probs=0.95 )\r\n\r\n\r\ny=log(y)\r\nEpars_m=log(Epars_m)\r\nEpars_9=log(Epars_9)\r\nEpars_5=log(Epars_5)\r\n\r\nsetwd(\"./covid19modelrisk/output/plot\")\r\n\r\npdf(paste(\"M2_\",abn[k],\"_log.pdf\",sep=\"\"), width = 16.5, height = 8.50) \r\n\r\nm=length(y)\r\n\r\nnewx=c(1:m)\r\npar(cex.axis=2.5,cex.lab=2.5,cex.main=2.5,mar=c(5, 5, 3, 0.5))\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M2 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\n\r\npolygon(c(rev(newx), newx), c(rev(Epars_9[1:m]), Epars_5[1:m]), col = \"grey80\", border = NA)\r\n\r\nlines(newx,Epars_9[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\nlines(newx, Epars_5[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\npar(new=TRUE)\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M2 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\nplot.ts(Epars_m[1:m],col=\"red\",ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M2 model for \",name[k],sep=\"\"),lwd = 2,xaxt = 'n')\r\nlegend(\"bottomright\", legend=c(\"Observed data\", \"Fitted data\"), col=c(\"black\", \"red\"), lty=c(1,1), lwd = 3,cex=2)\r\naxis(1, at=c(0,50,100,150,180), labels=c(Date[(k-1)*5+1],Date[(k-1)*5+2],Date[(k-1)*5+3],Date[(k-1)*5+4],Date[(k-1)*5+5])) \r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\n\r\nsetwd(\"./covid19modelrisk/output/M2\")\r\n\r\nload(paste(\"M4_\",abn[k],\".RData\",sep=\"\"))\r\n\r\n\r\nEpars_m <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , mean )\r\nEpars_5 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.05 )\r\nEpars_9 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.95 )\r\n\r\n\r\ny=log(y)\r\nEpars_m=log(Epars_m)\r\nEpars_9=log(Epars_9)\r\nEpars_5=log(Epars_5)\r\n\r\nsetwd(\"./covid19modelrisk/output/plot\")\r\n\r\npdf(paste(\"M4_\",abn[k],\"_log.pdf\",sep=\"\"), width = 16.5, height = 8.50) \r\n\r\nm=length(y)\r\n\r\nnewx=c(1:m)\r\npar(cex.axis=2.5,cex.lab=2.5,cex.main=2.5,mar=c(5, 5, 3, 0.5))\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M4 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\n\r\npolygon(c(rev(newx), newx), c(rev(Epars_9[1:m]), Epars_5[1:m]), col = \"grey80\", border = NA)\r\n\r\nlines(newx,Epars_9[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\nlines(newx, Epars_5[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\npar(new=TRUE)\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M4 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\nplot.ts(Epars_m[1:m],col=\"red\",ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M4 model for \",name[k],sep=\"\"),lwd = 2,xaxt = 'n')\r\nlegend(\"bottomright\", legend=c(\"Observed data\", \"Fitted data\"), col=c(\"black\", \"red\"), lty=c(1,1), lwd = 3,cex=2)\r\naxis(1, at=c(0,50,100,150,180), labels=c(Date[(k-1)*5+1],Date[(k-1)*5+2],Date[(k-1)*5+3],Date[(k-1)*5+4],Date[(k-1)*5+5])) \r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\n\r\nsetwd(\"./covid19modelrisk/output/M7\")\r\n\r\nload(paste(\"M7_\",abn[k],\".RData\",sep=\"\"))\r\n\r\n\r\nEpars_m <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , mean )\r\nEpars_5 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.05 )\r\nEpars_9 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.95 )\r\n\r\n\r\ny=log(y)\r\nEpars_m=log(Epars_m)\r\nEpars_9=log(Epars_9)\r\nEpars_5=log(Epars_5)\r\n\r\nsetwd(\"./covid19modelrisk/output/plot\")\r\n\r\npdf(paste(\"M7_\",abn[k],\"_log.pdf\",sep=\"\"), width = 16.5, height = 8.50) \r\n\r\nm=length(y)\r\n\r\nnewx=c(1:m)\r\npar(cex.axis=2.5,cex.lab=2.5,cex.main=2.5,mar=c(5, 5, 3, 0.5))\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M7 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\n\r\npolygon(c(rev(newx), newx), c(rev(Epars_9[1:m]), Epars_5[1:m]), col = \"grey80\", border = NA)\r\n\r\nlines(newx,Epars_9[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\nlines(newx, Epars_5[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\npar(new=TRUE)\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M7 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\nplot.ts(Epars_m[1:m],col=\"red\",ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M7 model for \",name[k],sep=\"\"),lwd = 2,xaxt = 'n')\r\nlegend(\"bottomright\", legend=c(\"Observed data\", \"Fitted data\"), col=c(\"black\", \"red\"), lty=c(1,1), lwd = 3,cex=2)\r\naxis(1, at=c(0,50,100,150,180), labels=c(Date[(k-1)*5+1],Date[(k-1)*5+2],Date[(k-1)*5+3],Date[(k-1)*5+4],Date[(k-1)*5+5])) \r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\nsetwd(\"./covid19modelrisk/output/M12\")\r\n\r\nload(paste(\"M12_\",abn[k],\".RData\",sep=\"\"))\r\n\r\n\r\nEpars_m <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , mean )\r\nEpars_5 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.05 )\r\nEpars_9 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.95 )\r\n\r\n\r\ny=log(y)\r\nEpars_m=log(Epars_m)\r\nEpars_9=log(Epars_9)\r\nEpars_5=log(Epars_5)\r\n\r\nsetwd(\"./covid19modelrisk/output/plot\")\r\n\r\npdf(paste(\"M12_\",abn[k],\"_log.pdf\",sep=\"\"), width = 16.5, height = 8.50) \r\n\r\nm=length(y)\r\n\r\nnewx=c(1:m)\r\npar(cex.axis=2.5,cex.lab=2.5,cex.main=2.5,mar=c(5, 5, 3, 0.5))\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M12 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\n\r\npolygon(c(rev(newx), newx), c(rev(Epars_9[1:m]), Epars_5[1:m]), col = \"grey80\", border = NA)\r\n\r\nlines(newx,Epars_9[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\nlines(newx, Epars_5[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\npar(new=TRUE)\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M12 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\nplot.ts(Epars_m[1:m],col=\"red\",ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M12 model for \",name[k],sep=\"\"),lwd = 2,xaxt = 'n')\r\nlegend(\"bottomright\", legend=c(\"Observed data\", \"Fitted data\"), col=c(\"black\", \"red\"), lty=c(1,1), lwd = 3,cex=2)\r\naxis(1, at=c(0,50,100,150,180), labels=c(Date[(k-1)*5+1],Date[(k-1)*5+2],Date[(k-1)*5+3],Date[(k-1)*5+4],Date[(k-1)*5+5])) \r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\nD_GM=c(\"Jan/27\",\"May/05\",\"Aug/13\",\"Nov/21\",\"Jan/10\")\r\nD_IT=c(\"Jan/31\",\"May/09\",\"Aug/17\",\"Nov/25\",\"Jan/14\")\r\nD_JP=c(\"Jan/22\",\"Apr/30\",\"Aug/08\",\"Nov/16\",\"Jan/5\")\r\nD_SP=c(\"Feb/01\",\"May/10\",\"Aug/18\",\"Nov/26\",\"Jan/15\")\r\nD_UK=c(\"Jan/31\",\"May/09\",\"Aug/17\",\"Nov/25\",\"Jan/14\")\r\nD_US=c(\"Jan/22\",\"Apr/30\",\"Aug/08\",\"Nov/16\",\"Jan/5\")\r\nD_AU=c(\"Jan/26\",\"May/04\",\"Aug/12\",\"Nov/20\",\"Jan/9\")\r\n\r\nDate=c(D_GM,D_IT,D_JP,D_SP,D_UK,D_US,D_AU)\r\n\r\n\r\n\r\nname=c(\"Germany\",\"Italy\",\"Japan\",\"Spain\",\"U.K.\",\"U.S.\",\"Australia\")\r\nabn=c(\"GM\",\"IT\",\"JP\",\"SP\",\"UK\",\"US\",\"AU\")\r\n\r\n\r\n\r\nsetwd(\"./covid19modelrisk/output/M2_extra\")\r\n\r\nload(paste(\"M2_\",abn[k],\".RData\",sep=\"\"))\r\n\r\n\r\nEpars_m <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(6+N):(5+2*N)] , 2 , mean )\r\nEpars_5 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(6+N):(5+2*N)] , 2 , quantile, probs=0.05 )\r\nEpars_9 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(6+N):(5+2*N)] , 2 , quantile, probs=0.95 )\r\n\r\n\r\ny=log(y)\r\nEpars_m=log(Epars_m)\r\nEpars_9=log(Epars_9)\r\nEpars_5=log(Epars_5)\r\n\r\nsetwd(\"./covid19modelrisk/output/plot\")\r\n\r\npdf(paste(\"M2_\",abn[k],\"_log_LONG.pdf\",sep=\"\"), width = 16.5, height = 8.50) \r\n\r\nm=length(y)\r\n\r\nnewx=c(1:m)\r\npar(cex.axis=2.5,cex.lab=2.5,cex.main=2.5,mar=c(5, 5, 3, 0.5))\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M2 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\n\r\npolygon(c(rev(newx), newx), c(rev(Epars_9[1:m]), Epars_5[1:m]), col = \"grey80\", border = NA)\r\n\r\nlines(newx,Epars_9[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\nlines(newx, Epars_5[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\npar(new=TRUE)\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M2 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\nplot.ts(Epars_m[1:m],col=\"red\",ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M2 model for \",name[k],sep=\"\"),lwd = 2,xaxt = 'n')\r\nlegend(\"bottomright\", legend=c(\"Observed data\", \"Fitted data\"), col=c(\"black\", \"red\"), lty=c(1,1), lwd = 3,cex=2)\r\naxis(1, at=c(0,100,200,300,350), labels=c(Date[(k-1)*5+1],Date[(k-1)*5+2],Date[(k-1)*5+3],Date[(k-1)*5+4],Date[(k-1)*5+5])) \r\n \r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\nsetwd(\"./covid19modelrisk/output/M7_extra\")\r\n\r\nload(paste(\"M7_\",abn[k],\".RData\",sep=\"\"))\r\n\r\n\r\nEpars_m <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , mean )\r\nEpars_5 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.05 )\r\nEpars_9 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.95 )\r\n\r\n\r\ny=log(y)\r\nEpars_m=log(Epars_m)\r\nEpars_9=log(Epars_9)\r\nEpars_5=log(Epars_5)\r\n\r\nsetwd(\"./covid19modelrisk/output/plot\")\r\n\r\npdf(paste(\"M7_\",abn[k],\"_log_LONG.pdf\",sep=\"\"), width = 16.5, height = 8.50) \r\n\r\nm=length(y)\r\n\r\nnewx=c(1:m)\r\npar(cex.axis=2.5,cex.lab=2.5,cex.main=2.5,mar=c(5, 5, 3, 0.5))\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M7 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\n\r\npolygon(c(rev(newx), newx), c(rev(Epars_9[1:m]), Epars_5[1:m]), col = \"grey80\", border = NA)\r\n\r\nlines(newx,Epars_9[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\nlines(newx, Epars_5[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\npar(new=TRUE)\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M7 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\nplot.ts(Epars_m[1:m],col=\"red\",ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M7 model for \",name[k],sep=\"\"),lwd = 2,xaxt = 'n')\r\nlegend(\"bottomright\", legend=c(\"Observed data\", \"Fitted data\"), col=c(\"black\", \"red\"), lty=c(1,1), lwd = 3,cex=2)\r\naxis(1, at=c(0,100,200,300,350), labels=c(Date[(k-1)*5+1],Date[(k-1)*5+2],Date[(k-1)*5+3],Date[(k-1)*5+4],Date[(k-1)*5+5])) \r\n \r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\nsetwd(\"./covid19modelrisk/output/M12_extra\")\r\n\r\nload(paste(\"M12_\",abn[k],\".RData\",sep=\"\"))\r\n\r\n\r\nEpars_m <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , mean )\r\nEpars_5 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.05 )\r\nEpars_9 <- apply( as.data.frame(fit@sim[[1]][[1]])[10001:100000,(7+N):(6+2*N)] , 2 , quantile, probs=0.95 )\r\n\r\n\r\ny=log(y)\r\nEpars_m=log(Epars_m)\r\nEpars_9=log(Epars_9)\r\nEpars_5=log(Epars_5)\r\n\r\nsetwd(\"./covid19modelrisk/output/plot\")\r\n\r\npdf(paste(\"M12_\",abn[k],\"_log_LONG.pdf\",sep=\"\"), width = 16.5, height = 8.50) \r\n\r\nm=length(y)\r\n\r\nnewx=c(1:m)\r\npar(cex.axis=2.5,cex.lab=2.5,cex.main=2.5,mar=c(5, 5, 3, 0.5))\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M12 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\n\r\npolygon(c(rev(newx), newx), c(rev(Epars_9[1:m]), Epars_5[1:m]), col = \"grey80\", border = NA)\r\n\r\nlines(newx,Epars_9[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\nlines(newx, Epars_5[1:m], lty = 'dashed', col = 'red',lwd = 2)\r\npar(new=TRUE)\r\nplot(y[1:m],ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M12 model for \",name[k],sep=\"\"),pch=19,xaxt = 'n')\r\npar(new=TRUE)\r\nplot.ts(Epars_m[1:m],col=\"red\",ylim=c(min(y[1:m]),max(y[1:m],Epars_9[1:m])),ylab=\"Count\",xlab=\"Time\",main=paste(\"In-sample fit results of M12 model for \",name[k],sep=\"\"),lwd = 2,xaxt = 'n')\r\nlegend(\"bottomright\", legend=c(\"Observed data\", \"Fitted data\"), col=c(\"black\", \"red\"), lty=c(1,1), lwd = 3,cex=2)\r\naxis(1, at=c(0,100,200,300,350), labels=c(Date[(k-1)*5+1],Date[(k-1)*5+2],Date[(k-1)*5+3],Date[(k-1)*5+4],Date[(k-1)*5+5])) \r\n \r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\n}\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n", "meta": {"hexsha": "1fa19bf9d7f2643ac589589bee197b1e4a406d1e", "size": 13922, "ext": "r", "lang": "R", "max_stars_repo_path": "R/New_log_plot.r", "max_stars_repo_name": "ichalkiad/covid19modelrisk", "max_stars_repo_head_hexsha": "eaf14f373411c0122c73439f089e9626164c87d1", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/New_log_plot.r", "max_issues_repo_name": "ichalkiad/covid19modelrisk", "max_issues_repo_head_hexsha": "eaf14f373411c0122c73439f089e9626164c87d1", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/New_log_plot.r", "max_forks_repo_name": "ichalkiad/covid19modelrisk", "max_forks_repo_head_hexsha": "eaf14f373411c0122c73439f089e9626164c87d1", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-04-05T00:19:10.000Z", "max_forks_repo_forks_event_max_datetime": "2021-04-05T00:19:10.000Z", "avg_line_length": 37.4247311828, "max_line_length": 190, "alphanum_fraction": 0.5905042379, "num_tokens": 5742, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7185944046238982, "lm_q2_score": 0.727975460709318, "lm_q1q2_score": 0.5231190927692203}} {"text": "# Functions for Multifractal analysis \n\n## p-model\n\n# Calculates pmodel using the C++ compiled program\n# the output is a file in \"sed\" format \n#\n# I use the following to have only one parameter:\n# $p2=p3=p4=(1-p1)/3$\n#\n# fname: output file name\n# p1: p1 parameter for pmodel\n# iter: (default=9) gives the image size $lado = 2^iter$\n# rnd: how I assign p at each step S->random with reposition\n#\n\ncalc_pmodel1 <- function(fname,p1,iter=9,rnd=\"S\")\n {\n if(p1==0 | iter<2) stop(\"p1 == 0 or iter<2\") \n if(file.exists(fname))\n file.remove(fname)\n p2 <- p3 <- p4 <- (1-p1)/3\n syst.txt <- paste(\"./pmodel\", p1,p2,p3,p4,iter,fname,rnd)\n system(syst.txt)\n }\n\n# Plot a sed file from pmodel using lattice\n#\n# fname: sed file with pmodel output\n# p1: parameter for title\n#\nplot_pmodel1 <- function(fname,p1)\n {\n require(lattice)\n per <- read.table(fname, skip=2,header=F)\n per <-data.matrix(per)*(2^(2*iter))\n col.l <- colorRampPalette(c('white', 'green', 'purple', 'yellow', 'brown'))(64) \n levelplot(per, scales = list(draw = FALSE),xlab =NULL, ylab = NULL,col.regions=col.l,\n useRaster=T,\n main=list( paste(\"p1=\",p1),cex=1))\n }\n\n# Plot a sed file\n#\n# fname: file name\n# gname: graph title \n# dX: range in columns to plot\n# col: vector of colors to make the color palette\n# shf: shift in the vector of colors \n#\nplot_image <- function(fname,gname,dX=0,col=0,shf=0)\n {\n require(lattice)\n require(RColorBrewer)\n per <-data.matrix(read.table(fname, skip=2,header=F))\n if(length(dX)>1) per <- per[,dX]\n mp = max(per)\n if(mp<50) {\n mp = 50\n seqat = seq( min(per),max(per),(max(per)-min(per))/50)\n }\n else\n {\n seqat = seq( min(per),mp,5)\n }\n if(length(col)==1) col.l <- colorRampPalette(c('white', 'green', 'purple', 'yellow', 'brown'))(mp) \n else col.l <- colorRampPalette(col)(mp) \n if( shf>0) col.l = col.l[shf:mp]\n levelplot(per, scales = list(draw = FALSE),xlab =NULL, ylab = NULL,col.regions=col.l,\n useRaster=T,at=seqat,\n main=list( gname,cex=1))\n }\n\n# Read a sed file in a matrix\n#\n#\nread_sed <- function(fname)\n{\n per <-data.matrix(read.table(fname, skip=2,header=F))\n}\n\n# Calculates the multifractal spectrum and \n# put the result in a frame\n# \ncalcDq_mfSBA <- function(fname,recalc=FALSE)\n {\n sname <- paste0(\"s.\", fname)\n if((!file.exists(sname)) | recalc)\n {\n syst.txt <- paste(\"./mfSBA \",fname, \"q.sed 2 512 20 S\")\n system(syst.txt)\n }\n pp <- read.table(sname, header=T)\n\n pp$Dq <- with(pp,ifelse(q==1,alfa,Tau/(q-1)))\n pp$SD.Dq <- with(pp,ifelse(q==1,SD.alfa,abs(SD.Tau/(q-1))))\n pp$R.Dq <- with(pp,ifelse(q==1,R.alfa,R.Tau))\n return(pp[,c(\"q\",\"Dq\",\"SD.Dq\",\"R.Dq\")])\n }\n\n# Calc mfSBA\n# draw histogram\n# add to a dataframe dq with a new variable site = siten\n#\naddDqRHist <- function(fname,dq=0,siten=\"\")\n{\n # Extract site label\n if( nchar(siten)==0 ){ \n siten <- strsplit(fname,\".sed\")[[1]][1]\n }\n \n # Make multifractal analysis\n \n dqnew <- calcDq_mfSBA(fname)\n dqnew$Site <- siten\n \n require(ggplot2)\n \n #print(ggplot(dqnew, aes(x=R.Dq)) + geom_histogram() + ggtitle(siten))\n hist(dqnew$R.Dq)\n if(is.data.frame(dq))\n return(rbind(dq,dqnew))\n else\n return(dqnew)\n}\n\n# Do repeated simulations of pmodel and store the results in a dataframe.\n#\nrepeat_pmodel <- function(p1,fname,reps) {\n \n for(i in 1:reps)\n {\n calc_pmodel1(fname,p1)\n #plot_pmodel1(fname,p1)\n s_Dq <- calcDq_mfSBA(fname,T) # recalc = TRUE\n s_Dq$p1 <- p1\n s_Dq <-na.omit(s_Dq)\n a_Dq <-if(!exists(\"a_Dq\")) s_Dq else rbind(a_Dq,s_Dq)\n }\n #plot_pmodel1(fname,p1)\n return(a_Dq)\n}\n\n\n# Auxiliar function for boot\n#\nmediaDq <- function(dq,i) mean(dq[i])\n\n# Calculates CI using boot \n#\n# sp: frame to put Dq|q\n# the following are to identify the output \n# p1: the value of p1 used in p-model \n# type: the type of the run \ncalcDq_bootCI <- function(sp,p1,type)\n {\n require(boot)\n require(plyr)\n sp <- na.omit(sp)\n testBoot <- with(sp,by(Dq,q,boot, mediaDq, R=1000))\n # boot pg 331 \n bt <-boot.ci(testBoot[[1]])\n testCI <- sapply(testBoot, boot.ci)\n compCI1<-data.frame(matrix(unlist(testCI),ncol=21,byrow=T)[,c(2,15,16)])\n names(compCI1)=c(\"Dq\",\"LowCI\",\"HighCI\")\n compCI1$q<-as.numeric(names(testBoot))\n compCI1$p1 <-p1\n compCI1$Type <-type\n compCI1$SD.Dq <-ldply(testBoot, function(e) sd(e$t))$V1\n return(compCI1)\n }\n\n\n# Function to graph multifractal spectra with CI\n#\n# El error dinnames es que no todas las variables son numéricas\n# Usa Type como variable de grupo y LowCI,HighCI para el CI\n#\nplot_DqCI <- function(DqCI, tit=\"Type\")\n {\n require(lattice)\n require(Hmisc)\n with(DqCI,\n xYplot(Cbind(Dq,LowCI,HighCI) ~ q, data=DqCI, nx=FALSE, groups=Type, type=\"l\",\n scales=list(tck=-1),\n cap=.01,\n ylim=c(min(DqCI$LowCI)-.01,max(DqCI$HighCI)+.01),\n \t panel=function(...){\n \t\t panel.abline(h=2, col=\"grey\", lty=2)\n \t\t panel.xYplot(...)},\n \t\t ylab=expression(italic(D[q])),\n \t\t xlab=expression(italic(q)),\n label.curves=F,\n auto.key=list(x=.65,y=.9,title=tit,cex.title=.9, lines=T,points=F,cex=.7),\n )\n )\n }\n\n# Calculates theoretic Dq from pmodel\n#\ncalcDqTeor <- function(q,p1) {\n if( q==1)\n q <- q+1e-10\n p2 <- p3 <- p4 <- (1-p1)/3\n f1 <- p1/(p1+p2+p3+p4)\n f2 <- p2/(p1+p2+p3+p4)\n f3 <- p3/(p1+p2+p3+p4)\n f4 <- p4/(p1+p2+p3+p4)\n dq <- log2(f1^q+f2^q+f3^q+f4^q)/(1-q)\n return(dq)\n }\n\n# Function for CI from SD \n#\ncalcCI <- function(dat,n,p) qt(1-p/2,n-1)*sqrt(dat*dat/n)\n\n# Function to plot Dq fit from t* files generated by mfSBA \n# Beware! Is fixed for q interval -5 to 5\n#\nplotDqFit <- function(fname,wtitle=\"\")\n{\n \n zq <- read.table(fname, sep=\"\\t\", skip=1,\n col.names=c(\"BoxSize\",\"logBox\",\"Xminus5\",\"Xminus4.5\",\"Xminus4\",\"Xminus3.5\",\"Xminus3\",\"Xminus2.5\",\"Xminus2\",\"Xminus1.5\",\"Xminus1\",\"Xminus0.5\",\"X0\",\"X0.5\",\"X1\",\"X1.5\",\"X2\",\"X2.5\",\"X3\",\"X3.5\",\"X4\",\"X4.5\",\"X5\")\n )\n #cna=seq(-5,5,by=0.5)\n cna =c(-5,-4.5,-4,-3.5,-3,-2.5,-2,-1.5,-1,-0.5,0,0.5,1,1.5,2,2.5,3,3.5,4,4.5,5)\n zq0=reshape(zq, timevar=\"q\",times=cna,v.names=c(\"logTr\"),\n varying=list(c(\"Xminus5\",\"Xminus4.5\",\"Xminus4\",\"Xminus3.5\",\"Xminus3\",\"Xminus2.5\",\"Xminus2\",\"Xminus1.5\",\"Xminus1\",\"Xminus0.5\",\"X0\",\"X0.5\",\"X1\",\"X1.5\",\"X2\",\"X2.5\",\"X3\",\"X3.5\",\"X4\",\"X4.5\",\"X5\")),\n direction=\"long\")\n \n \n library(lattice)\n zq1 <- subset(zq0, q==1 | q==2 | q==3 | q==4 | q==5 | q==0 | q==-1 | q==-2 | q==-3 | q==-4 | q==-5 )\n\n# oname <- paste(\"lsaravia_figS_W\",sem,\"_\",wnro,\".tif\",sep=\"\")\n \n# tiff(oname, width=4.86,height=4.86,units=\"in\",res=600,compression=c(\"lzw\"))\n #devAskNewPage(TRUE)\n \n trellis.par.set(superpose.symbol=list(pch=c(0,1,2,3,4,5,6,8,15,16,17)))\n trellis.par.set(superpose.symbol=list(cex=c(rep(0.6,11))))\n trellis.par.set(superpose.line=list(lty=3))\n \n #show.settings()\n \n print(xyplot(logTr~logBox , data =zq1, groups=q, type=c(\"r\",\"p\"), scales=list(tck=-1), \n main=list(wtitle,cex=0.9),\n auto.key=list(space = \"right\",title=expression(italic(\"q\")),cex.title=.7, points=TRUE,cex=.7),\n ylab=expression(italic(paste(\"log \", Z[q](epsilon) ))) , xlab=expression(italic(paste(\"log \",epsilon))) \n ))\n# dev.off() \n}\n\n\n# Calculates Dq from a data.frame read from the output of neutral model \n# auxiliar function for the next one\n#\ncalcDq_frame <- function(pp)\n {\n pp$Dq <- with(pp,ifelse(q==1,alfa,Tau/(q-1)))\n pp$SD.Dq <- with(pp,ifelse(q==1,SD.alfa,abs(SD.Tau/(q-1))))\n pp$R.Dq <- with(pp,ifelse(q==1,R.alfa,R.Tau))\n return(pp[,c(1:5,15:17)])\n } \n\n# Reads the output of multifractal spectra of neutral model\n# an calculates Dq \n#\nreadNeutral_calcDq <-function(fname)\n{\n md1 <- read.table(fname,header=F,skip=1)\n md1 <- md1[,c(2:4, 6:16)]\n names(md1)<-c(\"MortalityRate\",\"DispersalDistance\",\"ColonizationRate\",\"Time\",\"q\",\"Tau\",\"alfa\",\"f(alfa)\",\"R.Tau\",\"R.alfa\",\"R.f\",\"SD.Tau\",\"SD.alfa\",\"SD.f\")\n \n md1 <-calcDq_frame(md1)\n}\n\n\n# Save a matrix as a sed file with type BI (floating point)\n#\nsave_matrix_as_sed <- function(mat,fname)\n{\n header <- paste(nrow(mat),ncol(mat),\"BI\")\n write.table(header,file=fname,row.names=F,col.names=F,quote=F)\n write.table(mat,file=fname,row.names=F,col.names=F,quote=F,append=T)\n}\n", "meta": {"hexsha": "ffd8a782ed7f5f3c7b7fb60557325cfa30fbcb6f", "size": 8308, "ext": "r", "lang": "R", "max_stars_repo_path": "R/Fun_Multi_patt.r", "max_stars_repo_name": "lsaravia/SpeciesRankSurface", "max_stars_repo_head_hexsha": "33e71b7b50e1e86d8a420812efd1b96ffb22e2dc", "max_stars_repo_licenses": ["CC-BY-4.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/Fun_Multi_patt.r", "max_issues_repo_name": "lsaravia/SpeciesRankSurface", "max_issues_repo_head_hexsha": "33e71b7b50e1e86d8a420812efd1b96ffb22e2dc", "max_issues_repo_licenses": ["CC-BY-4.0"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2015-05-09T12:48:54.000Z", "max_issues_repo_issues_event_max_datetime": "2015-05-09T12:48:54.000Z", "max_forks_repo_path": "R/Fun_Multi_patt.r", "max_forks_repo_name": "lsaravia/SpeciesRankSurface", "max_forks_repo_head_hexsha": "33e71b7b50e1e86d8a420812efd1b96ffb22e2dc", "max_forks_repo_licenses": ["CC-BY-4.0"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2015-05-08T01:29:48.000Z", "max_forks_repo_forks_event_max_datetime": "2020-02-26T13:41:06.000Z", "avg_line_length": 28.8472222222, "max_line_length": 225, "alphanum_fraction": 0.6109773712, "num_tokens": 3000, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6825737473266735, "lm_q2_score": 0.7662936377487304, "lm_q1q2_score": 0.5230519198707394}} {"text": "LISP Interpreter Run\n\n[[[[[\n\n A LISP expression that asserts that it itself is unprovable!\n\n Let g(x): x -> (is-unprovable (value-of (('x)('x))))\n\n Then (is-unprovable (value-of (('g)('g))))\n asserts that it itself is not a theorem!\n\n]]]]]\n\ndefine (g x) \n let (L x y) cons x cons y nil [Makes x and y into list.]\n (L is-unprovable (L value-of (L (L \"' x) (L \"' x))))\n\ndefine g\nvalue (lambda (x) ((' (lambda (L) (L is-unprovable (L va\n lue-of (L (L ' x) (L ' x)))))) (' (lambda (x y) (c\n ons x (cons y nil))))))\n\n\n[Here we try g:]\n\n(g x)\n\nexpression (g x)\nvalue (is-unprovable (value-of ((' x) (' x))))\n\n\n[\n Here we calculate the LISP expression \n that asserts its own unprovability: \n]\n\n(g g)\n\nexpression (g g)\nvalue (is-unprovable (value-of ((' (lambda (x) ((' (lamb\n da (L) (L is-unprovable (L value-of (L (L ' x) (L \n ' x)))))) (' (lambda (x y) (cons x (cons y nil))))\n ))) (' (lambda (x) ((' (lambda (L) (L is-unprovabl\n e (L value-of (L (L ' x) (L ' x)))))) (' (lambda (\n x y) (cons x (cons y nil))))))))))\n\n\n[Here we extract the part that it uses to name itself:]\n\ncadr cadr (g g)\n\nexpression (car (cdr (car (cdr (g g)))))\nvalue ((' (lambda (x) ((' (lambda (L) (L is-unprovable (\n L value-of (L (L ' x) (L ' x)))))) (' (lambda (x y\n ) (cons x (cons y nil))))))) (' (lambda (x) ((' (l\n ambda (L) (L is-unprovable (L value-of (L (L ' x) \n (L ' x)))))) (' (lambda (x y) (cons x (cons y nil)\n )))))))\n\n\n[Here we evaluate the name to get back the entire expression:] \n\neval cadr cadr (g g)\n\nexpression (eval (car (cdr (car (cdr (g g))))))\nvalue (is-unprovable (value-of ((' (lambda (x) ((' (lamb\n da (L) (L is-unprovable (L value-of (L (L ' x) (L \n ' x)))))) (' (lambda (x y) (cons x (cons y nil))))\n ))) (' (lambda (x) ((' (lambda (L) (L is-unprovabl\n e (L value-of (L (L ' x) (L ' x)))))) (' (lambda (\n x y) (cons x (cons y nil))))))))))\n\n\n[Here we check that it worked:]\n\n= (g g) eval cadr cadr (g g)\n\nexpression (= (g g) (eval (car (cdr (car (cdr (g g)))))))\nvalue true\n\nEnd of LISP Run\n\nElapsed time is 0 seconds.\n", "meta": {"hexsha": "b40a8c8a69f98ff9d18cc400bb1ee396f0bef9dc", "size": 2257, "ext": "r", "lang": "R", "max_stars_repo_path": "book-examples/godel.r", "max_stars_repo_name": "darobin/chaitin-lisp", "max_stars_repo_head_hexsha": "a06fd5647a1d69d41ec725616fa0ebcc71e55bec", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2018-02-28T09:21:07.000Z", "max_stars_repo_stars_event_max_datetime": "2021-01-09T03:29:32.000Z", "max_issues_repo_path": "book-examples/godel.r", "max_issues_repo_name": "darobin/chaitin-lisp", "max_issues_repo_head_hexsha": "a06fd5647a1d69d41ec725616fa0ebcc71e55bec", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "book-examples/godel.r", "max_forks_repo_name": "darobin/chaitin-lisp", "max_forks_repo_head_hexsha": "a06fd5647a1d69d41ec725616fa0ebcc71e55bec", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2016-06-23T14:37:37.000Z", "max_forks_repo_forks_event_max_datetime": "2019-04-19T13:09:35.000Z", "avg_line_length": 26.869047619, "max_line_length": 63, "alphanum_fraction": 0.4957908728, "num_tokens": 743, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.712232184238947, "lm_q2_score": 0.7341195210831258, "lm_q1q2_score": 0.5228635499934844}} {"text": "#' Calculates several measures of fit for Linear Mixel Models\n#' based on Lou and Azen (2013) text.\n#' Models could be lmer or lme models\n#' @param m.null Null model (only with random intercept effects)\n#' @param m.full Full model \n#' @return lmmR2 class\n#' @export\nlmmR2<-function(m.null, m.full) {\n\tif(is(m.null,\"lme\"))\t {\n\t\treturn(lmmR2.lme(m.null,m.full))\n\t} else if (class(m.null)==\"mer\" | class(m.null)==\"lmerMod\" | class(m.null)==\"lmerTest\") {\n\t\treturn(lmmR2.mer(m.null,m.full))\n\t} else {\n\t\tstop(\"Not implemented for other classes than lme or lmer\") \n\t}\n}\n\n# Calcula los cuatro R^2 presentes en el texto de Lou y Azen (2013).\n# Debo extraer los interceptos para los modelos\nlmmR2.mer<-function(m.null,m.full) {\n\t# Primero, verifico que tengan la misma estructura de grupos. Si \n\t# no, estoy puro leseando\n\tv.0<-lme4::VarCorr(m.null)\n\tv.1<-lme4::VarCorr(m.full)\n\tif(!all.equal(names(v.0),names(v.1))) {\n\t\tstop(\"Groups should be equal\")\n\t}\n\tn.b<-length(names(v.0))\n\t# recojo los sigmas\n\tsigmas=c(sigma(m.null)^2,sigma(m.full)^2)\n\t# recojo los thetas\n\tthetas.0=sapply(v.0,function(x) {x[1,1]})\n\tthetas.1=sapply(v.1,function(x) {x[1,1]})\n\t# recojo el largo promedio\n\tnn=sapply(m.null@flist,function(x) {\n\t\tlength(levels(x)) / sum(1/table(x))\n\t})\n\trb.r2.1<-1-(sigmas[2]/sigmas[1])\n\trb.r2.2<-1-(sum(thetas.1)/sum(thetas.0))\n\tsb.r2.1<-1-((sigmas[2]+sum(thetas.1))/(sigmas[1]+sum(thetas.0)))\n\tsb.r2.2<-1-( sigmas[2]+sum(thetas.1*nn)) / (sigmas[1]+sum(thetas.0*nn))\n\tout<-list(sigmas=sigmas,t0=thetas.0,t1=thetas.1, nn=nn, rb.r2.1=rb.r2.1, rb.r2.2=rb.r2.2, sb.r2.1=sb.r2.1, sb.r2.2=sb.r2.2)\n\tclass(out)<-\"lmmR2\"\n\tout\n}\n# Extractor de varianzas para nlme\nvars.lme<-function(x) {\n\tvv<-nlme::VarCorr(x)\n\t# Si tiene solo dos filas, recoge un solo factor\n\tif(nrow(vv)==2) {\n\t\tout=list()\n\t\tout[[colnames(x$groups)]]=as.numeric(vv[1,1])\n\t\treturn(out)\n\t} else {\n\t\tout=list()\n\t\tvv.n<-rownames(vv)\n\t\tcurrent.var=NULL\n\t\tfor(i in 1:nrow(vv)) {\n\t\t\tc.name=vv.n[i]\n\t\t\t#print(vv.n[i])\n\t\t\taa=grep(\"(.+) =\",c.name,value=T)\n\t\t\tif(length(aa)) {\n\t\t\t\taa=sub(\" =\",\"\",aa)\n\t\t\t\tcurrent.var=aa\n\t\t\t} else if(!is.null(current.var) & c.name==\"(Intercept)\") {\n\t\t\t\tout[[current.var]]=vv[i,1]\n\t\t\t}\n\t\t}\n\t\tout\n\t}\n}\n# Debo extraer los interceptos para los modelos\nlmmR2.lme<-function(m.null,m.full) {\n\t# Primero, verifico que tengan la misma estructura de grupos. Si \n\t# no, estoy puro leseando\n\tv.0<-vars.lme(m.null)\n\tv.1<-vars.lme(m.full)\n\tif(!all.equal(names(v.0),names(v.1))) {\n\t\tstop(\"Groups should be equal\")\n\t}\n\n\tn.b<-length(names(v.0))\n\t# recojo los sigmas\n\tsigmas=c(m.null$sigma^2,m.full$sigma^2)\n\t# recojo los thetas\n\tthetas.0=sapply(v.0,function(x) {as.numeric(x[1])})\n\tthetas.1=sapply(v.1,function(x) {as.numeric(x[1])})\n\t#print(thetas.0)\n\t#print(thetas.1)\n\t# recojo el largo promedio\n\tnn=sapply(m.null$groups,function(x) {\n\t\tlength(levels(x)) / sum(1/table(x))\n\t})\n\t\n\trb.r2.1<-1-(sigmas[2]/sigmas[1])\n\trb.r2.2<-1-(sum(thetas.1)/sum(thetas.0))\n\tsb.r2.1<-1-((sigmas[2]+sum(thetas.1))/(sigmas[1]+sum(thetas.0)))\n\tsb.r2.2<-1-( sigmas[2]+sum(thetas.1*nn)) / (sigmas[1]+sum(thetas.0*nn))\n\tout<-list(sigmas=sigmas,t0=thetas.0,t1=thetas.1, nn=nn,rb.r2.1=rb.r2.1, rb.r2.2=rb.r2.2, sb.r2.1=sb.r2.1, sb.r2.2=sb.r2.2)\n\tclass(out)<-\"lmmR2\"\n\tout\n}\n\n#' @export\nprint.lmmR2<-function(x) {\n\tprint(summary.lmmR2(x))\n}\n\n#' @export\nprint.summary.lmmR2<-function(xx) {\n\tcat(\"Explanatory power of Multilevel Model\\n\")\n\tcat(\"=====================================\\n\")\n\tcat(\"Variances:\\n\")\n\tprint(xx$m1)\n\tcat(\"Indexes:\\n\")\n\tprint(xx$m2,row.names=F)\n\tcat(\"\\n\")\n\t\n}\n\n#' @export\nsummary.lmmR2<-function(x) {\n \n v.null<-c(x$sigmas[1],x$t0)\n v.full<-c(x$sigmas[2],x$t1)\n \n\tm1<-data.frame(avg.size=c(1,x$nn),null=v.null, null.r=v.null/(sum(v.null)), full=v.full, pseudo.r2= 1-(v.full/v.null) )\n\t#cat(\"Variances:\\n\")\n\trownames(m1)[1]<-\"Residual\"\n\t#print(m1)\n\t#cat(\"Indexes:\\n\")\n\tm2<-with(x, data.frame(indexes=c(\"R & B R1\",\"R & B R2\",\"S & B R1\",\"S & B R2\"), \n\tmeaning=c(\"Within-cluster variance(relative)\",\n\t\"Between-cluster variance(relative)\",\n\t\"Reduce individual error(total)\",\n\t\"Reduce cluster error(total)\"\n\t),\n\tvals=c(rb.r2.1,rb.r2.2,sb.r2.1,sb.r2.2)))\n\t#print(m2, row.names=F)\n\tout=list(m1=m1,m2=m2)\n\t\n\tclass(out)<-\"summary.lmmR2\"\n\treturn(out)\n}\n", "meta": {"hexsha": "e9d8e21946920562e5b79982b2e6dcce20cee3be", "size": 4203, "ext": "r", "lang": "R", "max_stars_repo_path": "R/lmmR2.r", "max_stars_repo_name": "clbustos/r-glmmextra", "max_stars_repo_head_hexsha": "f82f94e78cc37824c4f3655b3b99b0d280a25d1a", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/lmmR2.r", "max_issues_repo_name": "clbustos/r-glmmextra", "max_issues_repo_head_hexsha": "f82f94e78cc37824c4f3655b3b99b0d280a25d1a", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/lmmR2.r", "max_forks_repo_name": "clbustos/r-glmmextra", "max_forks_repo_head_hexsha": "f82f94e78cc37824c4f3655b3b99b0d280a25d1a", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 28.9862068966, "max_line_length": 124, "alphanum_fraction": 0.6364501547, "num_tokens": 1603, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7772998611746912, "lm_q2_score": 0.6723317123102955, "lm_q1q2_score": 0.5226033466421351}} {"text": "GrayEncode <- function(binary) {\n\tgray <- substr(binary,1,1)\n\trepeat {\n\tif (substr(binary,1,1) != substr(binary,2,2)) gray <- paste(gray,\"1\",sep=\"\")\n\telse gray <- paste(gray,\"0\",sep=\"\")\n\tbinary <- substr(binary,2,nchar(binary))\n\tif (nchar(binary) <=1) {\n\t\tbreak\n\t\t}\n\t}\nreturn (gray)\n}\nGrayDecode <- function(gray) {\n\tbinary <- substr(gray,1,1)\n\trepeat {\n\tif (substr(binary,nchar(binary),nchar(binary)) != substr(gray,2,2)) binary <- paste(binary ,\"1\",sep=\"\")\n\telse binary <- paste(binary ,\"0\",sep=\"\")\n\tgray <- substr(gray,2,nchar(gray))\n\n\tif (nchar(gray) <=1) {\n\t\tbreak\n\t\t}\n\t}\nreturn (binary)\n}\n", "meta": {"hexsha": "09785f729dad924e08096c67688ea2581d81c964", "size": 597, "ext": "r", "lang": "R", "max_stars_repo_path": "Task/Gray-code/R/gray-code.r", "max_stars_repo_name": "LaudateCorpus1/RosettaCodeData", "max_stars_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2018-11-09T22:08:38.000Z", "max_stars_repo_stars_event_max_datetime": "2018-11-09T22:08:38.000Z", "max_issues_repo_path": "Task/Gray-code/R/gray-code.r", "max_issues_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_issues_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Task/Gray-code/R/gray-code.r", "max_forks_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_forks_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2018-11-09T22:08:40.000Z", "max_forks_repo_forks_event_max_datetime": "2018-11-09T22:08:40.000Z", "avg_line_length": 22.9615384615, "max_line_length": 105, "alphanum_fraction": 0.6164154104, "num_tokens": 185, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.6791786861878392, "lm_q2_score": 0.7690802317779601, "lm_q1q2_score": 0.5223429013919938}} {"text": "## LOOP FOR:\n\n# Los \"loops\" (\"lazos\" o \"bucles\" en español) son comandos especiales que \n# sirven para hacer ejecutar una tarea una cantidad arbitraria de veces; se \n# llama iteración a cada una de estas repeticiones. Sirven para hacer en \n# segundos lo que manualmente llevaría horas, días o sería simplemente demasiado.\n\n# Los comandos más universales para hacer loops en programación son \"for\" y\n# \"while\", y en R esto no es distinto.\n\n# Se puede acceder a la ayuda de R con varios comandos, por ejemplo (no correr):\n?Control\n?\"for\"\n\n\n## SINTAXIS\n\n# Empecemos con un ejemplo muy sencillo:\n\nfor (i in 1:4) {\n print(i)\n}\n\n# Analicemos lo que ocurre:\n# En la consola vemos que se imprimen los números 1, 2, 3 y 4. Esto ocurre \n# porque dentro del loop se puso el comando print(i). En este caso i es un \n# objeto, al cual llamaremos la \"variable\" o \"variable de iteración\". Cuando \n# miramos en el paréntesis del for, vemos que dice \"i in 1:4\". Esto quiere \n# decir que la variable i va a tomar los valores contenidos en el vector 1:4, \n# es decir que en la primer iteración i vale 1, luego 2 y así. En suma, es \n# equivalente a escribir:\ni <- 1\nprint(i)\ni <- 2\nprint(i)\ni <- 3\nprint(i)\ni <- 4\nprint(i)\n# ... pero por supuesto que es mucho más económico y elegante.\n\n# Nota: debido a que hay una sola expresión dentro del loop (print(i)), no es \n# necesario usar las llaves para delimitar el loop; son equivalentes estas dos \n# opciones:\nfor (i in 1:4)\n print(i)\n# ó\nfor (i in 1:4) print(i)\n\n# Nota: puede verificar que i es además un objeto que existe en su sesión y \n# cuyo valor es el último asignado por el loop for. En este caso i == 4:\ni\n\n# Por supuesto que este es uno de los ejemplos más elementales que se pueden \n# imaginar de un loop for, pero hay un par de detalles que por costumbre o \n# convención se repiten casi siempre:\n# - El usar el nombre \"i\" para la variable de iteración. No es necesario que \n# tenga este nombre, eso es un valor arbitrario (puede llamarse, j, k o carlos \n# sin problemas), pero es aconsejable mantener este criterio para facilitar la \n# comprensión del código.\n# - El utilizar números naturales para el vector de valores que va a tomar i, \n# en este caso 1:4. Aquí tampoco hay límites en cuanto a los valores o clase \n# de vector que se elige. A continuación veremos porqué es más común y \n# conveniente utilizar estos tipos de vectores.\n\n# Un ejemplo un poco más ilustrativo: consideremos que existe un vector v \n# definido así:\nv <- c(5, 2, -2, 3)\n# Ahora, supongamos que queremos imprimirlo en la consola, un valor a la vez. \n# Podemos recurrir a este código:\nfor (i in v)\n print(i)\n\n# Como puede ver, ahora la variable i va tomando los valores del vector v y \n# print(i) los va imprimiendo en la consola. Es equivalente a:\ni <- v[1]\nprint(i)\ni <- v[2]\nprint(i)\ni <- v[3]\nprint(i)\ni <- v[4]\nprint(i)\n\n# De todas formas, este ejemplo *no refleja el uso más frecuente* de la \n# variable i. Usualmente lo que se hace es utilizar i para indizar al vector v, \n# así:\nfor (i in 1:4)\n print(v[i])\n\n# A través de los corchetes estamos usando i para imprimir el iésimo elemento \n# de v en cada iteración. Parece un camino más largo y por lo tanto \n# inconveniente. Sin embargo en la práctica rara vez se va a utilizar la \n# variable i para indizar solamente un vector, si no que suele usarse \n# repetidamente para indizar diferentes objetos, incluyendo matrices, \n# data.frames o listas. Veamos un ejemplo:\n\n# Preparación:\nm <- matrix(rpois(12, 8), 4, 3) # Una matriz de 4x3\ns <- numeric(4) # Un vector de longitud 4, valores == 0\n# Loop:\nfor (i in 1:4)\n s[i] <- sum(m[i,])\ns\n# Aquí se creo un vector s, inicialmente todos sus valores 0, cuyo objetivo \n# era contener los valores de las sumas por fila de la matriz m. Adentro del \n# loop se utilizó i en dos ocasiones: \n# (1) para indicar el número de fila de m al cual se debía aplicar la \n# función sum (\"sum(m[i,])\") y \n# (2) para indicar el iésimo valor de s, en donde se almacenó el resultado de \n# la operación anterior (\"s[i] <- ...\").\n\n# Nota: el vector s es idéntico al obtenido con el comando:\nrowSums(m)\n\n# Nota 2: es un error muy común definir s *adentro* del loop, en lugar de \n# *afuera*. Al cometer este error le estamos indicando a R que s se debe \n# *reiniciar* en cada iteración, de forma que borra todo lo hecho en \n# iteraciones anteriores. Veamos esto en un ejemplo:\nm <- matrix(rpois(12, 8), 4, 3) # Una matriz de 4x3\n# Loop:\nfor (i in 1:4) {\n s <- numeric(4) # Un vector de longitud 4, valores == 0\n s[i] <- sum(m[i,])\n} # Se agregan las llaves porque hay más de 1 comando\ns\n\n# Como se muestra en la observación anterior (s == rowSums(m) para el loop \n# correcto), no siempre es necesario hacer un loop para hacer estas operaciones \n# (aquí rowSums puede hacer lo mismo). En general en R es conveniente saber \n# cuándo podemos evitar hacer un loop, no sólo para \"simplificar\" la sintaxis, \n# si no porque puede haber una pérdida de eficiencia que, en caso de hacer \n# cálculos pesados, se paga con mucho tiempo perdido. En general el término que \n# refiere a optar por alternativas a los loops es \"vectorización\" (ver lección \n# 6.2-loop-for-extra.r por más detalles relativos a la vectorización a la \n# eficiencia del uso de recursos). Nótese que la vectorización no es posible en \n# todos los lenguajes de programación y que en lenguajes más tradicionales (como \n# pueden ser C o FORTRAN) es necesario usar loops para las operaciones más \n# simples (como, por ejemplo, sumar dos vectores de igual longitud).\n\n# Hay casos en los que usar loops en R es sencillamente inevitable. Vamos a ver \n# ahora uno de estos casos.\n\n## EJEMPLO: SERIE DE FIBONACCI\n\n# La serie de Fibonacci se puede definir de la siguiente manera: se trata de una \n# secuencia de valores que cumplen la regla general:\n# F_{i} == F_{i - 1} + F_{i - 2}\n\n# (Aquí F_{n} es el enésimo elemento de la serie F.)\n\n# Esto implica que es necesario definir de manera arbitraria los dos primeros \n# valores (F_{1} y F_{2}) para poder definir el resto de la serie. Si los \n# primeros valores son 0 y 1 entonces la serie es (hasta el octavo valor):\n# F = [0, 1, 1, 2, 3, 5, 8, 13]\n\n# Vamos entonces a crear un loop en R que permita calcular la serie de Fibonacci \n# hasta el enésimo valor. Crearemos la función fibo para este objetivo:\nfibo <- function(n = 20, inicio = 0:1) { \n # n e inicio tienen valores por defecto, por comodidad\n\n}\n# ... por ahora la función no hace nada, veamos cómo debemos completarla ...\n\n# Traduciendo la regla general a R, nos quedaría algo así:\nout[i] <- out[i - 1] + out[i - 2]\n# donde out es un vector en el cual vamos a almacenar todos los valores de la \n# serie. Nótese que esto implica que i no puede ser menor que 3; si i fuera 2, \n# por ejemplo, entonces out[i - 2] sería out[0], lo cual es un valor no definido \n# en R. A su vez, esto implica que out tiene al menos i elementos y que out[1] y \n# out[2] están definidos de antemano.\n\n# Estas condicionantes nos llevan a definir lo que llamamos la Preparación del \n# loop. Por un lado, out tiene que existir y tener al menos 3 elementos, pero de \n# hecho ya decidimos que tendrá n elementos. El valor n está definido como uno \n# de los argumentos de la función, así que no lo definiremos de vuelta. Por \n# otro lado, los dos primeros valores de out se definen con el otro argumento: \n# inicio. De esta forma la preparación sería algo así (no ejecutar):\nout <- numeric(n)\nout[1:2] <- inicio\n\n# Como puede confirmar, si asigna valores a n y a inicio (inicio debe ser un \n# vector de 2 elementos), out ya es un vector viable para aplicar la fórmula \n# general. Pero para eso tenemos que agregar el loop en sí (no ejecutar):\nfor (i in 3:n) {\n out[i] <- out[i - 1] + out[i - 2]\n}\n\n# Es imporante notar que el rango de valores de i es 3:n; 3 porque ya vimos que \n# i no puede ser menor y n porque es la cantidad de elementos de out.\n\n# Tenemos entonces todo para definir correctamente la función fibo:\nfibo <- function(n = 20, inicio = 0:1) {\n # Preparación:\n out <- numeric(n)\n out[1:2] <- inicio\n # Loop:\n for (i in 3:n) {\n out[i] <- out[i - 1] + out[i - 2]\n }\n out # La salida de la función\n}\n\n# Podemos comprobar de que fibo hace correctamente su tarea:\nfibo()\nfibo(22)\n\n# Es importante destacar los puntos claves a tener en cuenta en el proceso de \n# crear fibo:\n# 1. El vector out debe estar correctamente definido antes de empezar el loop. \n# Generalmente la Preparación es necesaria para aprontar vectores como out, \n# en el cual se almacenan los valores que nos interesan.\n# 2. La variable i debe estar contenida en el rango correcto de valores, en \n# este caso 3:n. Este es un punto en el cual es frecuente comenter errores.\n\n# Pero ¿por qué es que es inevitable usar un loop para este problema?\n# Se puede responder con un ejemplo: ¿qué ocurre al tratar de calcular el \n# cuarto elemento de out? Para obtener este valor es necesario primero definir \n# el tercer elemento de out, ya que out[4] = out[3] + out[2]. El mismo problema \n# surgirá cuando se trate de obtener out[5] y los demás valores, por lo que no \n# queda opción que \"ir calculando de a uno\" los valores de la serie de \n# Fibonacci. En general es fácil ver que siempre que los valores de una \n# secuencia dependen de los anteriores es necesario hacer un loop para poder \n# calcularlos. Dicho de otra forma, los loops son inevitables para calcular \n# secuencias definidas por Relaciones de Recurrencia \n# (https://es.wikipedia.org/wiki/Relaci%C3%B3n_de_recurrencia).\n# Nota: muchas veces estas relaciones de recurrencia también tienen soluciones\n# cerradas, lo que implica que en esos casos tampoco es estrictamente necesario\n# usar un loop (en R) para calcular los valores de la sucesión.\n\n## LOOPS FOR ANIDADOS:\n\n# Nada impide que adentro de un for (o un while) se escriba otro loop. Esto es \n# lo que llamamos un loop anidado. Las reglas que siguen son las mismas que antes, \n# pero hay que tener la precaución de utilizar diferentes nombres para las \n# variables de iteración. En el siguiente ejemplo se usan i y j como variables:\n\nm <- matrix(rpois(4, 8), 2, 2) # Una matriz de 2x2\nfor (i in 1:nrow(m)) {\n for (j in 1:ncol(m)) {\n print(m[i, j])\n }\n}\n\n# Como fácilmente se puede comprobar, este código simplemente imprime todos los\n# elementos de la matriz m. Nótese: que la variable i se usa para las filas y \n# la variable j para las columnas. \n\n\n## RESUMEN\n# Hemos visto el tipo de loop más comunmente utilizado en programación. El loop\n# 'for' se utiliza cuando \"sabemos\" de antemano el número de iteraciones que\n# necesitamos.\n# Dado que la sintaxis es relativamente elaborada, hay que tener especial\n# atención a los errores, particularmente los más comunes: omisión de una llave,\n# escribir mál la secuencia de números (1:n) o confundir el nombre de las\n# variables (i, j, k, etc ...).\n# En general siempre va a ser necesario tener una Preparación, es decir líneas\n# de código anteriores al loop en las que se definen correctamente uno o varios\n# objetos que serán utilizados dentro del mismo. Un error común es definir \n# alguno de estos objetos (como el out en la función fibo) *adentro* del loop,\n# cuando es necesario que esto se haga **afuera**, en la preparación.\n# Los ejemplos empleados dan cuenta de situaciones comunes en que podemos\n# utilizar esta herramienta, pero no está limitada a éstas. Siendo particular-\n# mente útil al realizar simulaciones con modelos o aplicar análisis a grandes\n# series de datos, su combinación dentro de funciones y con otras estructuras\n# de control (ej: condicionales) se hace más importante, y resulta fundamental\n# como herramienta para cualquier programador.\n\n", "meta": {"hexsha": "59c884d9458b37fc88e6699a918807ec79df99ba", "size": 11685, "ext": "r", "lang": "R", "max_stars_repo_path": "CODIGO CHURN/CODIGOS_UTILIZADOS/lecciones/6.2-loop-for.r", "max_stars_repo_name": "jcombari/Data-products-Course-Project", "max_stars_repo_head_hexsha": "bd3bc260d2bf3def34c8fc94bf847ca0482e7549", "max_stars_repo_licenses": ["FTL"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "CODIGO CHURN/CODIGOS_UTILIZADOS/lecciones/6.2-loop-for.r", "max_issues_repo_name": "jcombari/Data-products-Course-Project", "max_issues_repo_head_hexsha": "bd3bc260d2bf3def34c8fc94bf847ca0482e7549", "max_issues_repo_licenses": ["FTL"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "CODIGO CHURN/CODIGOS_UTILIZADOS/lecciones/6.2-loop-for.r", "max_forks_repo_name": "jcombari/Data-products-Course-Project", "max_forks_repo_head_hexsha": "bd3bc260d2bf3def34c8fc94bf847ca0482e7549", "max_forks_repo_licenses": ["FTL"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 43.2777777778, "max_line_length": 83, "alphanum_fraction": 0.7268292683, "num_tokens": 3439, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6334102636778401, "lm_q2_score": 0.8244619220634456, "lm_q1q2_score": 0.522222643446546}} {"text": "model_gaimean <- function (gAI = 0.0,\n tTWindowForPTQ = 0.0,\n deltaTT = 0.0,\n pastMaxAI_t1 = 0.0,\n listTTShootWindowForPTQ1_t1 = c(0.0),\n listGAITTWindowForPTQ_t1 = c(0.0)){\n #'- Name: GAImean -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: Average GAI on a specific thermal time window\n #' * Author: Loïc Manceau\n #' * Reference: -\n #' * Institution: INRA\n #' * Abstract: -\n #'- inputs:\n #' * name: gAI\n #' ** description : Green Area Index of the day\n #' ** inputtype : variable\n #' ** variablecategory : auxiliary\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 500.0\n #' ** unit : m2 leaf m-2 ground\n #' ** uri : \n #' * name: tTWindowForPTQ\n #' ** description : Thermal Time window for average\n #' ** inputtype : parameter\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 5000.0\n #' ** unit : °C d\n #' ** uri : \n #' * name: deltaTT\n #' ** description : Thermal time increase of the day\n #' ** inputtype : variable\n #' ** variablecategory : auxiliary\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 100.0\n #' ** unit : °C d\n #' ** uri : \n #' * name: pastMaxAI_t1\n #' ** description : Maximum Leaf Area Index reached the current day\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 5000.0\n #' ** unit : m2 leaf m-2 ground\n #' ** uri : \n #' * name: listTTShootWindowForPTQ1_t1\n #' ** description : List of daily shoot thermal time in the window dedicated to average\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLELIST\n #' ** default : [0.0]\n #' ** min : \n #' ** max : \n #' ** unit : °C d\n #' ** uri : \n #' * name: listGAITTWindowForPTQ_t1\n #' ** description : List of daily Green Area Index in the window dedicated to average\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLELIST\n #' ** default : [0.0]\n #' ** min : \n #' ** max : \n #' ** unit : m2 leaf m-2 ground\n #' ** uri : \n #'- outputs:\n #' * name: gAImean\n #' ** description : Mean Green Area Index\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** min : 0.0\n #' ** max : 500.0\n #' ** unit : m2 leaf m-2 ground\n #' ** uri : \n #' * name: pastMaxAI\n #' ** description : Maximum Leaf Area Index reached the current day\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** min : 0.0\n #' ** max : 5000.0\n #' ** unit : m2 leaf m-2 ground\n #' ** uri : \n #' * name: listTTShootWindowForPTQ1\n #' ** description : List of daily shoot thermal time in the window dedicated to average\n #' ** variablecategory : state\n #' ** datatype : DOUBLELIST\n #' ** min : \n #' ** max : \n #' ** unit : °C d\n #' ** uri : \n #' * name: listGAITTWindowForPTQ\n #' ** description : List of daily Green Area Index in the window dedicated to average\n #' ** variablecategory : state\n #' ** datatype : DOUBLELIST\n #' ** min : \n #' ** max : \n #' ** unit : m2 leaf m-2 ground\n #' ** uri : \n listTTShootWindowForPTQ1 <- vector()\n listGAITTWindowForPTQ <- vector()\n TTList <- vector()\n GAIList <- vector()\n count <- 0\n gai_ <- 0.0\n gaiMean_ <- 0.0\n countGaiMean <- 0\n for( i in seq(0, length(listTTShootWindowForPTQ1_t1)-1, 1)){\n TTList <- c(TTList, listTTShootWindowForPTQ1_t1[i+1])\n GAIList <- c(GAIList, listGAITTWindowForPTQ_t1[i+1])\n }\n TTList <- c(TTList, deltaTT)\n GAIList <- c(GAIList, gAI)\n SumTT <- sum(TTList)\n while( SumTT > tTWindowForPTQ){\n SumTT <- SumTT - TTList[count+1]\n count <- count + 1}\n for( i in seq(count, length(TTList)-1, 1)){\n listTTShootWindowForPTQ1 <- c(listTTShootWindowForPTQ1, TTList[i+1])\n listGAITTWindowForPTQ <- c(listGAITTWindowForPTQ, GAIList[i+1])\n }\n for( i in seq(0, length(listGAITTWindowForPTQ)-1, 1)){\n gaiMean_ <- gaiMean_ + listGAITTWindowForPTQ[i+1]\n countGaiMean <- countGaiMean + 1\n }\n gaiMean_ <- gaiMean_ / countGaiMean\n gai_ <- max(pastMaxAI_t1, gaiMean_)\n pastMaxAI <- gai_\n gAImean <- gai_\n return (list (\"gAImean\" = gAImean,\"pastMaxAI\" = pastMaxAI,\"listTTShootWindowForPTQ1\" = listTTShootWindowForPTQ1,\"listGAITTWindowForPTQ\" = listGAITTWindowForPTQ))\n}", "meta": {"hexsha": "262415271ee8fb0a661a6dcd2e1ec1dd6d4f8430", "size": 6824, "ext": "r", "lang": "R", "max_stars_repo_path": "src/r/SQ_Wheat_Phenology/Gaimean.r", "max_stars_repo_name": "Crop2ML-Catalog/SQ_Wheat_Phenology", "max_stars_repo_head_hexsha": "8e9ec229e5f0754d0f4b9d79ac96a084d75dde67", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-06-21T18:58:04.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-29T21:32:28.000Z", "max_issues_repo_path": "src/r/SQ_Wheat_Phenology/Gaimean.r", "max_issues_repo_name": "Crop2ML-Catalog/SQ_Wheat_Phenology", "max_issues_repo_head_hexsha": "8e9ec229e5f0754d0f4b9d79ac96a084d75dde67", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 27, "max_issues_repo_issues_event_min_datetime": "2018-12-04T15:35:44.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-11T08:25:03.000Z", "max_forks_repo_path": "src/r/SQ_Wheat_Phenology/Gaimean.r", "max_forks_repo_name": "Crop2ML-Catalog/SQ_Wheat_Phenology", "max_forks_repo_head_hexsha": "8e9ec229e5f0754d0f4b9d79ac96a084d75dde67", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 7, "max_forks_repo_forks_event_min_datetime": "2019-04-20T02:25:22.000Z", "max_forks_repo_forks_event_max_datetime": "2021-11-04T07:52:35.000Z", "avg_line_length": 49.0935251799, "max_line_length": 165, "alphanum_fraction": 0.3751465416, "num_tokens": 1593, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104788995148792, "lm_q2_score": 0.6442251064863697, "lm_q1q2_score": 0.5221308553449288}} {"text": "#Solve for optimal lobby effort under DGH97-style model with obj fcn W + e\n\n#reserve space for loop output\ntau = seq(0.001,.166,0.001) #this will be counter variable in loop\nPSx = matrix(NA,length(tau),1)\nCSx = matrix(NA,length(tau),1)\nTR = matrix(NA,length(tau),1)\nPSy = matrix(NA,length(tau),1)\nCSy = matrix(NA,length(tau),1)\n\n#calculate producer surplus, consumer surplus, tariff revenue for each possible\n#value of the tariff on the grid (just above zero to prohibitive tariff 1/6)\nfor (j in 1:length(tau)) {\n t = tau[j]\n\n PSx[j] = ((2 +2*t)^2)/49\n CSx[j] = .5*((3 -4*t)^2)/49\n TR[j] = (t - 6*t^2)/7\n CSy[j] = ((3 +3*t)^2)/98\n PSy[j] = ((4 -3*t)^2)/98\n}\n\n#Calculations for when lobby has all the bargaining power\n\n#calculate government welfare when tau = 0 (baseline)\nb = ((2 +2*0)^2)/49 + .5*((3 -4*0)^2)/49 + (0 - 6*0^2)/7 + ((3 +3*0)^2)/98 + ((4 -3*0)^2)/98\nW = PSx + CSx + TR + CSy + PSy #social welfare\ne = ((b - W)/PSx)^5 #gov't indifference condition when WG = W + e\npi = PSx - e #net profits\n\nvalue = max(pi) #the value at which profits are maximized (over non-negative values)\nind = which.max(pi) #the location at which profits are maximized\nRC = arrayInd(ind,c(dim(pi),dim(pi))) #row/column version of maximand location\n\n\n#Calculations for when government has all the bargaining power\nbl = ((2 +2*0)^2)/49 #baseline for lobby: profits when tau = 0\ne = PSx - bl #effort level giving all excess profits over tau=0 to gov't\n\ng = e^1.2 #little g(e) function to add to social welfare\nG = W + g #gov't welfare a la DGH97\nplot(G)\n\nvalue = max(G) #the value at which profits are maximized (over non-negative values)\nind = which.max(G) #the location at which profits are maximized\nRC = arrayInd(ind,c(dim(G),dim(G))) #row/column version of maximand location", "meta": {"hexsha": "4eb0d5c212881ae72d30eb8abce04ee54b27d508", "size": 1871, "ext": "r", "lang": "R", "max_stars_repo_path": "DGH.r", "max_stars_repo_name": "kbuzard/SOP_repeated", "max_stars_repo_head_hexsha": "9faf5cdb76e8a1f404823ab8a619012dcbe1d582", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2016-09-09T21:55:37.000Z", "max_stars_repo_stars_event_max_datetime": "2016-09-09T21:55:37.000Z", "max_issues_repo_path": "DGH.r", "max_issues_repo_name": "kbuzard/SOP_repeated", "max_issues_repo_head_hexsha": "9faf5cdb76e8a1f404823ab8a619012dcbe1d582", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 27, "max_issues_repo_issues_event_min_datetime": "2015-01-14T15:27:57.000Z", "max_issues_repo_issues_event_max_datetime": "2017-06-29T18:33:17.000Z", "max_forks_repo_path": "DGH.r", "max_forks_repo_name": "kbuzard/SOP_repeated", "max_forks_repo_head_hexsha": "9faf5cdb76e8a1f404823ab8a619012dcbe1d582", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 40.6739130435, "max_line_length": 92, "alphanum_fraction": 0.6360235168, "num_tokens": 596, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8918110483133801, "lm_q2_score": 0.5851011542032312, "lm_q1q2_score": 0.5217996736993522}} {"text": "# CLUSTER GLEDE NA ZADOVOLJSTVO PREBIVALCEV, ZAPOSLENOST, BDP TER STEVILO PREBIVALCEV\n\ncluster_evropa <- function(){\n evropa1 <- uvozi.svet() %>% filter (continent == 'Europe')\n zadovoljstvo <- uvozi.rating() %>% filter(leto == 2018)\n brezposelnost_cluster <- uvozi.zaposlenost()\n bdp <- uvozi.BDP() %>% filter(leto == 2018) %>% select('Drzava', 'BDP per capita')\n evropa2 <- merge(x = evropa1, y = zadovoljstvo, by.y = 'Drzava', by.x = 'name')\n evropa3 <- merge(x = evropa2, y = brezposelnost_cluster, by.y = 'name', by.x = 'name')\n evropa4 <- merge(x = evropa3, y = bdp, by.y = 'Drzava', by.x = 'name')\n podatki_cluster <- evropa4 %>% filter (leto == 2018) %>% select('name', 'pop_est','BDP per capita', 'Ocena','Brezposelnost')\n podatki_cluster2 <- podatki_cluster\n podatki_cluster$geometry = NULL\n cluster <- podatki_cluster %>% select(-name) %>% scale()\n rownames(cluster) <- podatki_cluster$name\n k <- kmeans(cluster, 5, nstart=1000)\n skupine <- data.frame(name=podatki_cluster2$name, skupina=factor(k$cluster))\n podatki_za_risat <- merge(podatki_cluster2, skupine, by ='name')\n zemljevid <- tm_shape(podatki_za_risat) + tm_polygons('skupina')\n zemljevid\n tmap_mode('view')\n return(zemljevid)\n}\ncluster <- cluster_evropa()\n\n\n# GRAF KI PRIKAZUJE GIBANJE REALNEGA BDP SLOVENIJE TER EVROPE NA PREBIVALCA IN NAPOVEDUJE PRIHODNOST\n\nnapovedovanje_BDP <- function(){\n BDP <- uvozi.BDP()\n bdp_po_letih_eu <- aggregate(BDP$'BDP per capita', by=list(leto=BDP$leto), FUN=mean)\n bdp_po_letih_eu$x <- as.numeric(bdp_po_letih_eu$x)\n bdp_po_letih_eu$leto <- as.numeric(bdp_po_letih_eu$leto)\n bdp_po_letih_eu$z <- 'evropa'\n \n bdp_po_letih_slo <- BDP %>% filter(Drzava == 'Slovenia') %>% select('leto', 'BDP per capita')\n bdp_po_letih_slo <- bdp_po_letih_slo %>% rename(x = 'BDP per capita')\n bdp_po_letih_slo$x <- as.numeric(bdp_po_letih_slo$x)\n bdp_po_letih_slo$leto <- as.numeric(bdp_po_letih_slo$leto)\n bdp_po_letih_slo$z <- 'slovenija'\n \n bdp_po_letih <- rbind(bdp_po_letih_eu, bdp_po_letih_slo)\n colnames(bdp_po_letih)[3] <- \"Legenda\"\n \n h <- ggplot(bdp_po_letih, aes(leto, x, shape=Legenda, colour=Legenda, fill=Legenda)) +\n geom_smooth(method=\"lm\", fullrange=TRUE) +\n geom_point(size=3) +\n fte_theme() +\n xlab(\"Leto\") +\n ylab(\"€\") +\n ggtitle(\"Gibanje realnega BDP v € na prebivalca\") +\n scale_x_continuous(breaks = 2013:2022) +\n expand_limits(x = 2020) +\n geom_hline(yintercept=0, size=0.4, color=\"black\") +\n theme(legend.position=\"right\")\n return(h)\n}\nnapovedovanje <- napovedovanje_BDP()\n", "meta": {"hexsha": "a42c6f1294e8da462ffab318dbeed9715d11b744", "size": 2593, "ext": "r", "lang": "R", "max_stars_repo_path": "analiza/analiza.r", "max_stars_repo_name": "BulaRebula/APPR-2019-20", "max_stars_repo_head_hexsha": "5977dfa4e86173ca0650906065ff2e9bc2a09563", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2020-03-01T11:20:57.000Z", "max_stars_repo_stars_event_max_datetime": "2020-03-01T11:20:57.000Z", "max_issues_repo_path": "analiza/analiza.r", "max_issues_repo_name": "BulaRebula/APPR-2019-20", "max_issues_repo_head_hexsha": "5977dfa4e86173ca0650906065ff2e9bc2a09563", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 6, "max_issues_repo_issues_event_min_datetime": "2019-12-15T15:58:06.000Z", "max_issues_repo_issues_event_max_datetime": "2020-03-23T17:59:16.000Z", "max_forks_repo_path": "analiza/analiza.r", "max_forks_repo_name": "BulaRebula/APPR-2019-20", "max_forks_repo_head_hexsha": "5977dfa4e86173ca0650906065ff2e9bc2a09563", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 43.9491525424, "max_line_length": 126, "alphanum_fraction": 0.6837639799, "num_tokens": 978, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929002541068, "lm_q2_score": 0.6688802603710086, "lm_q1q2_score": 0.5217218542095051}} {"text": "## Exercicio 1\r\n## Prof. James Hunter\r\n## from: https://rstudio.cloud/project/1177204\r\n## 11 de maio de 2020\r\n\r\n\r\n# 1 - Fazer o cálculo\r\n((42 * 95^2) + 6)/(16 - 3.5)\r\n\r\n# 2 - Atribuir à variável calc\r\ncalc <- ((42 * 95^2) + 6)/(16 - 3.5)\r\n\r\n# 3 - arrondar a 1 casa decimal\r\nround(calc, digits = 1)\r\n", "meta": {"hexsha": "560a8ab3bc6287ccb2972143bfbfa91fcff2a2af", "size": 299, "ext": "r", "lang": "R", "max_stars_repo_path": "exercicio_1.r", "max_stars_repo_name": "jameshunterbr/Sustentare_MAD_2020", "max_stars_repo_head_hexsha": "299a7f0af7e999e59cc53c57cf22618b6eb68092", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "exercicio_1.r", "max_issues_repo_name": "jameshunterbr/Sustentare_MAD_2020", "max_issues_repo_head_hexsha": "299a7f0af7e999e59cc53c57cf22618b6eb68092", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "exercicio_1.r", "max_forks_repo_name": "jameshunterbr/Sustentare_MAD_2020", "max_forks_repo_head_hexsha": "299a7f0af7e999e59cc53c57cf22618b6eb68092", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 19.9333333333, "max_line_length": 47, "alphanum_fraction": 0.5752508361, "num_tokens": 121, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929002541068, "lm_q2_score": 0.6688802537704063, "lm_q1q2_score": 0.5217218490610822}} {"text": "# This is adapted from the code in http://goo.gl/zw2eeq\n# It has been edited for clarity and style, and comments have been added.\n# This version is for math only\n\n# Arguments:\n# item_parameters_file: a csv of items and their parameters that\n# we'll optimize over\n# content_minimums: A list with the minimum values for each content category\n# The content categories should be named 1 - n, with each item in\n# content_minimums corresponding to the corresponding category.\n# num_forms: an int representing the total number of forms we're creating\n# items_per_form: an int representing the number of items per form\nsolver <- function(item_parameters_file, content_minumums, num_forms, n){\n\n # Constants\n # These are the ability values for which we care about maximizing the TIF\n theta = c(-1.5,0,1.5);\n # These are target information levels for the three ability levels we're\n # optimizing for in our optimization function\n d_theta = c(5.4, 10, 5.4);\n\n # Manually set constants end here!\n\n # Import the lpsolve library\n library(lpSolveAPI);\n\n # Read in the data. This should be of the form item,a,b,c,category\n # This example involves IRT for the Test Information Function\n s = read.csv(item_parameters_file);\n\n # The 3PL model for items has these three variables, a, b, and c. Learn more\n # on wikipedia: http://en.wikipedia.org/wiki/Item_response_theory\n\n # a is discrimination\n a = s$a;\n\n # b is difficulty\n b = s$b;\n\n # c is a psudo-guessing offset\n c = s$b;\n\n # Content is the different categories - like 1, 2, 3, 4, 5, 6...\n content = s$Content;\n\n # i is just the total number of items\n num_items = nrow(s);\n\n # This is the number of content categories. We assume they're labeled\n # 1-num_categories for the sake of simplicity\n num_categories = length(unique(content));\n\n # This is where we sort all of our items by content.\n content_items=list();\n for(k in 1:num_categories){\n content_items[[k]] = c(1:num_items)[content==k];\n }\n\n # This i x j array contains the amount of information item i gives us\n # about theta value j\n j=length(theta);\n info = array(0,c(num_items, j));\n\n # This calculates the information at each value\n # http://en.wikipedia.org/wiki/Item_response_theory#Three_parameter_logistic_model\n # euler's constant, e, can be represented as exp(1)\n e = exp(1)\n for (jj in 1:j){\n # P is the probability of answering correctly\n p = c + (1-c) / (1 + exp(-e * a * (theta[jj]-b)));\n q = 1 - p;\n\n # src:de Ayala, R.J. (2009). The Theory and Practice of Item Response\n # Theory, New York, NY: The Guilford Press. (6.12), p.144\n info[,jj] = (e^2) * (a^2) * ((p - c)/(1 - c))^2 * q / p;\n }\n\n # m is the total number of decision variables. We're deciding whether each item\n # is in each form, so we have i * f variables. (TODO: What is the last one)\n m = num_items * num_forms + 1;\n\n # CONFIGURE THE LINEAR PROGRAM\n\n # Here's our linear program object\n # We start with 0 rows and m columns. Rows are added as constraints are added.\n lprec = make.lp(0, m);\n\n # The sense control means we're minimizing the objective function\n # epsint is the tolerance to decide if a floating point number is an integer\n # The mip gap has to do with the algorithm used and what solution branches to\n # ignore to speed things up.\n # See http://lpsolve.sourceforge.net/5.5/set_mip_gap.htm\n lp.control(lprec, sense=\"min\", epsint=0.1, mip.gap = c(0.1,0.05));\n\n # All of the first i * f columns are binary, as they represent whether each\n # item is in each form\n set.type(lprec, columns=c(1:(num_forms * num_items)), type=\"binary\");\n\n # The last decision variable is real\n set.type(lprec, columns=m, type=\"real\");\n\n # Each variable must range between 0 and 1\n set.bounds(lprec, lower=rep(0, m), upper=rep(1, m));\n\n # ADD CONSTRAINTS TO THE LINEAR PROGRAM\n\n # This adds a constraint that each item can only exist in one form.\n # This actually will only work for two forms max.\n # It's basically guaranteeing that an array of ones times an array of\n # whether item i appears in form f for each i and all f is <= 1.\n for (i in 1:num_items){\n\n idxs = c();\n for (f in 1:num_forms){\n item_offset = num_items * (f - 1);\n idxs = c(idxs, i + item_offset);\n }\n add.constraint(lprec, rep(1, num_forms), \"<=\", 1, indices=idxs);\n }\n\n # Add constraints on the amount of content in each category\n for (f in 1:num_forms)\n {\n item_offset = num_items * (f - 1);\n for (k in 1:length(content_minumums)) {\n add.constraint(\n lprec,\n rep(1, length(content_items[[k]])), \">=\", content_minumums[k],\n indices=item_offset + content_items[[k]]);\n }\n }\n\n # Ensure there are exactly n items in each form\n for (f in 1:num_forms)\n {\n first_item = num_items * (f - 1) + 1;\n last_item = num_items * f;\n add.constraint(lprec,\n rep(1, num_items), \"=\", n,\n indices=first_item:last_item);\n }\n\n # Add constraints on the information level\n for (f in 1:num_forms)\n {\n first_item = num_items * (f - 1) + 1;\n last_item = num_items * f;\n\n # Add a constraint for each theta level\n for (k in 1:length(d_theta)) {\n add.constraint(lprec,\n c(info[,k], -1), \"<=\", d_theta[k],\n indices=c(first_item:last_item,m));\n add.constraint(lprec,\n c(info[,k], 1), \">=\", d_theta[k],\n indices=c(first_item:last_item,m));\n\n }\n }\n\n # Set the objective function\n set.objfn(lprec, 1, indices=m);\n\n res_flag = solve(lprec);\n\n res_flag;\n\n # Now x_opt contains the assignments of questions to forms\n # TODO: Print out an understandable list of Form #, Item #\n x_opt = get.variables(lprec);\n\n return(x_opt);\n}\n", "meta": {"hexsha": "e9b54e6007d75fe70601894629a1af27ffff1a05", "size": 6097, "ext": "r", "lang": "R", "max_stars_repo_path": "lpsolve/diao_ex_1.r", "max_stars_repo_name": "elifeasley/ProjectLearnSquared", "max_stars_repo_head_hexsha": "397f8c2478925c9b6d24ed9371cb291736f658fb", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "lpsolve/diao_ex_1.r", "max_issues_repo_name": "elifeasley/ProjectLearnSquared", "max_issues_repo_head_hexsha": "397f8c2478925c9b6d24ed9371cb291736f658fb", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "lpsolve/diao_ex_1.r", "max_forks_repo_name": "elifeasley/ProjectLearnSquared", "max_forks_repo_head_hexsha": "397f8c2478925c9b6d24ed9371cb291736f658fb", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.0402298851, "max_line_length": 86, "alphanum_fraction": 0.6242414302, "num_tokens": 1601, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8479677660619633, "lm_q2_score": 0.6150878555160665, "lm_q1q2_score": 0.5215746747738026}} {"text": "\n# coding: utf-8\n\n# # Práctico 5 - Cohesión y detección de comunidades en redes reales\n# \n# #1. Ambientes de trabajo\n# \n# ## 1.a) Ambiente COLAB remoto\n# \n# 1. Abrir en navegador: https://colab.research.google.com/\n# 2. Abrir el notebook de la tarea:\n# File-> Open Notebook -> Github -> https://github.com/prbocca/na101_master -> homework_5_communities\n# 3. Guardar el notebook en su Google Drive:\n# File -> Save a Copy in Drive... \n# 4. Renombrar el archivo `\"cedula ID\"_ar_hw5.ipynb`, por ejemplo *33484022_ar_hw5.ipynb*\n# 5. Al final usted deberá descargar el notebook. Asegurarse que se están guardando las salidas de ejecución en el notebook: File -> Download .ipynb\n# 6. Luego estos archivos deberán ser enviados a prbocca@fing.edu.uy \n# \n# ##1.b) Ambiente RSTUDIO local (opcional)\n# \n# Abrir el .r de la tarea en: https://github.com/prbocca/na101_master/tree/master/homework_5_communities\n\n# ## 1.c) Cargar Librerias\n# \n# Todas las librerías debe instalarse correctamente, si el proceso se interrumpe o alguna librería da error en la instalación, entonces habrá problemas en el código más adelante. Si tiene este tipo de problemas pruebe hacer `Runtime -> Factory reset runtime`, y volver a intentar.\n# \n\n# In[ ]:\n\n# cargar librerias\nload_libs <- function(libraries = libs, install=TRUE){\n if (install){ # instalar librerias no instaladas\n new.packages <- libs[!(libs %in% installed.packages()[,\"Package\"])]\n if(length(new.packages)) install.packages(new.packages)\n }\n #cargo librerias \n for (lib in libraries){\n require(lib, character.only=TRUE, quietly = FALSE)\n } \n} \n\nlibs = c(\"network\", \"sna\",\"rgexf\",\"ape\", \n \"R.matlab\", \"sand\",\"igraph\",\"igraphdata\") \n\nload_libs(libs)\n\n\n# ## 1.d) Descargar funciones auxiliares\n\n# In[ ]:\n\n#directorio donde se va a trabajar\ndata_path = \"/content/ar/hw5/\"\n\ndir.create(data_path, showWarnings = FALSE, recursive = TRUE)\nsetwd(data_path)\ngetwd()\nlist.files()\n\n# cargo funciones auxiliares\nsource(\"https://raw.githubusercontent.com/prbocca/na101_master/master/homeworks_common.r\")\n\n\n# # 2. Cohesión en grafos en `R`\n# \n# Seguir la Secciones 4.3 del libro [SANDR], ejecutando el código fuente incluido.\n\n# In[ ]:\n\n#4.3 Characterizing Network Cohesion\ndata(karate)\ntable(sapply(cliques(karate), length))\n# 1 2 3 4 5\n# 34 78 45 11 2\n# there are 34 nodes (cliques of size one) and 78 edges (cliques of size two), followed by 45 triangles (cliques of size three)\n#\n# the largest cliques are of size five, of which there are only two. \n# These two both involve four actors in common, including Mr Hi, the head instructor.\ncliques(karate)[sapply(cliques(karate), length) == 5]\n# [[1]]\n# + 5/34 vertices, named, from 4b458a1:\n# [1] Mr Hi Actor 2 Actor 3 Actor 4 Actor 14\n# \n# [[2]]\n# + 5/34 vertices, named, from 4b458a1:\n# [1] Mr Hi Actor 2 Actor 3 Actor 4 Actor 8\n#\ntable(sapply(maximal.cliques(karate), length))\n# 2 3 4 5 \n# 11 21 2 2 \n# In the karate network, the two largest cliques (formally called maximum cliques) are maximal, while, for example, the same can\n# be said of only two of the 11 cliques of size four.\n\n\n# In[ ]:\n\n# representacion en k-core\n# esta es una forma relativamente común para visualizar grafos muy grandes (>10k nodos)\ncores <- graph.coreness(karate)\nA <- get.adjacency(karate, sparse=FALSE)\nlibrary(network)\ng <- network::as.network.matrix(A)\nsna::gplot.target(g, cores, \n #circ.lab = FALSE,\n circ.col=\"skyblue\", usearrows = FALSE,\n vertex.col=cores, edge.col=\"darkgray\")\n#detach(\"package:sna\")\n#detach(\"package:network\")\n\n\n# In[ ]:\n\n# 4.3.2 Density and Related Notions of Relative Frequency\n# we see that the sub-graphs corresponding to each of the instructor and the administrator, in union with\n# their immediate respective neighborhoods—i.e., the ego-centric networks around\n# vertices 1 and 34—are noticeably more dense than the overall network.\nego.instr <- induced.subgraph(karate, neighborhood(karate, 1, 1)[[1]])\nego.admin <- induced.subgraph(karate, neighborhood(karate, 1, 34)[[1]])\ngraph.density(karate) #0.1390374\ngraph.density(ego.instr) #0.25\ngraph.density(ego.admin) #0.2091503\n#\n# global clustering, summarizing the relative frequency with which connected triples close to form triangles\ntransitivity(karate) # 0.2556818\n# In the case of the instructor and administrator of the karate network, for\n# example, we see that their local clustering is only 50–60 % that of the clustering for the network as a whole.\ntransitivity(karate, \"local\", vids=c(1,34)) #0.1500000 0.1102941\n#\n# reciprocity is defined as the total number of reciprocated edges divided by the total number of edges.\n# In the AIDS blog network, the reciprocity is quite low by either definition.\ndata(aidsblog)\nreciprocity(aidsblog, mode=\"default\") #[1] 0.03278689\nreciprocity(aidsblog, mode=\"ratio\") #[1] 0.01666667\n\n\n# In[ ]:\n\n# 4.3.3 Connectivity, Cuts, and Flows\ndata(yeast)\nis.connected(yeast)\n#\n#A census of all connected components within this graph, however, shows that there clearly is a giant component.\n# This single component contains 2375/2617 ≈ 90 %\ncomps <- decompose.graph(yeast)\ntable(sapply(comps, vcount))\n# 2 3 4 5 6 7 2375 \n# 63 13 5 6 1 3 1\n#\n# often attention would be restricted to this giant component alone in carrying out further analysis and modeling.\n# small world property:\nyeast.gc <- decompose.graph(yeast)[[1]]\naverage.path.length(yeast.gc) #5.09597\ndiameter(yeast.gc) #15\n# At the same time, the clustering in this network is relatively large\ntransitivity(yeast.gc)\n#\n# The vertex (edge) connectivity of G is the largest integer such that G is k-vertex- (k-edge-) connected.\n# In the case of the giant component of the yeast network, the vertex and edge connectivity are both equal to one.\nvertex.connectivity(yeast.gc) #1\nedge.connectivity(yeast.gc) #1\n# If the removal of a particular set of vertices (edges) in a graph disconnects the\n# graph, that set is called a vertex-cut (edge-cut). A single vertex that disconnects\n# the graph is called a cut vertex, or sometimes an articulation point.\n# In the giant component of the yeast network, almost 15 % of the vertices are cut vertices.\nyeast.cut.vertices <- articulation.points(yeast.gc)\nlength(yeast.cut.vertices) #[1] 350\n\n# Note that the distinction between strong and weak connectivity can be severe for some digraphs. \n# For example, the AIDS blog network is weakly connected\nis.connected(aidsblog, mode=c(\"weak\")) #2 [1] TRUE\n# but not strongly connected.\nis.connected(aidsblog, mode=c(\"strong\")) #[1] FALSE\n# And while there does exist a strongly connected component within the graph, there is only one and it has only four vertices.\naidsblog.scc <- clusters(aidsblog, mode=c(\"strong\"))\ntable(aidsblog.scc$csize)\n# 1 4 \n# 142 1 \n\n\n# # 3. Particionar grafos en `R`. \n# \n# Seguir la secciones 4.4 al 4.6 del libro [SANDR], ejecutando el código fuente incluido.\n# \n# \n\n# In[ ]:\n\n#4.4 Graph Partitioning\n#4.4.1 Hierarchical Clustering \n\nkc <- fastgreedy.community(karate)\nlength(kc) #[1] 3\nsizes(kc) \nmembership(kc)\n# Community sizes\n# 1 2 3 \n# 18 11 5 \nplot(kc, karate)\n#library(ape)\ndendPlot(kc, mode=\"phylo\")\n\n\n# In[ ]:\n\n# 4.4.2 Spectral Partitioning\nk.lap <- graph.laplacian(karate)\neig.anal <- eigen(k.lap)\nplot(eig.anal$values, col=\"blue\", ylab=\"Eigenvalues of Graph Laplacian\")\nf.vec <- eig.anal$vectors[, 33]\nfaction <- get.vertex.attribute(karate, \"Faction\")\nf.colors <- as.character(length(faction))\nf.colors[faction == 1] <- \"red\"\nf.colors[faction == 2] <- \"cyan\"\nplot(f.vec, pch=16, xlab=\"Actor Number\",\n ylab=\"Fiedler Vector Entry\", col=f.colors)\nabline(0, 0, lwd=2, col=\"lightgray\")\n\n\n# In[ ]:\n\n# 4.4.3 Validation of Graph Partitioning\nfunc.class <- get.vertex.attribute(yeast.gc, \"Class\")\ntable(func.class)\nyc <- fastgreedy.community(yeast.gc)\nc.m <- membership(yc)\ntable(c.m, func.class, useNA=c(\"no\"))\n\n\n# In[ ]:\n\n# 4.5 Assortativity and Mixing\n# The assortativity coefficient with categories\nassortativity.nominal(yeast, (V(yeast)$Class==\"P\")+1, directed=FALSE) #[1] 0.4965229\n# assortativity mixing with continuos attributes (Pearson correlation coefficient) \n\nassortativity.degree(yeast) #0.4610797\n\n\n# # 4. Particionar la red de blogs políticos de EE.UU.\n# \n# Estudiaremos la red de blogs políticos de EE.UU, \n# con el objetivo de particionarla en las dos comunidades políticas existenes: liberales (demócratas) y conservadores (republicanos).\n# Los datos son de la elección política de EE.UU. en 2004, \n# fueron recolectados por L. Adamic and N. Glance en 2005, \n# y pueden obtenerse de la colección de Mark Newman en: http://www-personal.umich.edu/~mejn/netdata/. \n# \n# En este ejercicio usaremos una versión no dirigida del grafo dirigido original,\n# donde las aristas corresponden a *hyperlinks* entre blogs. \n# La red tiene $N_v=1490$ blogs (vértices), \n# y se conoce la afiliación política de cada blogger (y por tanto de sus blogs), representada por un vector binario (0 es liberal, y 1 es conservador) para cada vértice.\n# \n\n# $\\newcommand \\ind [1] {{\\mathbb I \\left\\{#1\\right\\} } }$\n# $\\def\\bbA{{\\mathbf A}}$\n# $\\def\\bbB{{\\mathbf B}}$\n# \n# ## 4.a) Descargar y cargar los datos.\n# \n# Descargar los datos en formato `Matlab` de la blogósfera política en el archivo `political_blogs.mat` de la página del curso, o en el siguiente link: \n# https://github.com/prbocca/na101_master/raw/master/homework_5_communities/political_blogs.mat.\n# \n# Cargar los datos en `R`, que incluyen la matriz de adyacencia $\\bbA \\in \\{0,1\\}^{1490\\times 1490}$, \n# y el vector ${\\tt nodes}\\in\\{0,1\\}^{1490}$ que contiene la afiliación política de cada blog.\n\n# In[ ]:\n\n# download data\ndownload.file(url=\"https://github.com/prbocca/na101_master/raw/master/homework_5_communities/political_blogs.mat\", destfile=\"political_blogs.mat\", mode=\"wb\")\nlist.files()\n\n# cargar datos\nA = NA\nnodes = NA\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: cargar la matriz de adyacencia 'A', y el vector con datos de los nodos 'nodes'\n#\n#\n#\n##################################################################\nstr(A)\nstr(nodes)\n\n\n# Crear un grafo no dirigido a partir de los datos (matriz de adyacencia y afiliación). El resultaod debe ser similar al de la siguiente Figura: ![alt text](https://github.com/prbocca/na101_master/raw/master/homework_5_communities/political_blogs.png)\n# \n\n# In[ ]:\n\n# La matriz original tiene algunos loops, no son de nuestro interés y los eliminamos\ndiag(A) = 0\n\n# cargar el grafo no dirigido, g, con los datos de afinidad en un atributo \"nodes\" de los vértices\ng = NA\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: -\n#\n#\n#\n##################################################################\nsummary(g)\n\n#plot\nset.seed(1)\nplot(g, edge.color=\"gray\", edge.width=1, edge.lty=1, edge.arrow.size=.5, \n vertex.color=nodes, vertex.size=3, vertex.label=NA,\n layout=layout_components(g))\n\n\n# ## 4.b) Matriz de modularidad\n# \n# Calcular el grado $d_v$ de todos los vértices $v\\in V$ y el total de aristas $N_e$. Entonces calcular la matriz de modularidad $\\bbB$ de la red. Comparar el resultado con el de la función `modularity_matrix()`.\n# \n\n# In[ ]:\n\n#numero de aristas: 16715\nprintf(\"El numero de aristas es: %s\", length(E(g)))\n\n# calcular B\nB = NA\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: - calcular B a partir del grado\n#\n#\n#\n##################################################################\nstr(B)\n\n# usando igraph\nB2 = modularity_matrix(g, membership =rep(1,vcount(g)))\nstr(B2)\nidentical(B,B2)\n\n\n# ## 4.c) Particionado goloso\n# \n# Realizar el particionado goloso rápido (clustering jerárquico aglomerativo) en varias comunidades (`fastgreedy.community()`).\n# \n# \n\n# In[ ]:\n\ng.fg = fastgreedy.community(g, merges = T)\n\nprintf(\"Se encontraron %s comunidades\", length(g.fg))\nprintf(\"La mayoría de las comunidades son muy pequeñas:\")\nsizes(g.fg)\n\n\n# Dado que el metodo de particionado es jerárquico es posible definir la cantidad de comunidades que se quiere. Lamentablemente esto no siempre funciona, y este es el caso: no podemos particionar en dos usando este método. Ver el siguiente código.\n\n# In[ ]:\n\ncutat(g.fg, no=2) # no me deja partirlo en dos\n\n\n# Por tanto, voy a predecir la afiliación utilizando solo las dos comunidades más importantes. El resto de los vértices los dejo sin clasificar.\n# \n# Para esto me creo la matriz de confusión entre la afiliación real y la predicha. Conocer más de la matriz de confusión en: https://es.wikipedia.org/wiki/Matriz_de_confusi%C3%B3n\n\n# In[ ]:\n\n# me creo la matriz de confusión entre la afiliación real (nodes), y la predicción (usando solo las dos comunidades mas importantes)\ng.fg.mem = membership(g.fg)\ng.fg.t = as.data.frame(table(nodes, g.fg.mem), stringsAsFactors = F)\ng.fg.t = g.fg.t[order(g.fg.t$Freq, decreasing = T),][1:2,] #obtiene el mejor mapeo de las dos comunidades mas importantes\n\nif (g.fg.t[1,1]==0){\n comm_of_0 = g.fg.t[1,2]\n comm_of_1 = g.fg.t[2,2]\n} else{\n comm_of_0 = g.fg.t[2,2]\n comm_of_1 = g.fg.t[1,2]\n}\nprintf(\"El mejor mapeo para la afiliación 0 es la comunidad %s, y para la afiliación 1 es la comunidad %s\", comm_of_0, comm_of_1)\n\nprintf(\"La matriz de confusión es:\")\ng.fg.mem.bin = ifelse(g.fg.mem %in% c(comm_of_0, comm_of_1), g.fg.mem, \"other\")\ncm.fg = as.matrix(table(nodes, g.fg.mem.bin))\ncm.fg = cm.fg[,c(comm_of_0,comm_of_1, \"other\")] #reordeno las columnas \ncm.fg\n\nprintf(\"La exactitud del método es: %s\", sum(diag(cm.fg))/sum(cm.fg)) #porcentaje de bien detectados 0.7583893\n# También se pudo calcular con sum(g.fg.t$Freq)/vcount(g)\n\n\n# Puedo volver a graficar la red, agregando la información de cuando no realizo una buena predicción (vértices cuadrados), con el objetivo de buscar algún patrón en el error.\n\n# In[ ]:\n\n# plot\nset.seed(1)\nshapes = rep(\"square\",vcount(g)) #dibujo con cuadrado los que estan sin clasificiar o mal clasificados\nshapes[(nodes==0)&(g.fg.mem==comm_of_0)] = \"circle\" # dibujo con circulos los bien clasificados como 0\nshapes[(nodes==1)&(g.fg.mem==comm_of_1)] = \"circle\" # dibujo con circulos los bien clasificados como 1\nplot(g, edge.color=\"gray\", edge.width=1, edge.lty=1, edge.arrow.size=.5, \n vertex.color=nodes, vertex.size=3, vertex.label=NA,\n vertex.shape = shapes,\n layout=layout_components(g))\n\n# se ve que principalmente los cuadrados estan afuera de la componente gigante \n\n\n# ## 4.d) Particionado espectral\n# \n# Realizar el particionado espectral (maximización espectral de modularidad) en varias comunidades (`leading.eigenvector.community()`).\n\n# In[ ]:\n\ng.part = NA\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: similar a anterior\n#\n#\n#\n##################################################################\n\nprintf(\"Se encontraron %s comunidades\", length(g.part))\nprintf(\"La mayoría de las comunidades son muy pequeñas:\")\nsizes(g.part)\n\n\n# Dado que el metodo de particionado es jerárquico es posible definir la cantidad de comunidades que se quiere. ¿Es posible generar solo dos comunidades con este método? \n\n# In[ ]:\n\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: similar a anterior\n#\n#\n#\n##################################################################\n\n\n# Entonces, voy a predecir la afiliación utilizando solo las dos comunidades más importantes. El resto de los vértices los dejo sin clasificar.\n# \n# Crear la matriz de confusión entre la afiliación real y la predicha. Y calcular la exactitud del método.\n\n# In[ ]:\n\n\n# calcular la matriz de confusión\ncm.part = NA\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: similar a anterior\n#\n#\n#\n##################################################################\ncm.part\n\nprintf(\"La exactitud del método es: %s\", sum(diag(cm.part))/sum(cm.part)) #porcentaje de bien detectados 0.772483221\n# También se pudo calcular con sum(g.part.t$Freq)/vcount(g)\n\n\n# Graficar la red, agregando la información de cuando no realizo una buena predicción (vértices cuadrados), con el objetivo de buscar algún patrón en el error.\n\n# In[ ]:\n\n# plot\nset.seed(1)\nshapes = rep(\"square\",vcount(g)) #dibujo con cuadrado los que estan sin clasificiar o mal clasificados\nshapes[(nodes==0)&(g.part.mem==comm_of_0)] = \"circle\" # dibujo con circulos los bien clasificados como 0\nshapes[(nodes==1)&(g.part.mem==comm_of_1)] = \"circle\" # dibujo con circulos los bien clasificados como 1\nplot(g, edge.color=\"gray\", edge.width=1, edge.lty=1, edge.arrow.size=.5, \n vertex.color=nodes, vertex.size=3, vertex.label=NA,\n vertex.shape = shapes,\n layout=layout_components(g))\n\n# nuevamente se ve que los cuadrados estan principalmente afuera de la componente gigante\n\n\n# ## 4.e) Particionado espectral con matrices\n# \n# ¿Cómo detectar solo dos comunidades con los resultados o funciones anteriores? Si no es posible, entonces implementar el algoritmo espectral de maximización de modularidad visto en teórico. Además, crear la matriz de confusión entre la afiliación real y la predicha. Y calcular la exactitud del método.\n\n# In[ ]:\n\n# crear el dataframe g.mymod.t, con misma estructura de los casos anteriores,\n# pero hacerlo calculando la modularidad aplicando el método de teórico\ng.mymod.t = NA\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: calcular la modularidad aplicando el método de teórico\n#\n#\n#\n##################################################################\ng.mymod.t\n\n# crear la matriz de confusión\ncm.mymod = NA\n##################################################################\n# TU CÓDIGO ACÁ \n# Tip: similar a los anteriores\n#\n#\n#\n##################################################################\ncm.mymod\n\nprintf(\"La exactitud del método es: %s\", sum(diag(cm.mymod))/sum(cm.mymod)) #porcentaje de bien detectados 0.8711409\n# También se pudo calcular con sum(g.mymod.t$Freq)/vcount(g)\n\n\n# Graficar la red, agregando la información de cuando no realizo una buena predicción (vértices cuadrados), con el objetivo de buscar algún patrón en el error.\n\n# In[ ]:\n\n# plot\nset.seed(1)\nshapes = rep(\"square\",vcount(g)) #dibujo con cuadrado los que estan sin clasificiar o mal clasificados\nshapes[(nodes==0)&(g.mymod.mem==comm_of_0)] = \"circle\" # dibujo con circulos los bien clasificados como 0\nshapes[(nodes==1)&(g.mymod.mem==comm_of_1)] = \"circle\" # dibujo con circulos los bien clasificados como 1\nplot(g, edge.color=\"gray\", edge.width=1, edge.lty=1, edge.arrow.size=.5, \n vertex.color=nodes, vertex.size=3, vertex.label=NA,\n vertex.shape = shapes,\n layout=layout_components(g))\n\n#mejora visiblemente respecto a los anteriores\n\n\n# ## 4.f) Predicción de afinidad política.\n# \n# Si usaramos los métodos anteriores para predecir la afinidad política. ¿Cuál sería el mejor método de acuerdo a la exactitud?\n\n# In[ ]:\n\nprintf(\"La exactitud del método Goloso es: %s\", sum(diag(cm.fg))/sum(cm.fg)) #porcentaje de bien detectados 0.7583893\nprintf(\"La exactitud del método Particionado Espectral es: %s\", sum(diag(cm.part))/sum(cm.part)) #porcentaje de bien detectados 0.772483221\nprintf(\"La exactitud del método Particionado Espectral con matrices es: %s\", sum(diag(cm.mymod))/sum(cm.mymod)) #porcentaje de bien detectados 0.8711409\nprintf(\"El mejor método es el desarrollado por nosotros, el de maximizacion de modularidad: %s\", sum(diag(cm.mymod))/sum(cm.mymod))\n\n", "meta": {"hexsha": "27fc5b2b5f25d5bd7be9043c28bfadeb6678f04a", "size": 19869, "ext": "r", "lang": "R", "max_stars_repo_path": "r_deprecated/homework_5_communities/ar_hw5.r", "max_stars_repo_name": "prbocca/na101_master", "max_stars_repo_head_hexsha": "63640fbaa5c4d149052f8123587ebe360aea6fd1", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "r_deprecated/homework_5_communities/ar_hw5.r", "max_issues_repo_name": "prbocca/na101_master", "max_issues_repo_head_hexsha": "63640fbaa5c4d149052f8123587ebe360aea6fd1", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "r_deprecated/homework_5_communities/ar_hw5.r", "max_forks_repo_name": "prbocca/na101_master", "max_forks_repo_head_hexsha": "63640fbaa5c4d149052f8123587ebe360aea6fd1", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-16T19:06:12.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-16T19:06:12.000Z", "avg_line_length": 36.7944444444, "max_line_length": 304, "alphanum_fraction": 0.6704917208, "num_tokens": 5662, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6825737344123242, "lm_q2_score": 0.7634837581726991, "lm_q1q2_score": 0.5211339599790951}} {"text": "# NOTE: this is a comment.\npred_prey_euler <- function(a, b, h, x0, y0, const) {\n NOTE <- ((b - a) / h)\n XXX <- Y <- T <- double(n + 1)\n X[1] <- x0; Y[1] <- y0; T[1] <- a\n for (i in 1:n) { # XXX: here's another one.\n X[i + 1] <- X[i] + h * X[i] * (const[1] - const[2] * Y[i])\n Y[i + 1] <- Y[i] + h * Y[i] * (const[4] * X[i] - const[3])\n T[i + 1] <- T[i] + h\n }\n return(list(\"T\" = T, \"X\" = X, \"Y\" = Y))\n}\n", "meta": {"hexsha": "83a7a8b9e97b43fd4a994da25d2f7c2b3e93ea1c", "size": 419, "ext": "r", "lang": "R", "max_stars_repo_path": "fixtures/formats/test.r", "max_stars_repo_name": "TibsAtWork/vale", "max_stars_repo_head_hexsha": "01dc15dd4dff302ec2a4b5891fa5e1b5e9e4bc80", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1633, "max_stars_repo_stars_event_min_datetime": "2018-04-17T11:27:16.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-31T23:05:48.000Z", "max_issues_repo_path": "fixtures/formats/test.r", "max_issues_repo_name": "TibsAtWork/vale", "max_issues_repo_head_hexsha": "01dc15dd4dff302ec2a4b5891fa5e1b5e9e4bc80", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 346, "max_issues_repo_issues_event_min_datetime": "2018-04-17T11:17:30.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-31T16:30:11.000Z", "max_forks_repo_path": "fixtures/formats/test.r", "max_forks_repo_name": "TibsAtWork/vale", "max_forks_repo_head_hexsha": "01dc15dd4dff302ec2a4b5891fa5e1b5e9e4bc80", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 85, "max_forks_repo_forks_event_min_datetime": "2018-08-13T20:34:01.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-02T20:38:01.000Z", "avg_line_length": 32.2307692308, "max_line_length": 62, "alphanum_fraction": 0.4200477327, "num_tokens": 198, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.7956581000631541, "lm_q2_score": 0.6548947357776795, "lm_q1q2_score": 0.5210723012102297}} {"text": "\n\n# Snow crab --- Areal unit modelling Delta model variation\n# combination of three models via posterior simulation\n\n# 1. Poisson on positive valued numbers offset by swept are\n# 2. Meansize in space and time \n# 3 Presence-absence\n# the convolution of all three after simulation is called a Hurdle or Delta model\n \n# TODO::: move plotting calls to self-contained functions:\n\n\n# -------------------------------------------------\n# Part 1 -- construct basic parameter list defining the main characteristics of the study\n\n source( file.path( code_root, \"bio_startup.R\" ) )\n\n\n\n require(bio.snowcrab) # loadfunctions(\"bio.snowcrab\") \n\n year.assessment = 2021\n yrs = 1999:year.assessment\n spec_bio = bio.taxonomy::taxonomy.recode( from=\"spec\", to=\"parsimonious\", tolookup=2526 )\n\n runlabel= \"1999_present_fb\"\n\n snowcrab_filter_class = \"fb\" # fishable biomass (including soft-shelled )\n\n\n # params for number\n pN = snowcrab_parameters(\n project_class=\"carstm\",\n yrs=yrs, \n areal_units_type=\"tesselation\",\n family=\"poisson\",\n carstm_model_label= runlabel, \n selection = list(\n type = \"number\",\n biologicals=list( spec_bio=spec_bio ),\n biologicals_using_snowcrab_filter_class=snowcrab_filter_class\n )\n )\n\n # params for mean size .. mostly the same as pN\n pW = snowcrab_parameters(\n project_class=\"carstm\",\n yrs=yrs, \n areal_units_type=\"tesselation\",\n family = \"gaussian\",\n carstm_model_label= runlabel, \n selection = list(\n type = \"meansize\",\n biologicals=list( spec_bio=spec_bio ),\n biologicals_using_snowcrab_filter_class=snowcrab_filter_class\n )\n\n )\n\n # params for probability of observation\n pH = snowcrab_parameters( \n project_class=\"carstm\", \n yrs=yrs, \n areal_units_type=\"tesselation\", \n family = \"binomial\", # \"binomial\", # \"nbinomial\", \"betabinomial\", \"zeroinflatedbinomial0\" , \"zeroinflatednbinomial0\"\n carstm_model_label= runlabel, \n selection = list(\n type = \"presence_absence\",\n biologicals=list( spec_bio=spec_bio ),\n biologicals_using_snowcrab_filter_class=snowcrab_filter_class\n )\n )\n\n \n if (areal_units) {\n # polygon structure:: create if not yet made\n # for (au in c(\"cfanorth\", \"cfasouth\", \"cfa4x\", \"cfaall\" )) plot(polygon_managementareas( species=\"snowcrab\", au))\n xydata = snowcrab.db( p=pN, DS=\"areal_units_input\", redo=TRUE )\n xydata = snowcrab.db( p=pN, DS=\"areal_units_input\" )\n sppoly = areal_units( p=pN, xydata=xydata[ which(xydata$yr %in% pN$yrs), ], redo=TRUE, verbose=TRUE ) # create constrained polygons with neighbourhood as an attribute\n \n sppoly=areal_units( p=pN )\n\n plot(sppoly[\"npts\"])\n\n \n additional_features = snowcrab_features_tmap(pN) # for mapping below\n\n figure_area_based_extraction_from_carstm(DS=\"temperature\" ) # can only do done once we have an sppoly for snow crab\n \n M = snowcrab.db( p=pN, DS=\"carstm_inputs\", sppoly=sppoly, redo=TRUE ) # will redo if not found\n \n }\n\n sppoly=areal_units( p=pN )\n \n \n\n# ------------------------------------------------\n# Part 2 -- spatiotemporal statistical model\n\n if ( spatiotemporal_model ) {\n\n # total numbers\n sppoly = areal_units( p=pN )\n M = snowcrab.db( p=pN, DS=\"carstm_inputs\", sppoly=sppoly ) # will redo if not found\n\n io = which(M$tag==\"observations\")\n ip = which(M$tag==\"predictions\")\n iq = unique( c( which( M$totno > 0), ip ) )\n iw = unique( c( which( M$totno > 30), ip ) ) # need a good sample to estimate mean size\n\n \n # number \n fit = NULL; gc()\n fit = carstm_model( p=pN, data=M[ iq, ], sppoly=sppoly, \n posterior_simulations_to_retain=\"predictions\", improve.hyperparam.estimates=TRUE\n )\n\n # mean size\n fit = NULL; gc()\n fit = carstm_model( p=pW, data=M[ iw, ], sppoly = sppoly, \n posterior_simulations_to_retain=\"predictions\", improve.hyperparam.estimates=TRUE,\n control.inla = list( strategy=\"laplace\", int.strategy=\"eb\" )\n ) \n\n # model pa using all data\n fit = NULL; gc()\n fit = carstm_model( p=pH, data=M, sppoly=sppoly, \n posterior_simulations_to_retain=\"predictions\", improve.hyperparam.estimates=TRUE,\n # control.family=list(control.link=list(model=\"logit\")), # default\n control.inla = list( strategy=\"laplace\", int.strategy=\"eb\" )\n )\n\n # choose: \n p = pN\n p = pW\n p = pH\n\n if (0) {\n # extract results\n fit = carstm_model( p=p, DS=\"carstm_modelled_fit\", sppoly = sppoly ) # extract currently saved model fit\n fit$summary$dic$dic\n fit$summary$dic$p.eff\n plot(fit)\n plot(fit, plot.prior=TRUE, plot.hyperparameters=TRUE, plot.fixed.effects=FALSE )\n plot( fit, plot.prior=TRUE, plot.hyperparameters=TRUE, plot.fixed.effects=FALSE )\n plot( fit$marginals.hyperpar$\"Phi for space_time\", type=\"l\") # posterior distribution of phi nonspatial dominates\n plot( fit$marginals.hyperpar$\"Precision for space_time\", type=\"l\")\n plot( fit$marginals.hyperpar$\"Precision for setno\", type=\"l\")\n fit = NULL\n }\n\n res = carstm_model( p=p, DS=\"carstm_modelled_summary\", sppoly = sppoly ) # to load currently saved results\n\n if (0) {\n p = pH\n res = carstm_model( p=p, DS=\"carstm_modelled_summary\", sppoly = sppoly ) # to load currently saved results\n\n o = carstm_2D_effects_probability( \n res,\n xvar = \"inla.group(t, method = \\\"quantile\\\", n = 11)\", \n yvar = \"inla.group(z, method = \\\"quantile\\\", n = 11)\", \n xgrid = seq( -1, 10.5, by=0.5),\n ygrid = seq( 25, 350, by=25),\n xslice = 4,\n yslice = -200,\n nx=200, ny=200,\n theta = 140,\n phi = 15\n )\n \n # use a larger domain than sppoly for the following estimate:\n # sppoly is constrained to sampled locations, and so missing a lot of the inshore areas\n \n x11()\n crs_plot = st_crs( sppoly )\n domain = polygon_managementareas( species=\"maritimes\" )\n domain = st_transform( domain, crs_plot )\n data_mask = st_union( sppoly[which(sppoly$filter==1),1] ) \n # all = st_union( domain, data_mask )\n nearshore = st_cast( st_difference( domain, data_mask ), \"POLYGON\")[1]\n domain_new = st_union( data_mask, nearshore )\n \n o = carstm_optimal_habitat( \n res = res,\n xvar = \"inla.group(t, method = \\\"quantile\\\", n = 11)\", \n yvar = \"inla.group(z, method = \\\"quantile\\\", n = 11)\",\n depths=c(100, 350),\n probability_limit = 0.25,\n nsims = 100,\n domain=domain_new \n ) \n \n dev.new();\n print( o[\"depth_plot\"] )\n\n if (0) {\n u = readRDS('/home/jae/tmp/temp_depth_habitat.RDS')\n dev.new()\n plot( habitat~yr, u, type=\"b\", ylim=c(0.1, 0.33))\n lines( habitat_lb~yr, u)\n lines( habitat_ub~yr, u)\n abline(v=1993)\n abline(v=2012)\n \n dev.new()\n plot( habitat_sa~yr, u, type=\"b\", ylim=c( 25000, 75000))\n lines( habitat_sa_lb~yr, u)\n lines( habitat_sa_ub~yr, u)\n abline(v=1993)\n abline(v=2012)\n\n ll = loess(habitat~yr, u, span=0.25 )\n pp = predict( ll, u )\n lines(pp ~ u$yr)\n\n }\n\n outputdir = file.path( p$modeldir, p$carstm_model_label )\n fn_optimal = file.path( outputdir, \"optimal_habitat_temperature_depth_effect.RDS\" )\n saveRDS( o, file=fn_optimal, compress=FALSE )\n o = readRDS(fn_optimal)\n \n library(ggplot2)\n\n dev.new(width=14, height=8, pointsize=20)\n ggplot( o[[\"temperature_depth\"]], aes(yr, habitat ) ) +\n geom_ribbon(aes(ymin=habitat_lb, max=habitat_ub), alpha=0.2, colour=NA) +\n geom_line() +\n labs(x=\"Year\", y=\"Habitat probabtility\", size = rel(1.5)) +\n # scale_y_continuous( limits=c(0, 300) ) \n theme_light( base_size = 22 ) \n \n\n dev.new(width=14, height=8, pointsize=20)\n ggplot( o[[\"temperature_depth\"]], aes(yr, habitat_sa ) ) +\n geom_ribbon(aes(ymin=habitat_sa_lb, max=habitat_sa_ub), alpha=0.2, colour=NA) +\n geom_line() +\n labs(x=\"Year\", y=bquote(\"Habitat surface area;\" ~ km^2), size = rel(1.5)) +\n # scale_y_continuous( limits=c(0, 300) ) \n theme_light( base_size = 22 ) \n \n }\n\n\n\n if (0) {\n\n vn=c( \"random\", \"space\", \"combined\" )\n vn=c( \"random\", \"spacetime\", \"combined\" )\n vn=\"predictions\" # numerical density (km^-2)\n\n tmatch= as.character(year.assessment)\n\n carstm_map( res=res, vn=vn, tmatch=tmatch, \n sppoly = sppoly, \n palette=\"-RdYlBu\",\n plot_elements=c( \"compass\", \"scale_bar\", \"legend\" ),\n additional_features=additional_features,\n title =paste( vn, paste0(tmatch, collapse=\"-\"), \"no/m^2\" )\n )\n\n\n # map all :\n if ( number ) {\n p=pN\n res = carstm_model( p=p, DS=\"carstm_modelled_summary\", sppoly = sppoly ) # to load currently saved results\n outputdir = file.path( p$modeldir, p$carstm_model_label, \"predicted.numerical.densitites\" )\n if ( !file.exists(outputdir)) dir.create( outputdir, recursive=TRUE, showWarnings=FALSE )\n fn_root_prefix = \"Predicted_numerical_abundance\"\n fn_root = \"Predicted_numerical_abundance_persistent_spatial_effect\" \n outfilename = file.path( outputdir, paste(fn_root, \"png\", sep=\".\") )\n title= paste( snowcrab_filter_class, \"Predicted numerical density (no./m^2) - persistent spatial effect\" )\n }\n if ( meansize) {\n p=pW\n res = carstm_model( p=p, DS=\"carstm_modelled_summary\", sppoly = sppoly ) # to load currently saved results\n outputdir = file.path( p$modeldir, p$carstm_model_label, \"predicted.meansize\" )\n if ( !file.exists(outputdir)) dir.create( outputdir, recursive=TRUE, showWarnings=FALSE )\n fn_root_prefix = \"Predicted_meansize\"\n fn_root = \"Predicted_meansize_persistent_spatial_effect\" \n outfilename = file.path( outputdir, paste(fn_root, \"png\", sep=\".\") )\n title= paste( snowcrab_filter_class, \"Predicted meansize (kg) - persistent spatial effect\" ) \n }\n if ( presence_absence ) {\n p=pH\n res = carstm_model( p=p, DS=\"carstm_modelled_summary\", sppoly = sppoly ) # to load currently saved results\n outputdir = file.path( p$modeldir, p$carstm_model_label, \"predicted.presence_absence\" )\n if ( !file.exists(outputdir)) dir.create( outputdir, recursive=TRUE, showWarnings=FALSE )\n fn_root_prefix = \"Predicted_presence_absence\"\n fn_root = \"Predicted_presence_absence_persistent_spatial_effect\" \n outfilename = file.path( outputdir, paste(fn_root, \"png\", sep=\".\") )\n title= paste( snowcrab_filter_class, \"Predicted habitat probability - persistent spatial effect\") \n }\n\n vn = c( \"random\", \"space\", \"combined\" ) \n toplot = carstm_results_unpack( res, vn )\n brks = pretty( quantile(toplot[,\"mean\"], probs=c(0,0.975), na.rm=TRUE ) )\n\n tmout = carstm_map( res=res, vn=vn, \n sppoly = sppoly, \n breaks = brks,\n palette=\"-RdYlBu\",\n plot_elements=c( \"compass\", \"scale_bar\", \"legend\" ),\n additional_features=additional_features,\n outfilename=outfilename,\n title= title\n ) \n tmout\n \n\n vn=\"predictions\"\n toplot = carstm_results_unpack( res, vn )\n brks = pretty( quantile(toplot[,,\"mean\"], probs=c(0,0.975), na.rm=TRUE ) )\n\n for (y in res$time ){\n tmatch = as.character(y)\n fn_root = paste(fn_root_prefix, paste0(tmatch, collapse=\"-\"), sep=\"_\")\n outfilename = file.path( outputdir, paste(fn_root, \"png\", sep=\".\") )\n\n tmout = carstm_map( res=res, vn=vn, tmatch=tmatch,\n sppoly = sppoly, \n breaks =brks,\n palette=\"-RdYlBu\",\n plot_elements=c( \"compass\", \"scale_bar\", \"legend\" ),\n additional_features=additional_features,\n outfilename=outfilename,\n title=paste(fn_root_prefix, snowcrab_filter_class, paste0(tmatch, collapse=\"-\") )\n )\n tmout\n print(outfilename)\n \n }\n\n # plots with 95% PI\n outputdir = file.path( p$modeldir, p$carstm_model_label, \"effects\" )\n if ( !file.exists(outputdir)) dir.create( outputdir, recursive=TRUE, showWarnings=FALSE )\n\n (fn = file.path( outputdir, \"time.png\"))\n png( filename=fn, width=1024, height=1024, pointsize=12, res=196 )\n carstm_plotxy( res, vn=c( \"res\", \"random\", \"time\" ), \n type=\"b\", ylim=c(0, 1), xlab=\"Year\", ylab=\"Probabilty\", h=0, cex=1.25, cex.axis=1.25, cex.lab=1.25 )\n dev.off()\n\n (fn = file.path( outputdir, \"cyclic.png\"))\n png( filename=fn, width=1024, height=1024, pointsize=12, res=196 )\n carstm_plotxy( res, vn=c( \"res\", \"random\", \"cyclic\" ), \n type=\"b\", col=\"slategray\", pch=19, lty=1, lwd=2.5, ylim=c(0.35, 0.65),\n xlab=\"Season\", ylab=\"Probabilty\", cex=1.25, cex.axis=1.25, cex.lab=1.25 )\n dev.off()\n\n\n (fn = file.path( outputdir, \"temperature.png\"))\n png( filename=fn, width=1024, height=1024, pointsize=12, res=196 )\n carstm_plotxy( res, vn=c( \"res\", \"random\", \"inla.group(t, method = \\\"quantile\\\", n = 11)\" ), \n type=\"b\", col=\"slategray\", pch=19, lty=1, lwd=2.5, ylim=c(0, 0.8) ,\n xlab=\"Bottom temperature (degrees Celsius)\", ylab=\"Probabilty\", cex=1.25, cex.axis=1.25, cex.lab=1.25 )\n dev.off()\n\n\n (fn = file.path( outputdir, \"depth.png\"))\n png( filename=fn, width=1024, height=1024, pointsize=12, res=196 )\n carstm_plotxy( res, vn=c( \"res\", \"random\", \"inla.group(z, method = \\\"quantile\\\", n = 11)\" ), \n type=\"b\", col=\"slategray\", pch=19, lty=1, lwd=2.5, ylim=c(0, 0.9) ,\n xlab=\"Depth (m)\", ylab=\"Probabilty\", cex=1.25, cex.axis=1.25, cex.lab=1.25 )\n dev.off()\n\n\n fit = carstm_model( p=pW, DS=\"carstm_modelled_fit\", sppoly = sppoly ) # to load currently saved results\n \n plot( fit, plot.prior=TRUE, plot.hyperparameters=TRUE, plot.fixed.effects=FALSE )\n plot( fit$marginals.hyperpar$\"Phi for space_time\", type=\"l\") # posterior distribution of phi nonspatial dominates\n plot( fit$marginals.hyperpar$\"Precision for space_time\", type=\"l\")\n plot( fit$marginals.hyperpar$\"Precision for setno\", type=\"l\")\n\n }\n\n\n } # end spatiotemporal model\n\n\n# ----------------------\n\n assimilate_numbers_and_size = TRUE\n\n if (assimilate_numbers_and_size ) {\n\n # wgts_max = 1.1 # kg, hard upper limit\n # N_max = NULL\n # # quantile( M$totno[ipositive]/M$data_offset[ipositive], probs=0.95, na.rm=TRUE ) \n \n # posterior sims \n \n sims = carstm_posterior_simulations( pN=pN, pW=pW, pH=pH, sppoly=sppoly, pa_threshold=0.05, qmax=0.99 )\n sims = sims / 10^6 # 10^6 kg -> kt;; kt/km^2\n\n \n SM = aggregate_biomass_from_simulations( \n sims=sims, \n sppoly=sppoly, \n fn=carstm_filenames( pN, returnvalue=\"filename\", fn=\"aggregated_timeseries\" ), \n yrs=pN$yrs, \n method=\"sum\", \n redo=TRUE \n ) \n \n RES= SM$RES\n # RES = aggregate_biomass_from_simulations( fn=carstm_filenames( pN, returnvalue=\"filename\", fn=\"aggregated_timeseries\" ) )$RES\n\n outputdir = file.path( carstm_filenames( pN, returnvalue=\"output_directory\"), \"aggregated_biomass_timeseries\" )\n\n if ( !file.exists(outputdir)) dir.create( outputdir, recursive=TRUE, showWarnings=FALSE )\n\n\n ( fn = file.path( outputdir, \"cfa_all.png\") )\n png( filename=fn, width=3072, height=2304, pointsize=12, res=300 )\n plot( cfaall ~ yrs, data=RES, lty=\"solid\", lwd=4, pch=20, col=\"slateblue\", type=\"b\", ylab=\"Biomass index (kt)\", xlab=\"\")\n lines( cfaall_lb ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n lines( cfaall_ub ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n dev.off()\n\n\n ( fn = file.path( outputdir, \"cfa_south.png\") )\n png( filename=fn, width=3072, height=2304, pointsize=12, res=300 )\n plot( cfasouth ~ yrs, data=RES, lty=\"solid\", lwd=4, pch=20, col=\"slateblue\", type=\"b\", ylab=\"Biomass index (kt)\", xlab=\"\")\n lines( cfasouth_lb ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n lines( cfasouth_ub ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n dev.off()\n\n ( fn = file.path( outputdir, \"cfa_north.png\") )\n png( filename=fn, width=3072, height=2304, pointsize=12, res=300 )\n plot( cfanorth ~ yrs, data=RES, lty=\"solid\", lwd=4, pch=20, col=\"slateblue\", type=\"b\", ylab=\"Biomass index (kt)\", xlab=\"\")\n lines( cfanorth_lb ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n lines( cfanorth_ub ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n dev.off()\n\n ( fn = file.path( outputdir, \"cfa_4x.png\") )\n png( filename=fn, width=3072, height=2304, pointsize=12, res=300 )\n plot( cfa4x ~ yrs, data=RES, lty=\"solid\", lwd=4, pch=20, col=\"slateblue\", type=\"b\", ylab=\"Biomass index (kt)\", xlab=\"\")\n lines( cfa4x_lb ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n lines( cfa4x_ub ~ yrs, data=RES, lty=\"dotted\", lwd=2, col=\"slategray\" )\n dev.off()\n\n\n # map it ..mean density\n sppoly = areal_units( p=pN ) # to reload\n\n vn = paste(\"biomass\", \"predicted\", sep=\".\")\n\n outputdir = file.path( carstm_filenames( pN, returnvalue=\"output_directory\"), \"predicted_biomass_densitites\" )\n\n if ( !file.exists(outputdir)) dir.create( outputdir, recursive=TRUE, showWarnings=FALSE )\n\n B = apply( sims, c(1,2), mean ) \n B[ which(!is.finite(B)) ] = NA\n\n brks = pretty( log10( quantile( B[], probs=c(0.05, 0.95), na.rm=TRUE )* 10^6) )\n \n additional_features = snowcrab_features_tmap(pN) # for mapping below\n\n for (i in 1:length(pN$yrs) ) {\n y = as.character( pN$yrs[i] )\n sppoly[,vn] = log10( B[,y]* 10^6 )\n outfilename = file.path( outputdir , paste( \"biomass\", y, \"png\", sep=\".\") )\n tmout = carstm_map( sppoly=sppoly, vn=vn,\n breaks=brks,\n additional_features=additional_features,\n title=paste( \"log_10( Predicted biomass density; kg/km^2 )\", y ),\n palette=\"-RdYlBu\",\n plot_elements=c( \"compass\", \"scale_bar\", \"legend\" ), \n outfilename=outfilename\n )\n tmout\n \n }\n \n } # end assimilate size and numbers\n\n\n\n\n\n##########\n\n# this part is only relevent for R0 \n\nfishery_model = FALSE\n\nif (fishery_model) {\n\n # you need a stan installation on your system as well (outside of R), and the R-interface \"cmdstanr\":\n # install.packages(\"cmdstanr\", repos = c(\"https://mc-stan.org/r-packages/\", getOption(\"repos\")))\n \n require(cmdstanr)\n \n loadfunctions(\"bio.snowcrab\")\n \n # choose:\n pN$fishery_model_label = \"stan_surplus_production_2022_model_qc_uniform\"\n # pN$fishery_model_label = \"stan_surplus_production_2022_model_variation1_wider_qc_uniform\"\n # pN$fishery_model_label = \"stan_surplus_production_2022_model_variation1_wider_qc_normal\"\n # pN$fishery_model_label = \"stan_surplus_production_2022_model_qc_cauchy_wider\"\n\n # pN$fishery_model_label = \"stan_surplus_production_2022_model_qc_beta\" # qc_beta postive definite\n # pN$fishery_model_label = \"stan_surplus_production_2022_model_qc_cauchy\"\n # pN$fishery_model_label = \"stan_surplus_production_2019_model\"\n\n\n pN$fishery_model = fishery_model( DS = \"logistic_parameters\", p=pN, tag=pN$fishery_model_label )\n\n to_look = c(\"K\", \"r\", \"q\", \"qc\", \"log_lik\" )\n\n if ( model_version==\"framework_2019\" ) {\n # this is to create results for reviewers in 2022 that wanted a comparison with previous methods .. can be deleted in future \n # bring in unscaled abundance index\n a = fishery_model( DS=\"data_aggregated_timeseries\", p=pN )\n a$IOA[ !is.finite(a$IOA) ] = 0\n pN$fishery_model$standata$IOA = a$IOA\n to_look = c(\"K\", \"r\", \"q\", \"log_lik\" )\n\n } \n\n\n # str( pN$fishery_model)\n\n pN$fishery_model$stancode = stan_initialize( stan_code=fishery_model( p=pN, DS=pN$fishery_model_label ) )\n pN$fishery_model$stancode$compile()\n \n\n fit = pN$fishery_model$stancode$sample(\n data=pN$fishery_model$standata,\n iter_warmup = 4000,\n iter_sampling = 4000,\n seed = 1,\n chains = 3,\n parallel_chains = 3, # The maximum number of MCMC chains to run in parallel.\n max_treedepth = 18,\n adapt_delta = 0.99,\n refresh = 1000\n )\n\n fit$summary(to_look)\n\n require(loo)\n waic(fit$draws(\"log_lik\"))\n loo(fit$draws(\"log_lik\"))\n \n \n # save fit and get draws\n res = fishery_model( p=pN, DS=\"logistic_model\", tag=pN$fishery_model_label, fit=fit ) # from here down are params for cmdstanr::sample()\n\n if (0) {\n # reload saved fit and results\n res = readRDS(pN$fishery_model$fnres)\n fit = readRDS(pN$fishery_model$fnfit)\n\n }\n \n # frequency density of key parameters\n fishery_model( DS=\"plot\", vname=\"K\", res=res )\n fishery_model( DS=\"plot\", vname=\"r\", res=res )\n fishery_model( DS=\"plot\", vname=\"q\", res=res, xrange=c(0.5, 2.5))\n fishery_model( DS=\"plot\", vname=\"qc\", res=res, xrange=c(-1, 1))\n fishery_model( DS=\"plot\", vname=\"FMSY\", res=res )\n\n # timeseries\n fishery_model( DS=\"plot\", type=\"timeseries\", vname=\"biomass\", res=res )\n fishery_model( DS=\"plot\", type=\"timeseries\", vname=\"fishingmortality\", res=res)\n\n # Harvest control rules\n fishery_model( DS=\"plot\", type=\"hcr\", vname=\"default\", res=res )\n\n # Summary table of mean values for inclusion in document\n \n ( qs = apply( res$mcmc$K[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) ) # carrying capactiy\n\n ( qs = apply( res$mcmc$FMSY[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) ) # FMSY\n\n\n biomass = as.data.table( fit$summary(\"B\") )\n np = year.assessment+c(1:pN$fishery_model$standata$M)\n biomass$yr = rep( c(pN$yrs, np ), 3)\n nt = pN$fishery_model$standata$N +pN$fishery_model$standata$M\n biomass$region = c( rep(\"cfanorth\", nt), rep(\"cfasouth\", nt), rep(\"cfa4x\", nt) )\n (biomass)\n\n NN = res$pN$fishery_model$standata$N\n\n # densities of biomass estimates for the year.assessment\n ( qs = apply( res$mcmc$B[,NN,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # densities of biomass estimates for the previous year\n ( qs = apply( res$mcmc$B[,NN-1,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # densities of F in assessment year\n ( qs = apply( res$mcmc$F[,NN,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n ( qs = apply( res$mcmc$F[,NN,], 2, mean ) )\n\n # densities of F in previous year\n ( qs = apply( res$mcmc$F[,NN-1,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n ( qs = apply( res$mcmc$F[,NN-1,], 2, mean ) )\n\n\n\n if (0) {\n # obsolete:\n\n fishery_model( DS=\"plot\", vname=\"bosd\", res=res )\n fishery_model( DS=\"plot\", vname=\"bpsd\", res=res )\n fishery_model( DS=\"plot\", type=\"hcr\", vname=\"simple\", res=res )\n \n # biomass.summary.table()\n\n fit = fishery_model( p=pN, DS=\"fit\", tag=pN$fishery_model_label ) # to load samples (results)\n fit$summary(c(\"K\", \"r\", \"q\", \"qc\"))\n print( fit, max_rows=30 )\n fit$cmdstan_diagnose()\n fit$cmdstan_summary()\n\n # testing other samplers and optimizsers ... faster , good for debugging\n\n # (penalized) maximum likelihood estimate (MLE)\n fit_mle = pN$fishery_model$stancode$optimize(data =pN$fishery_model$standata, seed = 123)\n fit_mle$summary( to_look )\n u = stan_extract( as_draws_df(fit_mle$draws() ) )\n\n # mcmc_hist(fit$draws(\"K\")) + vline_at(fit_mle$mle(), size = 1.5)\n\n # # Variational Bayes\n fit_vb = pN$fishery_model$stancode$variational( data =pN$fishery_model$standata, seed = 123, output_samples = 4000)\n fit_vb$summary(to_look)\n fit_vb$cmdstan_diagnose()\n fit_vb$cmdstan_summary()\n\n\n u = stan_extract( as_draws_df(fit_vb$draws() ) )\n\n # bayesplot_grid(\n # mcmc_hist(fit$draws(\"K\"), binwidth = 0.025),\n # mcmc_hist(fit_vb$draws(\"K\"), binwidth = 0.025),\n # titles = c(\"Posterior distribution from MCMC\", \"Approximate posterior from VB\")\n # )\n\n # color_scheme_set(\"gray\")\n # mcmc_dens(fit$draws(\"K\"), facet_args = list(nrow = 3, labeller = ggplot2::label_parsed ) ) + facet_text(size = 14 )\n # mcmc_hist( fit$draws(\"K\"))\n\n # obtain mcmc samples from vb solution\n res_vb = fishery_model(\n DS=\"logistic_model\",\n p=pN,\n tag=pN$fishery_model_label,\n fit = fit_vb\n )\n\n names(res_vb$mcmc)\n\n # other diagnostics\n # fishery_model( DS=\"plot\", type=\"diagnostic.errors\", res=res )\n # fishery_model( DS=\"plot\", type=\"diagnostic.phase\", res=res )\n\n NN = res$pN$fishery_model$standata$N\n\n # bosd\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$bosd[,i] ), main=\"\")\n ( qs = apply( res$mcmc$bosd[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n\n # bpsd\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$bpsd[,i] ), main=\"\")\n ( qs = apply( res$mcmc$bpsd[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # rem_sd\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$rem_sd[,i] ), main=\"\")\n ( qs = apply( res$mcmc$rem_sd[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n\n # qc\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$qc[,i] ), main=\"\")\n ( qs = apply( res$mcmc$qc[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # b0\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$b0[,i] ), main=\"\")\n ( qs = apply( res$mcmc$b0[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n\n # K\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$K[,i] ), main=\"\")\n ( qs = apply( res$mcmc$K[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # R\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$r[,i] ), main=\"\")\n ( qs = apply( res$mcmc$r[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # q\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$q[,i] ), main=\"\")\n ( qs = apply( res$mcmc$q[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # FMSY\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$FMSY[,i] ), main=\"\")\n ( qs = apply( res$mcmc$FMSY[,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n\n # densities of biomass estimates for the year.assessment\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density(res$mcmc$B[,NN,i] ), main=\"\")\n ( qs = apply( res$mcmc$B[,NN,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # densities of biomass estimates for the previous year\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density( res$mcmc$B[,NN-1,i] ), main=\"\")\n ( qs = apply( res$mcmc$B[,NN-1,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n\n # densities of F in assessment year\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density( res$mcmc$F[,NN,i] ), xlim=c(0.01, 0.6), main=\"\")\n ( qs = apply( res$mcmc$F[,NN,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n ( qs = apply( res$mcmc$F[,NN,], 2, mean ) )\n\n # densities of F in previous year\n plot.new()\n layout( matrix(c(1,2,3), 3, 1 ))\n par(mar = c(4.4, 4.4, 0.65, 0.75))\n for (i in 1:3) plot(density( res$mcmc$F[,NN-1,i] ), xlim=c(0.01, 0.6), main=\"\")\n ( qs = apply( res$mcmc$F[,NN-1,], 2, quantile, probs=c(0.025, 0.5, 0.975) ) )\n ( qs = apply( res$mcmc$F[,NN-1,], 2, mean ) )\n\n # F for table ---\n summary( res$mcmc$F, median)\n\n } # end skip\n\n} # end fishery model\n\n\n\n\n\n# end\n", "meta": {"hexsha": "954090261dec3013366dd7e6b0fb3ff9fbbcd80e", "size": 28289, "ext": "r", "lang": "R", "max_stars_repo_path": "inst/scripts/03.snowcrab_carstm.r", "max_stars_repo_name": "jae0/snowcrab", "max_stars_repo_head_hexsha": "b168df368b739175004275c47f5bdf6907d066d9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "inst/scripts/03.snowcrab_carstm.r", "max_issues_repo_name": "jae0/snowcrab", "max_issues_repo_head_hexsha": "b168df368b739175004275c47f5bdf6907d066d9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "inst/scripts/03.snowcrab_carstm.r", "max_forks_repo_name": "jae0/snowcrab", "max_forks_repo_head_hexsha": "b168df368b739175004275c47f5bdf6907d066d9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.5019354839, "max_line_length": 171, "alphanum_fraction": 0.610095797, "num_tokens": 9364, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8902942173896132, "lm_q2_score": 0.5851011542032313, "lm_q1q2_score": 0.5209121741751253}} {"text": "# Setup\nset.seed(42)\ntarget= unlist(strsplit(\"METHINKS IT IS LIKE A WEASEL\", \"\"))\nchars= c(LETTERS, \" \")\nC= 100\n\n# Fitness function; high value means higher fitness\nfitness= function(x){\n sum(x == target)\n}\n\n# Mutate function\nmutate= function(x, rate= 0.01){\n idx= which(runif(length(target)) <= rate)\n x[idx]= replicate(n= length(idx), expr= sample(x= chars, size= 1, replace= T))\n x\n}\n\n# Evolve function\nevolve= function(x){\n parents= rep(list(x), C+1) # Repliction\n parents[1:C]= lapply(parents[1:C], function(x) mutate(x)) # Mutation\n idx= which.max(lapply(parents, function(x) fitness(x))) # Selection\n parents[[idx]]\n}\n\n# Initialize first parent\nparent= sample(x= chars, size= length(target), replace= T)\n\n# Main program\nwhile (fitness(parent) < fitness(target)) {\n parent= evolve(parent)\n cat(paste0(parent, collapse=\"\"), \"\\n\")\n}\n", "meta": {"hexsha": "da52147b9d62495d43b228306ab739eb49f95044", "size": 847, "ext": "r", "lang": "R", "max_stars_repo_path": "Task/Evolutionary-algorithm/R/evolutionary-algorithm-2.r", "max_stars_repo_name": "LaudateCorpus1/RosettaCodeData", "max_stars_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2021-05-05T13:42:20.000Z", "max_stars_repo_stars_event_max_datetime": "2021-05-05T13:42:20.000Z", "max_issues_repo_path": "Task/Evolutionary-algorithm/R/evolutionary-algorithm-2.r", "max_issues_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_issues_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Task/Evolutionary-algorithm/R/evolutionary-algorithm-2.r", "max_forks_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_forks_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 24.2, "max_line_length": 80, "alphanum_fraction": 0.6741440378, "num_tokens": 254, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8080672320414786, "lm_q2_score": 0.6442251064863698, "lm_q1q2_score": 0.5205771986100676}} {"text": "\r\n# Load UCBAdmissions dataset\r\ndata(UCBAdmissions)\r\n\r\n# Print dataset to console\r\nprint(UCBAdmissions)\r\n\r\n# Load broom package\r\nlibrary(broom)\r\n\r\n# Convert UCBAdmissions to tidy format\r\nucb_tidy <- tidy(UCBAdmissions)\r\n\r\n# Print tidy dataset to console\r\nprint(ucb_tidy)\r\n\r\n\r\n# Load the dplyr library\r\nlibrary(dplyr)\r\n\r\n# Aggregate over department\r\nucb_tidy_aggregated <- ucb_tidy %>% \r\n group_by(Admit, Gender) %>% \r\n summarize(n = sum(n)) %>% \r\n ungroup() %>% \r\n group_by(Gender) %>% \r\n mutate(prop = n/sum(n)) %>% \r\n filter(Admit == \"Admitted\")\r\n\r\n# Print aggregated dataset\r\nprint(ucb_tidy_aggregated)\r\n\r\n# Load the ggplot2 and scales packages\r\nlibrary(ggplot2)\r\nlibrary(scales)\r\n\r\n# Prepare the bar plot\r\ngg_bar <- ucb_tidy_aggregated %>% \r\n ggplot(aes(x = Gender, y = prop, fill = Gender)) +\r\n geom_col() +\r\n geom_text(aes(label = percent(prop)), vjust = -1) +\r\n labs(title = \"Acceptance rate of male and female applicants\",\r\n subtitle = \"University of California, Berkeley (1973)\",\r\n y = \"Acceptance rate\") +\r\n scale_y_continuous(labels = percent, limits = c(0,0.5)) +\r\n guides(fill = FALSE) # scales::percent.\r\n\r\n# Print the bar plot\r\nprint(gg_bar)\r\n\r\n# Calculate acceptance/rejection rate\r\nucb_by_dept <- ucb_tidy %>% \r\n group_by(Gender, Dept) %>% \r\n mutate(prop = n/sum(n)) %>% \r\n filter(Admit == \"Admitted\")\r\n\r\n# Print the dataset\r\nprint(ucb_by_dept)\r\n\r\n# Prepare the bar plot for each department\r\ngg_bar_faceted <- ucb_by_dept %>% \r\n ggplot(aes(Gender, prop, fill = Gender)) +\r\n geom_col() +\r\n geom_text(aes(label = percent(prop)), vjust = -1) +\r\n labs(title = \"Acceptance rate of male and female applicants\",\r\n subtitle = \"University of California, Berkeley (1973)\",\r\n y = \"Acceptance rate\") +\r\n scale_y_continuous(labels = scales::percent, limits = c(0, 1)) +\r\n facet_wrap(~Dept) +\r\n guides(fill = FALSE)\r\n\r\n# Print the bar plot for each department\r\nprint(gg_bar_faceted)\r\n\r\n\r\n# Define function that repeats each row in each column n times\r\nmultiply_rows <- function(column, n) {\r\n rep(column, n)\r\n} # It shown error if times = n here\r\n\r\n# Create new de-aggregated data frame using the multiply_rows function\r\nucb_full <- data.frame(Admit = multiply_rows(ucb_tidy$Admit, ucb_tidy$n),\r\n Gender = multiply_rows(ucb_tidy$Gender, ucb_tidy$n),\r\n Dept = multiply_rows(ucb_tidy$Dept, ucb_tidy$n))\r\n\r\n# Check the number of rows equals the number of students\r\nnrow(ucb_full) == 4526\r\n\r\nstr(ucb_full)\r\n\r\n# Load the forcats library\r\nlibrary(forcats)\r\n\r\n# Reverse the coding of the Admit variable\r\nucb_full$Admit <- fct_relevel(ucb_full$Admit,\r\n \"Rejected\", \"Admitted\")\r\n\r\n# Run the regression\r\nglm_gender <- glm(Admit ~ Gender, data = ucb_full, family = \"binomial\")\r\n\r\n# Summarize the results\r\nsummary(glm_gender)\r\n\r\n# Run the regression, including Dept as an explanatory variable\r\nglm_genderdept <- glm(Admit ~ Gender + Dept, data = ucb_full, family = \"binomial\")\r\n\r\n# Summarize the results\r\nsummary(glm_genderdept)\r\n\r\n\r\n# Filter for Department A\r\ndept_a <- ucb_full%>%\r\n filter(Dept == \"A\") # if ucb_tidy, dept_a$Admit datatype is \"char\", which will show error when run the regression\r\n\r\nstr(dept_a)\r\n\r\n# Run the regression\r\nglm_gender_depta <- glm(Admit ~ Gender, dept_a, family = \"binomial\")\r\n\r\n\r\n\r\n# Summarize the results\r\nsummary(glm_gender_depta)\r\n\r\n\r\n# Define bias\r\nbias <- \"Underreporting or misreporting of demographic, social or economic characteristics associated with one of the sexes.\"\r\n\r\n# Define discrimination\r\ndiscrimination <- \"the exercise of decision influenced by the sex of the applicant when that is immaterial to the qualifications for entry\"\r\n\r\n# Is bias equal to discrimination?\r\nbias == discrimination\r\n", "meta": {"hexsha": "43ea630c2c0440a17829afefbce9459654e59119", "size": 3774, "ext": "r", "lang": "R", "max_stars_repo_path": "07. Gender Bias in Graduate Admissions.r", "max_stars_repo_name": "Yafang611/DataScientistsProjects", "max_stars_repo_head_hexsha": "dfcde08aa3b218bc198e3e22dc19c81d2742ddde", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "07. Gender Bias in Graduate Admissions.r", "max_issues_repo_name": "Yafang611/DataScientistsProjects", "max_issues_repo_head_hexsha": "dfcde08aa3b218bc198e3e22dc19c81d2742ddde", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "07. Gender Bias in Graduate Admissions.r", "max_forks_repo_name": "Yafang611/DataScientistsProjects", "max_forks_repo_head_hexsha": "dfcde08aa3b218bc198e3e22dc19c81d2742ddde", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 28.1641791045, "max_line_length": 140, "alphanum_fraction": 0.6801801802, "num_tokens": 992, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6791787121629466, "lm_q2_score": 0.7662936324115011, "lm_q1q2_score": 0.5204503223999096}} {"text": "# Custom Math\n# Paul Sztorc\n# Written in R (v 3.1.1) using Rstudio (v 0.98.1028)\n\n# A collection of relatively basic functions.\n\n\nAsMatrix <- function(Vec) {\n # Takes a vector and transforms it into a 1 column matrix\n return(matrix(Vec,nrow=length(Vec)))\n}\n\n# Vector <- c(3,4,5)\n# > Vector\n# [1] 3 4 5\n# > AsMatrix(Vector)\n# [,1]\n# [1,] 3\n# [2,] 4\n# [3,] 5\n\nMeanNA <- function(Vec) {\n # Replaces NA instances with their mean instead\n m <- mean(Vec, na.rm = TRUE)\n Vec[is.na(Vec)] <- m\n return(Vec)\n}\n\n# > MeanNA(c(3,4,6,7,8))\n# [1] 3 4 6 7 8\n\n# > MeanNA(c(3,NA,6,7,8))\n# [1] 3 6 6 7 8\n\n# > MeanNA(c(0,0,0,1,NA))\n# [1] 0.00 0.00 0.00 1.00 0.25\n\n# > MeanNA(c(0,0,NA,1,NA))\n# [1] 0.0000000 0.0000000 0.3333333 1.0000000 0.3333333\n\n\n\nGetWeight <- function(Vec,AddMean=FALSE) {\n # Takes a Vector, absolute value, then proportional linear deviance from 0.\n new <- abs(Vec)\n if(AddMean==1) new <- new + mean(new)\n if(sum(new)==0) new <- new + 1\n new <- new/sum(new)\n return(new)\n}\n\n# > GetWeight(c(1,1,1,1))\n# [1] 0.25 0.25 0.25 0.25\n# > GetWeight(c(10,10,10,10))\n# [1] 0.25 0.25 0.25 0.25\n# > GetWeight(c(4,5,6,7))\n# [1] 0.1818182 0.2272727 0.2727273 0.3181818\n# > GetWeight(c(4,5,6,7),TRUE)\n# [1] 0.2159091 0.2386364 0.2613636 0.2840909\n\n\n\nCatch <- function(X,Tolerance=0) {\n # x is the ConoutRAW numeric, Tolerance is the length of the midpoint corresponding to .5 \n # The purpose here is to handle rounding for Binary Decisions (Scaled Decisions use weighted median)\n \n if(X<(.5-(Tolerance/2))) return(0)\n else if(X>(.5+(Tolerance/2))) return(1)\n else return(.5)\n \n}\n\n# > Catch(.4)\n# [1] 0\n# > Catch(.5)\n# [1] 0.5\n# > Catch(.6)\n# [1] 1\n# > Catch(.6,Tolerance=.1)\n# [1] 1\n# > Catch(.6,Tolerance=.2)\n# [1] 0.5\n\n\nweighted.median <- function(x, w, na.rm=TRUE, ties=NULL) {\n # Thanks to https://stat.ethz.ch/pipermail/r-help/2002-February/018614.html\n if (missing(w))\n w <- rep(1, length(x));\n \n # Remove values that are NA's\n if (na.rm == TRUE) {\n keep <- !(is.na(x) | is.na(w));\n x <- x[keep];\n w <- w[keep];\n } else if (any(is.na(x)))\n return(NA);\n \n # Assert that the weights are all non-negative.\n if (any(w < 0))\n stop(\"Some of the weights are negative; one can only have positive\n weights.\");\n \n # Remove values with weight zero. This will:\n # 1) take care of the case when all weights are zero,\n # 2) make sure that possible tied values are next to each others, and\n # 3) it will most likely speed up the sorting.\n n <- length(w);\n keep <- (w > 0);\n nkeep <- sum(keep);\n if (nkeep < n) {\n x <- x[keep];\n w <- w[keep];\n n <- nkeep;\n }\n \n # Are any weights Inf? Then treat them with equal weight and all others\n # with weight zero.\n wInfs <- is.infinite(w);\n if (any(wInfs)) {\n x <- x[wInfs];\n n <- length(x);\n w <- rep(1, n);\n }\n \n # Are there any values left to calculate the weighted median of?\n if (n == 0)\n return(NA);\n \n # Order the values and order the weights accordingly\n ord <- order(x);\n x <- x[ord];\n w <- w[ord];\n \n wcum <- cumsum(w);\n wsum <- wcum[n];\n wmid <- wsum / 2;\n \n # Find the position where the sum of the weights of the elements such that\n # x[i] < x[k] is less or equal than half the sum of all weights.\n # (these two lines could probably be optimized for speed).\n lows <- (wcum <= wmid);\n k <- sum(lows);\n \n # Two special cases where all the weight are at the first or the\n # last value:\n if (k == 0) return(x[1]);\n if (k == n) return(x[n]);\n \n # At this point we know that:\n # 1) at most half the total weight is in the set x[1:k],\n # 2) that the set x[(k+2):n] contains less than half the total weight\n # The question is whether x[(k+1):n] contains *more* than\n # half the total weight (try x=c(1,2,3), w=c(1,1,1)). If it is then\n # we can be sure that x[k+1] is the weighted median we are looking\n # for, otherwise it is any function of x[k:(k+1)].\n \n wlow <- wcum[k]; # the weight of x[1:k]\n whigh <- wsum - wlow; # the weight of x[(k+1):n]\n if (whigh > wmid)\n return(x[k+1]);\n \n if (is.null(ties) || ties == \"weighted\") { # Default!\n (wlow*x[k] + whigh*x[k+1]) / wsum;\n } else if (ties == \"max\") {\n x[k+1];\n } else if (ties == \"min\") {\n x[k];\n } else if (ties == \"mean\") {\n (x[k]+x[k+1])/2;\n } else if (ties == \"both\") {\n c(x[k], x[k+1]);\n }\n}\n\n# > weighted.median(x=c(3,4,5),w=c(.2,.2,.6))\n# [1] 5\n# > weighted.median(x=c(3,4,5),w=c(.2,.2,.5))\n# [1] 5\n# > weighted.median(x=c(3,4,5),w=c(.2,.2,.4))\n# [1] 4.5\n\n\n\nRescale <- function(UnscaledMatrix, ScalingData) {\n # Forces a matrix of raw (user-supplied) information (for example, # of House Seats, or DJIA) to conform to svd-appropriate range.\n # Practically, this is done by subtracting min and dividing by scaled-range (which itself is max-min).\n \n # Calulate multiplicative factors\n InvSpan = ( 1/ ( ScalingData[\"Max\",] - ScalingData[\"Min\",]) )\n \n #Recenter\n OutMatrix <- sweep(UnscaledMatrix, 2, ScalingData[\"Min\",])\n \n #Rescale\n NaIndex <- is.na(OutMatrix) #NA-Preempt\n OutMatrix[NaIndex] <- 0\n OutMatrix <- OutMatrix %*% diag(InvSpan)\n OutMatrix[NaIndex] <- NA #Restore NA's\n \n #Relabel\n row.names(OutMatrix) <- row.names(UnscaledMatrix)\n colnames(OutMatrix) <- colnames(UnscaledMatrix)\n \n return(OutMatrix)\n}\n\n# Scales <- matrix( data=c( 0, 0, 0, 0, 1, 1,\n# 0, 0, 0, 0, 0, 8000,\n# 1, 1, 1, 1, 435, 20000 ), nrow=3, byrow=TRUE)\n# \n# colnames(Scales) <- c(\"C1.1\", \"C2.1\", \"C3.0\", \"C4.0\", \"C5.233\", \"C6.1602759\")\n# row.names(Scales) <- c(\"Scaled\", \"Min\", \"Max\") \n# \n# Scales\n# \n# M <- matrix( data=c(\n# 1, 1, 0, 0, 233, 16027.59,\n# 1, 0, 0, 0, 199, NA,\n# 1, 1, 0, 0, 233, 16027.59,\n# 1, 1, 1, 0, 250, NA,\n# 0, 0, 1, 1, 435, 8001.00,\n# 0, 0, 1, 1, 435, 19999.00),\n# nrow=6,byrow=TRUE)\n# \n# colnames(M) <- c(\"C1.1\", \"C2.1\", \"C3.0\", \"C4.0\", \"C5.233\", \"C6.1602759\")\n# row.names(M) <- paste(\"Voter\",1:6)\n# \n# Rescale(M,Scales)\n# \n# C1.1 C2.1 C3.0 C4.0 C5.233 C6.1602759\n# Voter 1 1 1 0 0 0.5356322 6.689658e-01\n# Voter 2 1 0 0 0 0.4574713 NA\n# Voter 3 1 1 0 0 0.5356322 6.689658e-01\n# Voter 4 1 1 1 0 0.5747126 NA\n# Voter 5 0 0 1 1 1.0000000 8.333333e-05\n# Voter 6 0 0 1 1 1.0000000 9.999167e-01\n\n\n\nInfluence <- function(Weight) {\n # Takes a normalized Vector (one that sums to 1), and computes relative strength of the indicators.\n # this is because by-default the conformity of each Author and Judge is expressed relatively.\n \n Expected <- rep(1/length(Weight),length(Weight))\n return( Weight / Expected)\n}\n\n# > Influence(c(.25,.25,.25,.25))\n# [1] 1 1 1 1\n# > Influence(c(.3,.3,.4))\n# [1] 0.9 0.9 1.2\n# > Influence(c(.99,.0025,.0025,.0025,.0025))\n# [1] 4.9500 0.0125 0.0125 0.0125 0.0125\n\n\nReWeight <- function(Vec,exclude=is.na(Vec)) {\n # Get the relative influence of numbers, treat NA as influence-less\n \n if(!identical(Vec,abs(Vec))) {\n print(\"Warning: Expected all positive.\")\n }\n \n # force well-behaved\n Out <- abs(Vec)\n Out[exclude] <- 0\n \n # division by zero error\n if(sum(Out)==0) Out <- Out + 1\n \n # Reweight\n Out <- Out/sum(Out)\n \n return(Out)\n}\n\n# > ReWeight(c(1,1,1,1))\n# [1] 0.25 0.25 0.25 0.25\n# > ReWeight(c(NA,1,NA,1))\n# [1] 0.0 0.5 0.0 0.5\n# > ReWeight(c(2,4,6,12))\n# [1] 0.08333333 0.16666667 0.25000000 0.50000000\n# > ReWeight(c(2,4,NA,12))\n# [1] 0.1111111 0.2222222 0.0000000 0.6666667\n\n\nReverseMatrix <- function(Mat) {\n #Inverts a binary matrix\n return((Mat-1)*-1)\n}\n\n# M <- matrix(\n# nrow=3,\n# byrow=TRUE,\n# data=c(\n# 0,0,1,\n# 1,1,0,\n# 0,0,0))\n# \n# [,1] [,2] [,3]\n# [1,] 0 0 1\n# [2,] 1 1 0\n# [3,] 0 0 0\n# > ReverseMatrix(M)\n# [,1] [,2] [,3]\n# [1,] 1 1 0\n# [2,] 0 0 1\n# [3,] 1 1 1\n\n\n\nWeightedPrinComp <- function(X, Weights, Verbose=FALSE) {\n # Takes Matrix X and vector of row-weights \"Weights\"\n # Manually computes the statistical procedure known as Principal Components Analysis (PCA)\n # This version of the procedure is so basic, that it can also be thought of as merely a singular-value decomposition on a weighted covariance matrix.\n \n if(missing(Weights)) {\n Weights <- ReWeight(rep(1,nrow(X)))\n } \n \n if(length(Weights)!=nrow(X)) {\n print(\"Error: Weights must be equal to nrow(X)\")\n return(NULL)\n }\n \n wCVM <- cov.wt(x=X, wt=Weights) # Weighted Covariance Matrix\n \n # http://en.wikipedia.org/wiki/Singular_value_decomposition\n L <- svd(wCVM$cov)$u[,1] # extract first Loading (first column of U matrix)\n S <- as.vector( scale(X, center=wCVM$center, scale=FALSE) %*% L) # manually calculate the first Score, using the input matrix, covariance matrix, and first loading\n # (center subtracts the weight-adjusted-means from X's columns)\n # (scale=FALSE means that no scaling is done)\n # %*% is matrix multiplication\n \n if(Verbose) {\n Ls <- svd(wCVM$cov)$u # the entire u matrix\n Ss <- scale(X, center=wCVM$center, scale=FALSE) %*% Ls # all of the scores\n print(\"Loadings: \")\n print(Ls)\n print(\" \")\n print(\"Scores:\")\n print(Ss)\n print(\" \")\n }\n \n Out <- list(\"Scores\"=S,\"Loadings\"=L)\n return(Out)\n}\n\n# M <- matrix(\n# nrow=3,\n# byrow=TRUE,\n# data=c(\n# 0,0,1,\n# 1,1,0,\n# 0,0,0))\n# \n# M2 <- cbind(M,c(.7,.5,.2))\n\n# > WeightedPrinComp(M)\n# $Scores\n# [1] 0.7251092 -0.9905177 0.2654084\n# $Loadings\n# [1] -0.6279630 -0.6279630 0.4597008\n# \n# \n# > WeightedPrinComp(M,c(.33333,.33333,.33333))\n# $Scores\n# [1] 0.7251092 -0.9905177 0.2654084\n# $Loadings\n# [1] -0.6279630 -0.6279630 0.4597008\n# \n# > WeightedPrinComp(M,c(.1,.1,.8))\n# $Scores\n# [1] 0.2762801 -1.2732354 0.1246194\n# $Loadings\n# [1] -0.6989274 -0.6989274 0.1516607\n# \n# > WeightedPrinComp(M2)\n# $Scores\n# [1] 0.7385624 -0.9866042 0.2480418\n# $Loadings\n# [1] -0.62428014 -0.62428014 0.46733010 0.04638101\n# \n# > WeightedPrinComp(M2,c(.1,.1,.8))\n# $Scores\n# [1] 0.1269693 -1.2954637 0.1460618\n# $Loadings\n# [1] -0.69461121 -0.69461121 0.06807931 -0.17434371\n", "meta": {"hexsha": "799d79e4ec20bbe1ff90c01679d8df98a3ddb7be", "size": 10326, "ext": "r", "lang": "R", "max_stars_repo_path": "lib/consensus/CustomMath.r", "max_stars_repo_name": "endolith/Truthcoin", "max_stars_repo_head_hexsha": "448b35fb94f27e61f5989ead7ef87e03da2e9237", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 161, "max_stars_repo_stars_event_min_datetime": "2015-01-11T20:52:37.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-14T04:44:13.000Z", "max_issues_repo_path": "lib/consensus/CustomMath.r", "max_issues_repo_name": "endolith/Truthcoin", "max_issues_repo_head_hexsha": "448b35fb94f27e61f5989ead7ef87e03da2e9237", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2016-04-21T10:17:06.000Z", "max_issues_repo_issues_event_max_datetime": "2022-01-09T14:38:06.000Z", "max_forks_repo_path": "lib/consensus/CustomMath.r", "max_forks_repo_name": "endolith/Truthcoin", "max_forks_repo_head_hexsha": "448b35fb94f27e61f5989ead7ef87e03da2e9237", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 40, "max_forks_repo_forks_event_min_datetime": "2015-01-19T16:44:14.000Z", "max_forks_repo_forks_event_max_datetime": "2022-01-09T14:09:49.000Z", "avg_line_length": 26.5449871465, "max_line_length": 167, "alphanum_fraction": 0.5746658919, "num_tokens": 4056, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7057850278370111, "lm_q2_score": 0.7371581568543044, "lm_q1q2_score": 0.520275190255695}} {"text": "## * Utilities\nupper1st <- function(x) {\n if (is.factor(x))\n x0 <- levels(x) else x0 <- as.character(x)\n nc <- nchar(x0)\n x0[nc > 1] <- sprintf('%s%s', toupper(substr(x0[nc > 1], 1, 1)), substr(x0[nc > 1], 2, nchar(x0[nc > 1])))\n x0[nc == 1] <- toupper(x0[nc == 1])\n if (is.factor(x))\n levels(x) <- x0 else x <- x0\n return(x)\n}\n\n## * Statistical\nestBetaParams <- function(mu, var) {\n ## Estimate alpha and beta of beta distribution from mean and variance\n ## from http://stats.stackexchange.com/questions/12232/calculating-the-parameters-of-a-beta-distribution-using-the-mean-and-variance\n alpha <- ((1 - mu) / var - 1 / mu) * mu ^ 2\n beta <- alpha * (1 / mu - 1)\n return(params = list(alpha = alpha, beta = beta))\n}\n\n\n## * For PST\n## formula from Neil & Lebreton 2008\nfunNL<- function(x,a,s) (exp((a+s/(x-s))^-1)-x)^2\nlmax_nl <- function(a,s) return(optimise(funNL, c(1,2), a=a, s=s, tol=1e-10)$minimum)\nLmax_nl <- function(a,s,usemc=F) {\n if (usemc) {\n library(parallel)\n return(mcmapply(lmax_nl, a, s))\n } else return(mapply(lmax_nl, a, s))\n}\nRmax_NL <- function(s,a,...) return(Lmax_nl(a,s,...)-1)\n", "meta": {"hexsha": "28cd612a088367f446bd00c72cecc01573f11c2f", "size": 1169, "ext": "r", "lang": "R", "max_stars_repo_path": "functions.r", "max_stars_repo_name": "dragonfly-science/seabird-risk-assessment", "max_stars_repo_head_hexsha": "97d1bd13d3b3eee88ab9b9428dfc6eafc1acd305", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2018-02-22T20:16:08.000Z", "max_stars_repo_stars_event_max_datetime": "2018-02-22T20:16:08.000Z", "max_issues_repo_path": "functions.r", "max_issues_repo_name": "seabird-risk-assessment/seabird-risk-assessment", "max_issues_repo_head_hexsha": "97d1bd13d3b3eee88ab9b9428dfc6eafc1acd305", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 21, "max_issues_repo_issues_event_min_datetime": "2017-10-09T08:18:07.000Z", "max_issues_repo_issues_event_max_datetime": "2017-10-29T22:13:43.000Z", "max_forks_repo_path": "functions.r", "max_forks_repo_name": "dragonfly-science/seabird-risk-assessment", "max_forks_repo_head_hexsha": "97d1bd13d3b3eee88ab9b9428dfc6eafc1acd305", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2018-04-18T22:56:04.000Z", "max_forks_repo_forks_event_max_datetime": "2020-10-16T13:57:07.000Z", "avg_line_length": 34.3823529412, "max_line_length": 136, "alphanum_fraction": 0.5893926433, "num_tokens": 399, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7772998508568416, "lm_q2_score": 0.6688802603710086, "lm_q1q2_score": 0.5199205266274703}} {"text": "# This file includes a demonstration of how to use the method \"svl\" to identify the source\n# vent location of tephra fall deposits based on thickness or maximum clast size\n# measurements. \n# Author: Qingyuan Yang, Marcus Bursik, and E. Bruce Pitman.\n# GPL: License. Use at your own risk. \n###\n# The work was supported by National Science Foundation Hazard SEES grant number 1521855 \n# to G. Valentine, M. Bursik, E.B. Pitman and A.K. Patra, and National Science Foundation\n# DMS grant number 1621853 to A.K. Patra, M. Bursik, and E.B. Pitman.\n# We appreciate your comments, suggestions, and feedback. \n# Please feel free to contact us through vhub or email: qyang5@buffalo.edu \n########################################## Demonstration ################################\n# In this demonstration, we use a thickness dataset of the North Mono Bed 2 \n# (Sieh and Bursik, 1986) as an example. North Mono Bed 2 was erupted from Upper Dome, which is a\n# part of the Mono-Inyo Craters (eastern central California).\n# It is noted that the method can also be applied to maximum clast size measurements.\n\n# Set working directory.\nsetwd(\"/the/directory/where/you/put/your/data/\") \n\n#---\n\n# The input dataset should have three columns:\n#\tThe first two columns are the coordinates (x or E and y or N) of sample sites in the UTM system.\n#\tThe third column should include the corresponding thickness or maximum clast size \n#\tmeasurements in millimeters, with a minimum value of 1 mm.\n# The input dataset should be a \".csv\" file, and separated by \",\".\n# The head or column names of the three columns should be \"x\", \"y\", and \"zr\".\n# In the case of North Mono Bed 2, the x and y coordinates are in UTM Zone 11. \n# Read dataset from the working directory. \ntd = read.table(\"nmb2.csv\", header=TRUE, sep = \",\")\n\n#--- \n\n# Check data structure (optional command).\nstr(td)\n# Return:\n\t#'data.frame': 114 obs. of 3 variables:\n\t#$ x : num 320364 328807 327410 329283 327802 ...\n\t#$ y : num 4194955 4192135 4193500 4194283 4194929 ...\n\t#$ zr: num 0 0 0 0 0 0 0 0 0 0 ...\n# NOTE: if the data include observations of zero thickness; it is necessary to exclude them for the \n#\tmethod to work (optional command).\n#\n#\n#---\n\n# Exclude zero-thickness observations.\ntd = subset(td, zr!=0)\n \n#---\n\n# Check data structure in a different way (optional command).\n# Command below shows the first six rows of the dataset.\nhead(td)\n# Return:\n\t# x y zr\n\t#12 326341.2 4199522 22\n\t#15 327685.3 4203216 10\n\t#26 320851.0 4197091 38\n\t#27 321276.5 4197466 82\n\t#28 320946.3 4197606 96\n\t#29 321606.7 4197460 78\n# Note: the 1st column with # is just sample numbers; these are not\n# part of the dataset. The data are not in order (1, 2, 3,...)\n# because zero-thickness observations have been excluded. \n\n#---\n\n# Define the initial guess of source vent location.\n# These coordinates should be within the area where the vent is likely \n# to be. \n# We recommend users to try different coordinates within that area to \n# run the method,\n# and then check if the results converge to the same point.\nsv_assumed = cbind(x = 322227, y = 4181034)\n# In the example case, assuming that we known the vent is from the Mono-Inyo Craters,\n# the values of the initial guess are the coordinates of Obsidian Dome, which is > 14 km\n# south of Upper Dome, the true vent location of North Mono Bed 2.\n\n#---\n\n# Source functions of the method \"svl\".\n# Note: the semi-empirical model proposed by Yang and Bursik (2016) \n# is used here.\nsource(\"/where/you/put/your/source/code/pub_svl_2.0_exponential.r\")\n\n#---\n\n# Run the method \"svl\".\n\n# Brief description of the method:\n#\tStarting with an initial guess, \"sv_assumed\", on vent location, the\n#\tvalue is updated in each iteration (loop).\t\t\t\t\t \n# Start loop:\t\t\t\t\t\t\t\t <----------------|\n#\tThe method proposes different possible vent locations around \"sv_assumed\" \t |\t\n#\tthat are at a certain distance (distance: h = 1000 m in this case) from it, |\n#\tand compares if they are closer or farther to the true vent compared \t\t |\n#\twith \"sv_assumed\".\t\t\t\t\t\t\t\t |\n# \t\t\t\t\t\t\t\t\t\t |\n#\tIf \"sv_assumed\" is closer to the true vent: |\n#\t\twe shrink the search radius (controlled by r * h, 0.7 * 1000 = 700 here) |\n#\t\tto improve the resolution,\t\t\t\t\t\t |\n#\t\tand do the same thing with the smaller search radius; --------------->----/\n#\tIf one of the nearby points is closer to the true vent:\t\t\t\t |\t\n#\t\tthe guess on vent location, \"sv_assumed\", \t\t\t\t |\n#\t\tis updated/replaced (see function \"compare\" for more details),\t\t |\n#\t\tand with the same search radius, the method does\t\t\t |\n#\t\tthe same thing with the new guess on vent location.------------------>----/\n#\t\n#\tIn the actual implementation of the method, the search radius is updated\n#\tin a slightly more efficient manner (see detailed description in our work, whose \n#\tlink will be provided soon). \n#\tThis does not affect the notation in this file.\n#\n#\tAs described above, after each iteration, \n#\tthe search radius gets either smaller or unchanged. \n#\tWhen the search radius is sufficiently small, the vent is found.\n# End loop.\n\n#\tThe number of iterations is controlled by \"runs\".\n#---\n# How to assign values to the input parameters:\n#\tsv = sv_assumed, the initial guess on the source vent location;\n#\ttd: the input dataset;\n#\truns: the number of iterations; \n#\th: the initial search radius (in meters);\n#\tr: the shrink ratio in the search radius. \n#\t\tSuggested value: 0.7\n#\tnlps, h_ratio: these are related to estimating the wind direction in each\n#\t\titeration. It is sufficient to keep them fixed as 20 and 1, respectively. \n#\tnumb: the number of rows of data points within \"td\" that will be used as input. \n#\t To use the complete dataset, set it to \"1:nrow(td)\".\n#---\n#***\tValues of \"runs\" and \"h\" need to be selected with care. ***\n#\tHere we use the example of the North Mono Bed 2 to show how the values are determined.\n#\n# \tAssuming that we know the tephra was erupted from the Mono-Inyo Craters, \n#\twhich cover a range of >15 km in the y direction, and about 5 km in the x-direction.\n#\n#\tBy specifying the initial search radius to be 1000 m (h = 1000), we ensure that it takes at most \n#\t~15 steps/iterations before the method shrinks the search radius. \n#\tBy the time it shrinks the search radius, it can be assumed approximately that the true vent\n#\tis within ~1500 m of the current point. \n#\n# \tConsider the worst-case scenario, namely the search radius keeps shrinking. This means that\n#\tthe searching resolution is improving in each iteration at the lowest rate.\n#\tIn this case, the initial search radius keeps shrinking at a rate of 0.7 (r = 0.7). \n#\tFor the resolution to be at ~10 m scale, the method requires 13 more iterations: 1000*(0.7)^13 = 9.61.\n# \tBased on this, it is ensured that runs = 40 (greater than 15+13) \n#\titerations are sufficient.\n#\n# \tBased on the argument above, we recommend users to follow this strategy to assign values of\n#\t\"h\" and \"runs\":\n#\th = (the maximum height or width of the region of interest)/10\n#\tFor \"runs\", if we want to reach the solution at a ~10 m scale, then \n#\twe need to find out the value \"x\" in the brackets of the function: h*0.7^(x) = ~10. \n#\tAnd then set \"runs = 10 + x + E\", where E could be an integer ranging from 5 to 20, a fudge-factor. \n#\tGreater value in E means more additional iterations, \n#\twhich is likely to yield a more accurate estimate. \nresult_linear = gd_simplified(sv = sv_assumed, td, runs = 40, h = 1000, r = 0.7, nlps = 20, h_ratio = 1, numb = 1:nrow(td))\n\n# Check the output.\nresult_linear\n# Return:\n # x y output_ang h (Intercept) dist dd ssr rsquare\n #322224.4 4197601 172.3507 0.06281024 2.231973 -0.0003055393 -0.0002597007 2.877154 0.9419849\n \n# Interpretation:\n#\tx and y: \testimated vent coordinates of the North Mono Bed 2. This location is within the extent of \n# \t\t\tits true source, Upper Dome.\n#\toutput_ang: \testimated wind direction (from north clockwise). In this case, it is the UPWIND direction.\n#\t\t\tSee note below on how to tell if the estimated wind direction\n#\t\t\tis downwind or upwind.\n#\th:\t\tthe length of the search radius (m) from the last iteration. If this value is small\n#\t\t\tenough (e.g., <10), then the method converged. \n#\t\t\tIf this value is comparable to the initial search radius (say 490 = 1000*0.7*0.7), \n#\t\t\tthen users should try to increase the number of iterations, \n#\t\t\tnamely increase the value \"runs\" (e.g., set it to be 60),\n#\t\t\tand run the method again.\n#\t\t\t\tIf the method is run with more iterations, \n#\t\t\t\tbut the resultant \"h\" is still comparable \n#\t\t\t\tto the initial search radius, that means the method fails\n#\t\t\t\tto converge. \n#\t\t\t\tThen users should try to change the initial\n#\t\t\t\tguess on vent location (sv_assumed), and run the method.\n#\t(intercept):\tfitted coefficients for the semi-empirical model.\n#\tdist:\t\tsame as above.\t\n#\tdd:\t\tsame as above.\n#\tssr:\t\tsum of squared residuals from the fitting given estimated vent location and \n#\t\t\twind direction.\n#\trsquare:\tr-squared value for the final estimate given estimated vent location and wind direction.\n# \n# Given sparse data, it is important to check if the fitted coefficients are physical.\n# For the exponential model, this can be done by checking if dist+dd<0. If their sum dist+dd > 0 or dist > 0, this means that \n# along the dispersal direction, the thickness or maximum clast size increases with distance, which is generally unphysical. \n\n#------------------------------------------\n# Use the semi-empirical (power-law) model proposed by Gonzalez-Mellado and De la Cruz-Reyna (2010).\n# Source the functions of the method \"svl\".\nsource(\"/where/you/put/your/source/code/pub_svl_2.0_power-law.r\")\n\n#---\n\n# Run the method \"svl\" with identical inputs.\nresult_power = gd_simplified(sv = sv_assumed, td, runs = 40, h = 1000, r = 0.7, nlps = 20, h_ratio = 1, numb = 1:nrow(td))\n# Check the result.\nresult_power\n# Return:\n# \tx y output_ang h (Intercept) dif log(dist) ssr rsquare\n# \t322859.7 4198078 337.9795 0.5338782 5.194675 -0.0001700573 -0.4008759 2.084604 0.9579659\n\n# Interpretation:\n# \tSee lines 166-190.\n#\tThe estimated vent location is within the area of Upper Dome, the true vent of \n#\tthis tephra deposit.\n#\t### the output_ang in this case is the DOWNwind direction. \n#\tSee note below on how to tell if the estimated wind direction is downwind or upwind.\n# \tIf log(dist) > 0 or (Intercept) < 0, then unphysical prediction occurs.\n\n######\n\n#---\n\n# Note on how to tell if the estimated wind direction (output_ang) is downwind or upwind:\n# Due to the design of \"svl\", the output \"output_ang\" could be either the upwind or downwind direction.\n# This is related to the gradient descent method adopted in \"svl\". \n# To tell if it is upwind or downwind,\n# users need to examine the relationship between the fitted coefficients:\n\n\t# If \"svl\" is combined with the exponential method of Yang and Bursik (2016):\n\t#\tif dd < 0, then \"output_ang\" is the upwind direction;\n\t#\tif dd > 0, then \"output_ang\" is the downwind direction.\n\n\t# If \"svl\" is combined with the power-law method of Gonzalez-Mellado and De la Cruz-Reyna (2010):\n\t#\tif dif < 0, then \"output_ang\" is the downwind direction;\n\t#\tif dif > 0, then \"output_ang\" is the upwind direction.\n\n# References:\n#\tGonzalez-Mellado, A. O., and S. De la Cruz-Reyna. \"A simple semi-empirical approach to model thickness of \n# ash-deposits for different #eruption scenarios.\" Natural Hazards and Earth System Sciences 10.11 (2010): 2241.\n# \tSieh K, Bursik M. Most recent eruption of the Mono Craters, eastern central California. Journal of Geophysical \n# Research: Solid Earth, 1986, 91(B12): 12539-12571.\n# \tYang Q, Bursik M. A new interpolation method to model thickness, isopachs, extent, and volume of tephra fall \n# deposits. Bulletin of Volcanology, 2016, 78(10): 68.\n", "meta": {"hexsha": "5d035663ec8f84e18f5910c1336d90d0d374ff74", "size": 11928, "ext": "r", "lang": "R", "max_stars_repo_path": "pub_demo_svl_2.0.r", "max_stars_repo_name": "yiqioyang/svl", "max_stars_repo_head_hexsha": "12f949a7f944fd6e4158ca4082035271185126f1", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "pub_demo_svl_2.0.r", "max_issues_repo_name": "yiqioyang/svl", "max_issues_repo_head_hexsha": "12f949a7f944fd6e4158ca4082035271185126f1", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "pub_demo_svl_2.0.r", "max_forks_repo_name": "yiqioyang/svl", "max_forks_repo_head_hexsha": "12f949a7f944fd6e4158ca4082035271185126f1", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 48.487804878, "max_line_length": 126, "alphanum_fraction": 0.6981891348, "num_tokens": 3323, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7981867777396211, "lm_q2_score": 0.651354857898194, "lm_q1q2_score": 0.5199028351908083}} {"text": "#' Calculate the proportion of observations that lie within a given bandwidth\n#' \n#' Facilitates a \\code{knn}-type bandwidth by calculating the proportion \\eqn{sum(w) / n}. Called repeatedly by \\code{optimize} in \\code{lagr.dispatch} to home in on the desired \\code{knn}.\n#' \n#' @param bw bandwidth for which to compute \\code{sum(w)}\n#' @param loc location around which to center the kernel\n#' @param coords matrix of observation locations\n#' @param dist vector of distances from central location to the observation locations\n#' @param kernel kernel function for generating the local observation weights\n#' @param target targeted \\code{knn} bandwidth\n#' @param prior.weights vector of prior observation weights provided by the user\n#' @param total.weight sum of prior weights\n#' @param verbose print detailed information about our progress?\n#' \n#' @return difference between the calculated \\code{sum(w) / total.weight} and the target\n#' \nlagr.knn = function(bw, loc, coords, dist, kernel, target, prior.weights, total.weight, verbose) {\n kernel.weights = kernel(dist, bw)\n w = kernel.weights * prior.weights\n prop = sum(w)/total.weight\n\n #return the difference between the calculated sum(w) and the target\n return(abs(prop-target))\n}\n", "meta": {"hexsha": "33e813b676b0786004fb538aded1a98ea0ce7f5b", "size": 1249, "ext": "r", "lang": "R", "max_stars_repo_path": "R/lagr.knn.r", "max_stars_repo_name": "wrbrooks/lagr", "max_stars_repo_head_hexsha": "7ac867f0bf091e8835917205cab44ea806d3c65e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/lagr.knn.r", "max_issues_repo_name": "wrbrooks/lagr", "max_issues_repo_head_hexsha": "7ac867f0bf091e8835917205cab44ea806d3c65e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/lagr.knn.r", "max_forks_repo_name": "wrbrooks/lagr", "max_forks_repo_head_hexsha": "7ac867f0bf091e8835917205cab44ea806d3c65e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 49.96, "max_line_length": 189, "alphanum_fraction": 0.7453963171, "num_tokens": 284, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.798186768138228, "lm_q2_score": 0.651354857898194, "lm_q1q2_score": 0.5199028289368942}} {"text": "Edge Path 0:\na -> b (5)\nb -> c (6)\nc -> a (7)\na -> d (1)\nd -> c (2)\nc -> e (3)\ne -> a (4)\n", "meta": {"hexsha": "7851d6c5de0b2a2438b3458442a71b799ec36e9d", "size": 90, "ext": "r", "lang": "R", "max_stars_repo_path": "Demos/QuikGraph/Tests/Cases/Observers/EdgePredecessor/Paths/02.r", "max_stars_repo_name": "TXCodeDancer/OpenSource", "max_stars_repo_head_hexsha": "18c6442be2a8b6459a46c5021dd9b68698811811", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Demos/QuikGraph/Tests/Cases/Observers/EdgePredecessor/Paths/02.r", "max_issues_repo_name": "TXCodeDancer/OpenSource", "max_issues_repo_head_hexsha": "18c6442be2a8b6459a46c5021dd9b68698811811", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Demos/QuikGraph/Tests/Cases/Observers/EdgePredecessor/Paths/02.r", "max_forks_repo_name": "TXCodeDancer/OpenSource", "max_forks_repo_head_hexsha": "18c6442be2a8b6459a46c5021dd9b68698811811", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 10.0, "max_line_length": 12, "alphanum_fraction": 0.3333333333, "num_tokens": 47, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6992544085240401, "lm_q2_score": 0.7431680029241321, "lm_q1q2_score": 0.5196635023187061}} {"text": "#\n# # Versions of pcond vectors for R-vine\n# # with different pair-copula family for each edge of the vine\n# # for each row u[1:d] of hypercube data,\n# # return C_{2|1}(u2|u1), C_{3|12}(u3|u1,u2), ... C_{d|1..d-1}(ud|u1,...u[d-1])\n#\n# # parvec = vector of parameters of pair-copulas\n# # udat = nxd matrix with uniform scores\n# # A = dxd vine array with 1:d on diagonal\n# # ntrunc = truncated level, assume >=1\n# # pcondmat = matrix of names of conditional cdfs for trees 1,...,ntrunc\n# # (assuming only one needed for permutation symmetric pair-copulas)\n# # pcondmat is empty for diagonal and lower triangle,\n# # and could have ntrunc rows or be empty for rows ntrunc+1 to d-1\n# # np = dxd where np[ell,j] is size for parameter th[ell,j]\n# # for pair-copula in tree ell, variables j and A[ell,j]\n# # np=0 on and below diagonal\n# # Output:\n# # return C_{2|1}(u2|u1), C_{3|12}(u3|u1,u2), ... C_{d|1..d-1}(u[d]|u[1:(d-1)])\n# #rvinellkv.trunc2=function(parvec,udat,A,ntrunc,logdcopmat,pcondmat,np)\n# rvinepcond.trunc=function(parvec,udat,A,ntrunc,pcondmat,np)\n# { d=ncol(A) # or ncol(udat)\n# npmax=max(np); th=array(0,c(npmax,d,d)) # format of parameter for vine\n# ii=0;\n# for(ell in 1:ntrunc)\n# { for(j in (ell+1):d)\n# { ipp=1:np[ell,j]\n# th[ipp,ell,j]=parvec[ii+ipp]\n# ii=ii+np[ell,j]\n# }\n# }\n# out=varray2M(A)\n# M=out$mxarray\n# icomp=out$icomp\n# n=nrow(udat)\n# #llkv=rep(0,n)\n# vinepcond=matrix(0,n,d)\n# vinepcond[,1]=udat[,1]\n# v=matrix(0,n,d); vp=matrix(0,n,d); s=matrix(0,n,d);\n# nllk=0\n# # tree 1\n# #for(j in 2:d)\n# #{ ipp=1:np[1,j]\n# #logdcop=match.fun(logdcopmat[1,j])\n# #llkv=llkv+logdcop(udat[,A[1,j]],udat[,j],th[ipp,1,j])\n# #}\n# # tree 2\n# if(ntrunc>=2)\n# { for(j in 2:d)\n# { ipp=1:np[1,j]\n# pcond=match.fun(pcondmat[1,j])\n# if(icomp[1,j]==1) vp[,j]=pcond(udat[,A[1,j]],udat[,j],th[ipp,1,j])\n# v[,j]=pcond(udat[,j],udat[,A[1,j]],th[ipp,1,j])\n# }\n# vinepcond[,2]=v[,2]\n# for(j in 3:d)\n# { if(A[2,j]=3)\n# { for(ell in 3:ntrunc)\n# { for(j in ell:d)\n# { ipp=1:np[ell-1,j]\n# pcond=match.fun(pcondmat[ell-1,j])\n# if(icomp[ell-1,j]==1) vp[,j]=pcond(s[,j],w[,j],th[ipp,ell-1,j])\n# v[,j]=pcond(w[,j],s[,j],th[ipp,ell-1,j])\n# }\n# vinepcond[,ell]=v[,ell]\n# for(j in (ell+1):d)\n# { if(A[ell,j]1 & ntrunc .5*pi] <- theta[dtheta > .5*pi] - pi\n theta[dtheta < -.5*pi] <- pi + theta[dtheta < -.5*pi]\n if ( any(abs(dtheta) > .5*pi) )\n {\n theta <- convertThetaIter(theta, thetabar)\n }\n return(theta)\n}\n\n## convertThetaNew <- function(theta, thetabar)\n## {\n## dtheta <- theta - thetabar\n## if ( any(abs(dtheta) > pi ) )\n## {\n## cat(\"something wrong with theta!\\n\")\n## exit()\n## }\n## theta[dtheta > .5*pi] <- theta[dtheta > .5*pi] - pi\n## theta[dtheta < -.5*pi] <- pi + theta[dtheta < -.5*pi]\n## if ( any(abs(dtheta) > pi) )\n## {\n## cat(\"something wrong with theta 2!\\n\")\n## exit()\n## }\n## return(theta)\n## }\n\n## convertTheta <- function(theta)\n## {\n## if ( any(abs(theta) > pi ) )\n## {\n## cat(\"something wrong with theta!\\n\")\n## exit()\n## }\n## theta[theta > .5*pi] <- theta[theta > .5*pi] - pi\n## theta[theta < -.5*pi] <- pi + theta[theta < -.5*pi]\n## if ( any(abs(theta) > pi) )\n## {\n## cat(\"something wrong with theta 2!\\n\")\n## exit()\n## }\n## return(theta)\n## }\n\nsimu <- function(lambda2, k1, k2, alpha, ec, A, index)\n{\n dt <- 0.01\n a0 <- 0.01\n e0 <- 1.0\n lambda1 <- 1\n Ncell <- 50\n Nstep <- 40000\n v <- a0*A\n DispWindow <- seq(28000, 40000)\n ##Selection <- 28000 + 2400*seq(5)\n Selection <- 28000 + 100*seq(120)\n\n nt <- Ncell\n lt <- 1\n output1 <- NULL\n\n e <- seq(1.2, 1.2, length.out=Ncell)\n theta <- runif(Ncell, -pi/2, pi/2)\n thetabar <- 0\n S <- 0\n \n ## ############### model I\n for ( i in seq(Nstep) )\n {\n if ( i %% 20 == 0 )\n {\n n <- cbind(cos(theta), sin(theta))\n Q <- cbind(n[,1]*n[,1]-0.5, n[,1]*n[,2], n[,2]*n[,1], n[,2]*n[,2]-0.5)\n Q <- colSums((e-1)*Q)/sum((e-1))\n M <- matrix(Q, nrow=2, ncol=2, byrow=TRUE)\n ##M <- matrix(colMeans(Q), nrow=2, ncol=2, byrow=TRUE)\n eigenvec <- eigen(M)$vectors[,1]\n S <- 2.0*eigen(M)$values[1]\n if(eigenvec[1] < 0)\n {\n eigenvec <- -eigenvec\n }\n thetabar <- sign(eigenvec[2])*atan(abs(eigenvec[2])/pmax(0.000001, abs(eigenvec[1])))\n }\n Lt <- exp(v*dt*i)\n Force <- max(0, alpha*(Lt-lt))\n e <- e + dt*(Force*abs(cos(theta)) - lambda1*(e-1)) + k1*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n e <- pmax(e, 1)\n theta <- theta - dt*lambda2*S*(e-1)*(theta-thetabar) + k2*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n ##theta <- theta - dt*lambda2*S*(theta-thetabar) + k2*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n theta <- convertThetaIter(theta, thetabar)\n ##theta <- atan2(sin(theta), cos(theta))\n \n theta1 <- atan(abs(sin(theta))/pmax(0.000001, abs(cos(theta))))\n ndiv=rbinom(1, nt, dt*a0)\n divtheta <- sample(theta1, ndiv)\n lt<-lt*(1+sum(cos(divtheta))/nt)\n nt<-nt+ndiv\n theta<-c(theta, runif(ndiv, -pi/2, pi/2))\n e<-c(e, rep(e0, ndiv))\n A1 <- mean(cos(theta1))\n A2 <- mean(cos(theta1)) - mean(sin(theta1))\n\n output1 <- rbind(output1, c(Lt, lt, A1, A2, Force, nt, thetabar, S))\n }\n\n eModelI <- e\n thetaModelI <- theta\n \n################################ model II\n output2 <- NULL\n nt=Ncell\n lt=1\n e <- seq(1.2, 1.2, length.out=Ncell)\n theta <- runif(Ncell, -pi/2, pi/2)\n\n for ( i in seq(Nstep) )\n {\n if ( i %% 20 == 0 )\n {\n n <- cbind(cos(theta), sin(theta))\n Q <- cbind(n[,1]*n[,1]-0.5, n[,1]*n[,2], n[,2]*n[,1], n[,2]*n[,2]-0.5)\n Q <- colSums((e-1)*Q)/sum((e-1))\n M <- matrix(Q, nrow=2, ncol=2, byrow=TRUE)\n ##M <- matrix(colMeans(Q), nrow=2, ncol=2, byrow=TRUE)\n eigenvec <- eigen(M)$vectors[,1]\n S <- 2.0*eigen(M)$values[1]\n if(eigenvec[1] < 0)\n {\n eigenvec <- -eigenvec\n }\n thetabar <- sign(eigenvec[2])*atan(abs(eigenvec[2])/pmax(0.000001, abs(eigenvec[1])))\n }\n Lt <- exp(v*dt*i)\n Force <- max(0, alpha*(Lt-lt))\n e <- e + dt*(Force*abs(cos(theta)) - lambda1*(e-1)) + k1*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n e <- pmax(e, 1)\n theta <- theta - dt*lambda2*S*(e-1)*(theta-thetabar) + k2*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n ##theta <- theta - dt*lambda2*S*(theta-thetabar) + k2*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n theta <- convertThetaIter(theta, thetabar)\n ##theta <- atan2(sin(theta), cos(theta))\n \n theta1 <- atan(abs(sin(theta))/pmax(0.000001, abs(cos(theta))))\n ecut <- e < ec\n theta2 <- theta1 * (1 - ecut) + ecut * runif(nt, 0, pi/2)\n \n ndiv=rbinom(1, nt, dt*a0)\n \n divtheta <- sample(theta2, ndiv)\n lt<-lt*(1+sum(cos(divtheta))/nt)\n nt<-nt+ndiv\n theta<-c(theta, runif(ndiv, -pi/2, pi/2))\n e<-c(e, rep(e0, ndiv))\n A1 <- mean(cos(theta1))\n A2 <- mean(cos(theta1)) - mean(sin(theta1))\n \n output2<-rbind(output2, c(Lt, lt, A1, A2, Force, nt, thetabar, S))\n }\n\n\n ##if ( index <= 3 )\n {\n png(paste(\"plotlt\", \"_A_\", A, \"_alpha_\", alpha,\n \"_lambda2_\", lambda2,\n \"_k1_\", k1, \"_k2_\", k2,\n \"_\", index, \".png\", sep=''),\n width=1200, height=600)\n par(mfrow=c(2,5))\n plot(eModelI, thetaModelI*180/pi, ylim=c(-90,90))\n abline(h=c(-30,30))\n plot(output1[DispWindow, 1]-output1[DispWindow, 2],\n col = 'blue', ylab=\"L(t)-l(t)\",\n ylim=c(-1,1),\n xlab=\"timesteps\", cex=0.5\n )##, ylim=c(-1,1), cex=0.5)\n abline(h=0, lwd = 2, col='grey')\n plot(output1[DispWindow,5], cex=0.25)\n plot(output1[DispWindow, 3], col='blue', ylim=c(-1, 1))\n lines(output1[DispWindow, 4], col='red')\n lines(output2[DispWindow, 7], col='green', pch=5, type='p') \n abline(h=c(-0.3))\n plot(output1[DispWindow, 8], col='blue', ylim=c(0, 1)) \n\n plot(e, theta*180/pi, ylim=c(-90,90))\n abline(h=c(-30,30))\n plot(output2[DispWindow, 1]-output2[DispWindow, 2],\n col = 'blue', ylab=\"L(t)-l(t)\",\n ylim=c(-1,1),\n xlab=\"timesteps\", cex=0.5\n )\n abline(h=0, lwd = 2, col='grey')\n plot(output2[DispWindow,5], cex=0.25)\n plot(output2[DispWindow, 3], col='blue', ylim=c(-1, 1))\n lines(output2[DispWindow, 4], col='red')\n lines(output2[DispWindow, 7], col='green', pch=5, type='p')\n abline(h=c(-0.3))\n plot(output2[DispWindow, 8], col='blue', ylim=c(0, 1)) \n\n dev.off()\n }\n\n simu <- NULL\n simu$model1 <- output1[Selection, 1] - output1[Selection, 2]\n simu$model2 <- output2[Selection, 1] - output2[Selection, 2]\n return(simu)\n}\n\nbaseModels <- function(lambda2, k1, k2, ec, index, Force)\n{\n dt <- 0.01\n a0 <- 0.01\n e0 <- 1.0\n lambda1 <- 1\n Ncell <- 2000\n Nstep <- 10000\n\n nt <- Ncell\n e <- seq(1.2, 1.2, length.out=Ncell)\n theta <- runif(Ncell, -pi/2, pi/2)\n thetabar <- 0.0\n S <- 0\n \n ## ############### model I\n for ( i in seq(Nstep) )\n {\n if ( i %% 20 == 0 )\n {\n n <- cbind(cos(theta), sin(theta))\n Q <- cbind(n[,1]*n[,1]-0.5, n[,1]*n[,2], n[,2]*n[,1], n[,2]*n[,2]-0.5)\n Q <- colSums((e-1)*Q)/sum((e-1))\n M <- matrix(Q, nrow=2, ncol=2, byrow=TRUE)\n eigenvec <- eigen(M)$vectors[,1]\n S <- 2.0*eigen(M)$values[1]\n if(eigenvec[1] < 0)\n {\n eigenvec <- -eigenvec\n }\n if (abs(eigenvec[1]) < 0.000001)\n {\n thetabar <- sign(eigenvec[2])*pi/2\n }\n else\n {\n thetabar <- sign(eigenvec[2])*atan(abs(eigenvec[2])/pmax(0.000001, abs(eigenvec[1])))\n }\n ##cat(S, thetabar, '\\n')\n }\n e <- e + dt*(Force*abs(cos(theta)) - lambda1*(e-1)) + k1*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n e <- pmax(e, 1)\n theta <- theta - dt*lambda2*S*(e-1)*(theta-thetabar) + k2*sqrt(dt)*rnorm(nt, mean=0, sd=1)\n theta <- convertThetaIter(theta, thetabar)\n }\n \n theta1 <- atan(abs(sin(theta))/pmax(0.000001, abs(cos(theta))))\n ecut <- e < ec\n theta2 <- theta1 * (1 - ecut) + ecut * runif(nt, 0, pi/2)\n A1 <- mean(cos(theta1))\n A2 <- mean(cos(theta1)) - mean(sin(theta1))\n B1 <- mean(cos(theta2))\n B2 <- mean(cos(theta2)) - mean(sin(theta2))\n\n output <- NULL\n output$values <- c(A1, A2, B1, B2, S, thetabar)\n output$e <- e\n output$theta <- theta\n \n baseModels <- output\n return(baseModels)\n}\n\nlibrary(foreach)\nlibrary(doParallel)\n\nlambda2 <- 0.25\nk1 <- 0.1\nk2 <- 0.15\nalpha <- 2\nec <- 1.6\nA <- 0.8\n\nAlist <- seq(0.6, 1, 0.05)\nlambda2list <- seq(0.1, 3, 0.2)\neclist <- seq(1.1, 2, 0.1)\nalphalist <- c(0.1, 0.2, 0.3, 0.5, 1, 1.5, 2, 5, 10, 20, 40, 60, 100)\nk1list <- seq(0.1, 2, 0.2)\nk2list <- seq(0.1, 2, 0.2)\n\nsingleRun <- function(Nsample)\n{\n resAll <- NULL\n for ( i in seq(Nsample))\n {\n cat(i, '\\n')\n res <- simu(lambda2, k1, k2, alpha, ec, A, i)\n resAll$model1 <- c(resAll$model1, res$model1)\n resAll$model2 <- c(resAll$model2, res$model2)\n }\n singleRun <- c(sd(resAll$model1), sd(resAll$model2))\n return(singleRun)\n}\n\nparaRun <- function(Nsample)\n{\n paraRun <- foreach ( i=seq(Nsample), .export=c(\"convertTheta\", \"simu\", \"lambda2\", \"k1\", \"k2\", \"alpha\", \"ec\", \"A\"), .combine = list, .multicombine = TRUE ) %dopar%\n simu(lambda2, k1, k2, alpha, ec, A, i)\n return(paraRun)\n}\n\ncomputeSDpara <- function(nthreads, Nsample)\n{\n zz <- file(\"sd.txt\", \"w\")\n close(zz)\n \n cl<-makeCluster(nthreads)\n registerDoParallel(cl)\n ##for ( k1 in k1list )\n {\n ##for ( k2 in k2list )\n {\n res <- paraRun(Nsample)\n sd1 <- sd(unlist(lapply(res, \"[[\", 1)))\n sd2 <- sd(unlist(lapply(res, \"[[\", 2)))\n cat(lambda2, k1, k2, alpha, ec, A, sd1, sd2, '\\n',\n file=\"sd.txt\", sep=',', append=TRUE)\n }\n }\n stopCluster(cl)\n}\n\nfigure4 <- function(Nsample)\n{\n mycol = c(\"#99CCFF\",\"darkgrey\",\"#FFCCCC\",\"#FF99FF\",\"#99FFCC\")\n ##mycol = c(\"#6666FF\",\"darkgrey\",\"#FFCC99\",\"#FF99FF\",\"#99FFCC\")\n mylty = c(1, 1, 1, 1, 1)\n mypch = c(0,1,5,16,2)\n \n pdf(\"figure4A.pdf\", height=7, width=9)\n par(mar=c(4,6,4,5))\n plot(0, 0, type='n', \n col = 'blue', ylab=\"L(t)-l(t)\",\n ylim=c(-.75,.75),\n xlim=c(0, 120),\n xlab=\"t\", cex.lab=2, cex.axis=1.5\n )\n for ( i in seq(Nsample))\n {\n cat(i, '\\n')\n res <- simu(lambda2, k1, k2, alpha, ec, A, i)\n lines(res$model1, pch=mypch[i], lty=mylty[i], col=mycol[i], type='o', lwd=1.5, cex=0.6)\n }\n dev.off()\n\n pdf(\"figure4B.pdf\", height=7, width=9)\n par(mar=c(4,6,4,5))\n plot(0, 0, type='n', \n col = 'blue', ylab=\"L(t)-l(t)\",\n ylim=c(-.75,.75),\n xlim=c(0, 120),\n xlab=\"t\", cex.lab=2, cex.axis=1.5\n )\n for ( i in seq(Nsample))\n {\n cat(i, '\\n')\n res <- simu(lambda2, k1, k2, alpha, ec, A, i)\n lines(res$model2, pch=mypch[i], lty=mylty[i], col=mycol[i], type='o', lwd=1.5, cex=0.6)\n }\n dev.off()\n}\n\nfigure3 <- function(nthreads, Nsample)\n{\n Force <- 0.5\n ForceList <- seq(0, 1.5, 0.05)\n Astat <- NULL\n Bstat <- NULL\n Amean <- NULL\n Asd <- NULL\n Bmean <- NULL\n Bsd <- NULL\n cl<-makeCluster(nthreads)\n registerDoParallel(cl)\n for (Force in ForceList )\n {\n cat(Force, '\\n')\n res <- foreach ( i=seq(Nsample),\n .export=c(\"convertThetaIter\", \"baseModels\", \"lambda2\", \"k1\", \"k2\", \"ec\"),\n .combine = list, .multicombine = TRUE ) %dopar%\n baseModels(lambda2, k1, k2, ec, i, Force)\n Astat <- unlist(lapply(lapply(res, \"[[\", 1), \"[[\", 2))\n Bstat <- unlist(lapply(lapply(res, \"[[\", 1), \"[[\", 4))\n ##sd2 <- sd(unlist(lapply(res, \"[[\", 2)))\n ##res <- baseModels(lambda2, k1, k2, ec, i, Force)\n ##Astat <- c(Astat, res$values[2])\n ##Bstat <- c(Bstat, res$values[4])\n ##cat(res$values, '\\n')\n Amean <- c(Amean, mean(Astat))\n Asd <- c(Asd, sd(Astat))\n Bmean <- c(Bmean, mean(Bstat))\n Bsd <- c(Bsd, sd(Bstat))\n }\n stopCluster(cl)\n pdf(\"figure3.pdf\", height=7, width=8)\n par(mar=c(4,6,4,4))\n plot(ForceList, Amean, lwd=3,\n xlab=\"F\", ylab=\"A\",\n lty=2,\n cex.lab=1.5, cex.axis=1.5,\n cex=0.5,\n type='o', ylim=c(-0.1,1))\n ##abline(v=0, lty=4, lwd=2)\n legend(\"bottomright\", c(\"one population\", \"two populations\"),\n lty=c(2, 1), lwd=2, cex=1.5)\n arrows(ForceList, Amean-3*Asd, ForceList, Amean+3*Asd, lty=1, code=3, angle=90, length=0.05)\n lines(ForceList, Bmean, lwd=3,\n type='o', lty=1, cex=0.5)\n arrows(ForceList, Bmean-3*Bsd, ForceList, Bmean+3*Bsd, lty=1, code=3, angle=90, length=0.05)\n dev.off()\n}\n\nfigure3(nthreads=20, Nsample=100)\nfigure4(5)\n\n##computeSDpara(2, 2)\n\n##system(\"cat run.r > sd.txt\")\n\n##z = baseModels(lambda2, k1, k2, ec, i, -0.1)\n##cat(z$values, '\\n')\n", "meta": {"hexsha": "4b86d8c22fbaad417588c72686e10d6163e7de2a", "size": 13466, "ext": "r", "lang": "R", "max_stars_repo_path": "MeanFieldModel/selfOrdering.r", "max_stars_repo_name": "hydrays/CellModel", "max_stars_repo_head_hexsha": "c5aef3494869d7438a3b3d8615afb2271f2202d6", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "MeanFieldModel/selfOrdering.r", "max_issues_repo_name": "hydrays/CellModel", "max_issues_repo_head_hexsha": "c5aef3494869d7438a3b3d8615afb2271f2202d6", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "MeanFieldModel/selfOrdering.r", "max_forks_repo_name": "hydrays/CellModel", "max_forks_repo_head_hexsha": "c5aef3494869d7438a3b3d8615afb2271f2202d6", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 31.1712962963, "max_line_length": 166, "alphanum_fraction": 0.4871528293, "num_tokens": 4827, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8152325073083131, "lm_q2_score": 0.6370307806984444, "lm_q1q2_score": 0.519328200581365}} {"text": "model_canopytemperature <- function (minTair = 0.7,\n maxTair = 7.2,\n cropHeatFlux = 447.912,\n conductance = 598.685,\n lambdaV = 2.454,\n rhoDensityAir = 1.225,\n specificHeatCapacityAir = 0.00101){\n #'- Name: CanopyTemperature -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: CanopyTemperature Model\n #' * Author: Pierre Martre\n #' * Reference: Modelling energy balance in the wheat crop model SiriusQuality2:\n #' Evapotranspiration and canopy and soil temperature calculations\n #' * Institution: INRA/LEPSE Montpellier\n #' * Abstract: It is calculated from the crop heat flux and the boundary layer conductance \n #'- inputs:\n #' * name: minTair\n #' ** description : minimum air temperature\n #' ** datatype : DOUBLE\n #' ** variablecategory : auxiliary\n #' ** min : -30\n #' ** max : 45\n #' ** default : 0.7\n #' ** unit : degC\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' * name: maxTair\n #' ** description : maximum air Temperature\n #' ** datatype : DOUBLE\n #' ** variablecategory : auxiliary\n #' ** min : -30\n #' ** max : 45\n #' ** default : 7.2\n #' ** unit : degC\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' * name: cropHeatFlux\n #' ** description : Crop heat flux\n #' ** variablecategory : rate\n #' ** inputtype : variable\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 10000\n #' ** default : 447.912\n #' ** unit : g/m**2/d\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' * name: conductance\n #' ** description : the boundary layer conductance\n #' ** variablecategory : state\n #' ** inputtype : variable\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 10000\n #' ** default : 598.685\n #' ** unit : m/d\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' * name: lambdaV\n #' ** description : latent heat of vaporization of water\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 2.454\n #' ** min : 0\n #' ** max : 10\n #' ** unit : MJ/kg\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: rhoDensityAir\n #' ** description : Density of air\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 1.225\n #' ** unit : kg/m**3\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #' * name: specificHeatCapacityAir\n #' ** description : Specific heat capacity of dry air\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.00101\n #' ** unit : MJ/kg/degC\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : parameter\n #'- outputs:\n #' * name: minCanopyTemperature\n #' ** description : minimal Canopy Temperature \n #' ** datatype : DOUBLE\n #' ** variablecategory : state\n #' ** min : -30\n #' ** max : 45\n #' ** unit : degC\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' * name: maxCanopyTemperature\n #' ** description : maximal Canopy Temperature \n #' ** datatype : DOUBLE\n #' ** variablecategory : state\n #' ** min : -30\n #' ** max : 45\n #' ** unit : degC\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n minCanopyTemperature <- minTair + (cropHeatFlux / (rhoDensityAir * specificHeatCapacityAir * conductance / lambdaV * 1000.0))\n maxCanopyTemperature <- maxTair + (cropHeatFlux / (rhoDensityAir * specificHeatCapacityAir * conductance / lambdaV * 1000.0))\n return (list (\"minCanopyTemperature\" = minCanopyTemperature,\"maxCanopyTemperature\" = maxCanopyTemperature))\n}", "meta": {"hexsha": "1d0b06475a67c2f3d882897907610d56a833f1b7", "size": 5929, "ext": "r", "lang": "R", "max_stars_repo_path": "test/Models/energybalance_pkg/src/r/Canopytemperature.r", "max_stars_repo_name": "brichet/PyCrop2ML", "max_stars_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-06-21T18:58:04.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-29T21:32:28.000Z", "max_issues_repo_path": "test/Models/energybalance_pkg/src/r/Canopytemperature.r", "max_issues_repo_name": "brichet/PyCrop2ML", "max_issues_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 27, "max_issues_repo_issues_event_min_datetime": "2018-12-04T15:35:44.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-11T08:25:03.000Z", "max_forks_repo_path": "test/Models/energybalance_pkg/src/r/Canopytemperature.r", "max_forks_repo_name": "brichet/PyCrop2ML", "max_forks_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 7, "max_forks_repo_forks_event_min_datetime": "2019-04-20T02:25:22.000Z", "max_forks_repo_forks_event_max_datetime": "2021-11-04T07:52:35.000Z", "avg_line_length": 57.5631067961, "max_line_length": 129, "alphanum_fraction": 0.3924776522, "num_tokens": 1247, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8705972717658209, "lm_q2_score": 0.5964331462646255, "lm_q1q2_score": 0.5192530699286878}} {"text": "# Copy of the levinson() function from the signals package with\n# a correction to remove deprecation warnings.\n#\nlevnsn <- function (x, p = NULL) \n{\n fit <- function(acf, p) {\n ref <- numeric(p)\n g <- -acf[2]/acf[1]\n a <- g\n v <- Re((1 - g * Conj(g)) * acf[1])\n ref[1] <- g\n for (t in 2:p) {\n g <- -(acf[t + 1] + a %*% acf[seq(t, 2, by = -1)])/v\n a <- c((a + as.vector(g) * Conj(a[seq(t - 1, 1, -1)])), g)\n v <- v * (1 - Re(g * Conj(g)))\n ref[t] <- g\n }\n a <- c(1, a)\n return(list(a = a, v = v, ref = ref))\n }\n if ((!is.null(p) && (p != as.integer(p))) || (p < 2)) \n stop(\"p must be integer >= 2.\")\n if (is.numeric(x) && is.null(dim(x))) {\n lx <- length(x)\n if (is.null(p) || p >= lx) \n p <- lx - 1\n r <- fit(x, p)\n }\n else {\n if (is.numeric(x) && !(is.null(dim(x)))) {\n lx <- dim(x)\n if (is.null(p) || p >= lx[1]) \n p <- lx[1] - 1\n zr <- apply(x, 2, function(y) fit(y, p))\n zr <- matrix(unlist(zr), nrow = lx[2], byrow = TRUE)\n a <- zr[, 1:(p + 1), drop = FALSE]\n v <- zr[, p + 2]\n ref <- t(zr[, -(1:(p + 2)), drop = FALSE])\n r <- list(a = a, v = v, ref = ref)\n }\n else {\n stop(\"x must be a numeric vector or matrix.\")\n }\n }\n return(r)\n}\n", "meta": {"hexsha": "a2ac8c57c19d1bc781276f889aa4705d149a17ae", "size": 1275, "ext": "r", "lang": "R", "max_stars_repo_path": "R/levnsn.r", "max_stars_repo_name": "NemoursResearch/FormantTracking", "max_stars_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2021-09-01T14:22:23.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-25T03:47:04.000Z", "max_issues_repo_path": "R/levnsn.r", "max_issues_repo_name": "NemoursResearch/FormantTracking", "max_issues_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/levnsn.r", "max_forks_repo_name": "NemoursResearch/FormantTracking", "max_forks_repo_head_hexsha": "7053e6add8672dc00aa70f6549d90ff935f96748", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-08-31T18:20:28.000Z", "max_forks_repo_forks_event_max_datetime": "2021-08-31T18:20:28.000Z", "avg_line_length": 27.1276595745, "max_line_length": 64, "alphanum_fraction": 0.4329411765, "num_tokens": 486, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8244619263765707, "lm_q2_score": 0.6297746074044134, "lm_q1q2_score": 0.5192251860036912}} {"text": "###############################################################################\n# package: package source\n# By puglisij Copyright (C) 2015, All rights reserved.\n#\n###############################################################################\n\n\n#' linked_list\n#'\n#' @return function list\nlinked_list <- function() {\n head <- list(0)\n length <- 0\n \n methods <- list()\n \n methods$add <- function(val) {\n length <<- length + 1\n head <<- list(head, val)\n }\n \n methods$as.list <- function() {\n b <- vector('list', length)\n h <- head\n for (i in length:1) {\n b[[i]] <- head[[2]]\n head <- head[[1]]\n }\n return(b)\n }\n methods\n}\n\n#' @title Golden Ratio\n#' @description Determine a local minimum of a continuous real function f\n#'\n#' @param f function\n#' @param a numeric\n#' @param b numeric > a\n#' @param eps numeric (default 1e-16)\n#' @param maxiter maximum number of iterations\n#' @return list containing approximation of local minimum,\n#' value at local minimum and some control data\n#' @export\ngolden_ratio <- function(f,\n a,\n b,\n eps = 1e-16,\n maxiter = 100) {\n stopifnot(\n is.numeric(a) & is.finite(a),\n is.numeric(b) & is.finite(b),\n b > a,\n is.numeric(eps) & is.finite(eps) & eps > 0,\n is.numeric(maxiter) & is.finite(maxiter) & maxiter > 0\n )\n \n phi <- (sqrt(5) - 1) / 2\n convergence <- 1\n \n for (i in 1:maxiter) {\n # Update xl, xp\n xl <- b - phi * (b - a)\n xp <- a + phi * (b - a)\n if (f(xl) > f(xp)) {\n a <- xl\n } else {\n b <- xp\n }\n \n # Check if method converged\n if (abs(b - a) < eps) {\n convergence <- 0\n break\n }\n }\n \n if (convergence == 1)\n warning(\"Method didn't converge\")\n \n list(\n par = (a + b) / 2,\n value = f((a + b) / 2),\n counts = i,\n convergence = convergence,\n message = NULL\n )\n}\n\n\n#' @title Elliot wave - puglisij\n#' @description Criteria must match as follows: Climb in instrument begins as wave 1, followed by a retracement\n#' which is wave 2, followed by another climb wave 3, followed by a retracement wave 4, and\n#' then followed by the final climb, wave 5.\n#'\n#' @param v vector of instrument price data\n#' @param eps numeric\n#' @param maxiter maximum number of iterations\n#' @param fib vector of fibonacci ratios\n#' @param time_restrictive boolean each wave is restricted to its individual time interval\n#' @return list containing entries and golden ratio computed position scaled\n#' @author puglisij\n#' @export\n#'\nelliot_wave <- function(v,\n eps = 1e-16,\n maxiter = 10000,\n fib = c(0.236, 0.382, 0.5, 0.618),\n time_restrictive = TRUE) {\n stopifnot(is.vector(v) && is.vector(fiblevel))\n \n x <- vector(numeric, length = length(v))\n j <- as.vector(v)\n z <- vector(mode = \"complex\", length = 0)\n \n peaks_valleys <- cbind(x, z)\n entrys <- linked_list()\n \n for (i in 2:length(v) - 1) {\n if (v(i) <= v(i + 1) &&\n v(i - 1) >= v(i) || v(i) >= v(i + 1) && v(i - 1) <= v(i)) {\n x = c(x, v(i))\n z = c(z, j(i))\n \n x = x[1, diff(x) != 0]\n z = z[1, diff(x) != 0]\n peaks_valleys = c(x, z)\n }\n }\n \n entry <-\n function(fib_ratio,\n instrument_vector,\n index)\n ((0.999 * instrument_vector(index + 3)) - ((\n instrument_vector(index - 1) - ((\n instrument_vector(index - 1) - instrument_vector(index - 2)\n ) * fib_ratio)\n ) * 1.001)) / ((instrument_vector(index - 1) - ((instrument_vector(index -\n 1) - instrument_vector(index - 2)) * fib_ratio\n )) * 1.001)\n \n for (n in 3:length(x) - 1) {\n if (x(n) < x(n - 1) &&\n x(n) >= x(n - 2) &&\n x(n + 1) > x(n - 1) && x(n + 2) > x(n - 1)) {\n if (x(n + 1) - x(n) >= x(n - 1) - x(n - 2) || x(n + 1) - x(n) >= x\n (n + 3) - x(n + 2)) {\n if (time_restrictive == TRUE) {\n if (z(n) - z(n - 1) <= (0.382 * (z(n - 1) - z(n - 2))) &&\n z(n + 1) - z(n) <= (1.618 * (z(n - 1) - z(n - 2))) &&\n z(n + 2) - z(n + 1) <= (0.382 * (z(n + 1) - z(n))) &&\n z(n + 3) - z(n + 2) <= (1.618 * (z(n + 1) - z(n)))) {\n \n } else {\n next\n }\n }\n for (fr in fib) {\n if (x(n) <= (x(n - 1) - ((x(n - 1) - x(n - 2)) * fr)) * 1.001 &&\n x(n) >= (x(n - 1) - ((x(n - 1) - x(n - 2)) * fr)) * 0.999) {\n e <- entry(\n fib_ratio = fr,\n instrument_vector = x,\n index = n\n )\n \n gr <-\n golden_ratio(\n function(x)\n x * cos(0.1 * exp(x)) * sin(0.1 * exp(x)),\n a = e,\n b = x(n - 1),\n maxiter = maxiter\n )\n \n entrys$add(e, gr, n, x(n))\n }\n }\n \n }\n }\n }\n \n entrys\n}\n", "meta": {"hexsha": "b22143008710d6c202f57d856f9ef849cb833cc2", "size": 5097, "ext": "r", "lang": "R", "max_stars_repo_path": "R/elliot.r", "max_stars_repo_name": "gitdek/elliotwave", "max_stars_repo_head_hexsha": "a3fc19b64afa01aa75434cf6f7cb0375f3552da9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2018-03-25T00:31:46.000Z", "max_stars_repo_stars_event_max_datetime": "2020-12-03T13:00:06.000Z", "max_issues_repo_path": "R/elliot.r", "max_issues_repo_name": "gitdek/elliotwave", "max_issues_repo_head_hexsha": "a3fc19b64afa01aa75434cf6f7cb0375f3552da9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/elliot.r", "max_forks_repo_name": "gitdek/elliotwave", "max_forks_repo_head_hexsha": "a3fc19b64afa01aa75434cf6f7cb0375f3552da9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 3, "max_forks_repo_forks_event_min_datetime": "2018-03-25T00:31:50.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-03T22:49:23.000Z", "avg_line_length": 27.256684492, "max_line_length": 121, "alphanum_fraction": 0.4435942711, "num_tokens": 1548, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8104789269812079, "lm_q2_score": 0.6406358479787609, "lm_q1q2_score": 0.5192218546555224}} {"text": "#' mb2mmHG\n#'\n#' Conversion from Millibar [hPa] to mmHG.\n#'\n#' @param numeric mb Air pressure in Millibar [hPa] \n#' @return \n#'\n#'\n#' @author Istituto di Biometeorologia Firenze Italy Alfonso Crisci \\email{a.crisci@@ibimet.cnr.it}\n#' @keywords mbtommHG \n#' \n#' @export\n#'\n#'\n#'\n#'\nmb2mmHg<-function(mb)\n{\n return (mb* 0.750062);\n}\n\n", "meta": {"hexsha": "fa8418ff44edb7f893c10bcff263de19a71ad9d4", "size": 336, "ext": "r", "lang": "R", "max_stars_repo_path": "R/mb2mmHg.r", "max_stars_repo_name": "alfcrisci/biometeoR", "max_stars_repo_head_hexsha": "e9ee73da6ecc515ecd471cb9ae059a4370a51493", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2019-06-13T15:54:46.000Z", "max_stars_repo_stars_event_max_datetime": "2019-06-13T15:54:46.000Z", "max_issues_repo_path": "R/mb2mmHg.r", "max_issues_repo_name": "alfcrisci/biometeoR", "max_issues_repo_head_hexsha": "e9ee73da6ecc515ecd471cb9ae059a4370a51493", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/mb2mmHg.r", "max_forks_repo_name": "alfcrisci/biometeoR", "max_forks_repo_head_hexsha": "e9ee73da6ecc515ecd471cb9ae059a4370a51493", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 15.2727272727, "max_line_length": 101, "alphanum_fraction": 0.6398809524, "num_tokens": 127, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.8006920020959544, "lm_q2_score": 0.6477982179521103, "lm_q1q2_score": 0.5186868520862666}} {"text": "library(tensorflow)\n\nsess = tf$Session()\n\nhello <- tf$constant('Hello, TensorFlow!')\nsess$run(hello)\n\na <- tf$constant(10L)\nb <- tf$constant(32L)\nsess$run(a + b)\n\n#OR gate input data\ntrainin = rbind(c(1,1), c(1,-1), c(-1,1), c(-1,-1));\n#OR gate output data\ntrainout = rbind(1, 1, 1, 0);\n#Combined OR gate data\nORdat=cbind(trainout,trainin)\n\n# fit neural network with no hidden layers\nset.seed(2)\nNN = neuralnet(ORdat[,1]~., ORdat[,-1], hidden = 0 , threshold = 0.001,\nstepmax = 1e+05, linear.output = FALSE)\n#visualise the NN\nplot(NN)\n\nNN$weights\n\ntestin= rbind(c(1,1))\npredict_testNN = compute(NN, testin)\n\npredict_out = as.numeric(predict_testNN$net.result>0.5)\nprint(predict_out)\n\n\n#set up the input sequence\ntestin=rbind(c(1,1),c(1,-1),c(-1,1), c(-1,-1))\npredict_testNN = compute(NN, testin)\npredict_testNN$neurons\npredict_testNN$net.result\npredict_out = as.numeric(predict_testNN$net.result>0.5)\npredict_out\n\n#XOR gate input data\ntrainin = rbind(c(1,1), c(1,-1), c(-1,1), c(-1,-1));\n#XOR gate output data\ntrainout = rbind(0, 1, 1, 0);\n#Combined XOR gate data\nXORdat=cbind(trainout,trainin)\n#train a neural network on the XOR data\nset.seed(2)\nNN = neuralnet(XORdat[,1]~., XORdat[,-1], hidden = c(3,3) , threshold =\n0.001, stepmax = 1e+05, linear.output = FALSE)\n\ntestin = rbind(c(1,1), c(1,-1), c(-1,1), c(-1,-1));\ntestout=rbind(0,1,1,0)\npredict_testNN = compute(NN, testin)\npredict_testNN$neurons\npredict_testNN$net.result\npredict_out = as.numeric(predict_testNN$net.result>0.5)\npredict_out\n", "meta": {"hexsha": "6a028110f87e09b45fb0d18684250d9d84b68f32", "size": 1496, "ext": "r", "lang": "R", "max_stars_repo_path": "main.r", "max_stars_repo_name": "IKMalik/AI-R", "max_stars_repo_head_hexsha": "91e0eee13d0e50f9cb27e70dae4bea23d7197941", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "main.r", "max_issues_repo_name": "IKMalik/AI-R", "max_issues_repo_head_hexsha": "91e0eee13d0e50f9cb27e70dae4bea23d7197941", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "main.r", "max_forks_repo_name": "IKMalik/AI-R", "max_forks_repo_head_hexsha": "91e0eee13d0e50f9cb27e70dae4bea23d7197941", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 24.5245901639, "max_line_length": 71, "alphanum_fraction": 0.6978609626, "num_tokens": 548, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7279754371026367, "lm_q2_score": 0.7122321720225278, "lm_q1q2_score": 0.51848752674666}} {"text": "# Tempering algorithm.\n\nif (TRUE) {\n .libPaths(\"~/Rlib\")\n setwd(\"~/Code/ising_model/scripts\")\n source(\"algorithms.r\")\n source(\"tools.r\")\n}\n\ntempering <- function(A, beta_vector, init_matrix,\n n_exchange, n_mc_iter, B = 0,\n f1, f2 = f1, \n which_sampler = rep(1, length(beta_vector)),\n n_states = 2,\n is_potts = FALSE) {\n # Returns samples for graph A, across all temperatures\n # specified by the beta_vector. Uses tempering (or exchange)\n # Monte Carlo to improve sampling at high betas, i.e. cold temperatures.\n #\n # A: graph\n # beta_vector: betas for each replica\n # init_matrix: inits for each replica\n # n_exchange: number of exchanges.\n # n_mc_iter: number of sampling iterations between exchanges.\n # B: exterior magnetic field\n # f1: function for first sampler. Should accept A, beta, B,\n # an init vector, and n_mc_iter as arguments.\n # f2: function for second sampler. Same as above.\n # which_sampler: which sampler to use for each replica. This\n # should be a vector with the same length as\n # beta_vector with entries 1 and 2.\n #\n n_replica <- length(beta_vector)\n n_iter <- n_exchange * n_mc_iter\n n_particles <- nrow(A)\n\n X <- array(NA, dim = c(n_replica, n_iter, n_particles))\n\n current_init <- init_matrix\n\n exchange_rate <- rep(0, n_replica - 1)\n \n for (i in 1:n_exchange) {\n first_index <- (i - 1) * n_mc_iter + 1\n last_index <- i * n_mc_iter\n for (b in 1:n_replica) {\n if (which_sampler[b] == 1) {\n X[b, first_index:last_index, ] <-\n f1(A, beta_vector[b], B, current_init[b, ],\n n_mc_iter)\n } else {\n X[b, first_index:last_index, ] <-\n f2(A, beta_vector[b], B, current_init[b, ],\n n_mc_iter)\n }\n }\n\n # exchange step\n # NOTE: pass 1 instead of beta to ising kernel to not double-count beta.\n for (b in 1:(n_replica - 1)) {\n if (b == 1) {\n if (!is_potts) {\n E_low = - ising_kernel(1, A, X[b, last_index, ], B, T)\n } else {\n E_low = - 0.5 * potts_log_kernel(1, A, X[b, last_index, ], \n n_states = n_states)\n }\n } else { \n E_low = E_high\n }\n\n if (!is_potts) {\n E_high = - ising_kernel(1, A, X[b + 1, last_index, ], B, T)\n } else {\n E_high = - 0.5 * potts_log_kernel(1, A, X[b + 1, last_index, ],\n n_states = n_states)\n }\n\n delta = (beta_vector[b + 1] - beta_vector[b]) * (E_high - E_low)\n\n if (runif(1) < exp(delta)) {\n saved_state = X[b + 1, last_index, ]\n X[b + 1, last_index, ] = X[b, last_index, ]\n X[b, last_index, ] = saved_state\n\n E_high = E_low # update energy before next exchange\n\n exchange_rate[b] <- exchange_rate[b] + 1\n }\n\n current_init[b, ] = X[b, last_index, ] }\n\n current_init[n_replica, ] = X[n_replica, last_index, ]\n }\n print(exchange_rate / n_exchange)\n\n X\n}\n\n###############################################################################\n## Unit test\nunit_test <- FALSE\n# WARNING: unit tests have not been updated with latest signature\n# from tempering (with which_sampler argument)\nif (unit_test) {\n n_part <- 64\n A <- adjacency_graph(n_part, type = \"gaussian\")\n beta_vector <- c(0.5, 1.8, 2.3, 2.6, 3.0, 3.25, 3.5, 3.70, 3.90, 4.0) \n # c(seq(from = 0.5, to = 4, by = 0.35))\n # c(1.0, 1.0, 1.5, 2.0, 2.5) # c(0.5, 0.75, 1.0, 1.25)\n B <- 0\n \n n_replica <- length(beta_vector)\n init_matrix <- matrix(sample(c(-1, 1), n_part * n_replica, replace = T),\n nrow = n_replica, ncol = n_part)\n \n n_mc_iter <- 5e3\n n_exchange <- 40\n \n f1 <- sumit_sampler\n f2 <- function(A, beta, B, init, n_iter) { \n hb_sampler(A, beta, B, init, n_iter, \n sub_sample = ceiling(nrow(A) / 10))\n }\n threshold = 4.1\n\n sample <- tempering(A, beta_vector, init_matrix,\n n_exchange, n_mc_iter, B = 0,\n f1, f2, threshold)\n\n temp_estimates <- monte_carlo_estimate(sample[, -(1:1000), ])\n\n threshold = 0\n sample_hb <- tempering(A, beta_vector, init_matrix,\n n_exchange, n_mc_iter, B, f1, f2, threshold)\n \n temp_hb_estimates <- monte_carlo_estimate(sample_hb[, -(1:1000), ])\n \n ## Benchmark against sumit's sampler without tempering\n n_iter <- n_mc_iter * n_exchange\n sample_const <- array(NA, c(n_replica, n_iter, n_part))\n for (i in 1:n_replica) {\n sample_const[i, , ] <- sumit_sampler(A, beta_vector[i], n_iter = n_iter,\n init = init_matrix[i, ])\n }\n\n const_estimates <- \n monte_carlo_estimate(sample_const[, -(1:1000), ])\n\n temp_estimates\n temp_hb_estimates\n \n const_estimates\n}\n", "meta": {"hexsha": "37ee71faaeb48823e9ae3f7b8ef6b43ff752bf11", "size": 4869, "ext": "r", "lang": "R", "max_stars_repo_path": "scripts/algorithm_tempering.r", "max_stars_repo_name": "charlesm93/potts_simulation", "max_stars_repo_head_hexsha": "5f0247d58b5d83b777679b10d2d6e424fee3606b", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2021-10-21T16:25:09.000Z", "max_stars_repo_stars_event_max_datetime": "2021-10-21T16:25:09.000Z", "max_issues_repo_path": "scripts/algorithm_tempering.r", "max_issues_repo_name": "charlesm93/potts_simulation", "max_issues_repo_head_hexsha": "5f0247d58b5d83b777679b10d2d6e424fee3606b", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "scripts/algorithm_tempering.r", "max_forks_repo_name": "charlesm93/potts_simulation", "max_forks_repo_head_hexsha": "5f0247d58b5d83b777679b10d2d6e424fee3606b", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 31.4129032258, "max_line_length": 79, "alphanum_fraction": 0.5625385089, "num_tokens": 1451, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7520125737597972, "lm_q2_score": 0.6893056104028799, "lm_q1q2_score": 0.5183664861861378}} {"text": "\n# 加载库\nlibrary(WGCNA);\n\n# 读取文件\n# The following setting is important, do not omit.\n# 如果没有显式地指定“stringsAsFactors=FALSE”,默认会将所有的字符串转换为因子,导致数据处理速度较慢\noptions(stringsAsFactors = FALSE)\n# Read in the female liver data set 读取135个雌性小鼠的数据\nfemData = read.csv(\"./data/LiverFemale3600.csv\")\n# Take a quick look at what is in the data set:\n# 查看数据的维度\ndim(femData)\n# 预览数据\nhead(femData)\n\n# 删除冗余数据-c(1:8)删除前8列数据，t()转置数据\ndatExpr0 = as.data.frame(t(femData[, -c(1:8)]));\nhead(datExpr0)\n\n# 将原数据的行列名复制过来\nnames(datExpr0) = femData$substanceBXH;\nrownames(datExpr0) = names(femData)[-c(1:8)];\nhead(datExpr0)\n\ngsg = goodSamplesGenes(datExpr0, verbose = 3)\ngsg$allOK;\n\nif (!gsg$allOK)\n{\n # Optionally, print the gene and sample names that were removed:\n # 打印删除的基因和样本名称\n if (sum(!gsg$goodGenes)>0)\n printFlush(paste(\"Removing genes:\", paste(names(datExpr0)[!gsg$goodGenes], collapse = \", \")));\n if (sum(!gsg$goodSamples)>0)\n printFlush(paste(\"Removing samples:\", paste(rownames(datExpr0)[!gsg$goodSamples], collapse = \", \")));\n # Remove the offending genes and samples from the data:\n # 从数据中删除有问题的基因和样本\n datExpr0 = datExpr0[gsg$goodSamples, gsg$goodGenes]\n}\n\n# hclusts聚类算法, dist计算基因之间的距离\nsampleTree = hclust(dist(datExpr0), method = \"average\");\n# Plot the sample tree: Open a graphic output window of size 12 by 9 inches\n# The user should change the dimensions if the window is too large or too small.\n# 绘制聚类树，sizeGrWindow设置绘图窗口大小\n# sizeGrWindow(16,9)\npdf(file = \"./plot/sampleClustering.pdf\", width = 12, height = 9);\n# 设置文字大小\npar(cex = 0.5);\n# 设置图像边距c(bottom, left, top, right) \n# par(mar = c(0,4,2,0))\n# 画图 main标题，sub子标题，xlab x轴标题，cex.lab标题字体大小，cex.axis坐标轴刻度大小，cex.main主标题字体\nplot(sampleTree, main = \"Sample clustering to detect outliers\", sub=\"\", xlab=\"\", cex.lab = 1.5, cex.axis = 1.5, cex.main = 2)\ndev.off()\n\n# Plot a line to show the cut\n# 设置文字大小\npar(cex = 0.5);\nplot(sampleTree, main = \"Sample clustering to detect outliers\", sub=\"\", xlab=\"\", cex.lab = 1.5, cex.axis = 1.5, cex.main = 2)\n# 在上图上画红线\nabline(h = 15, col = \"red\");\n# Determine cluster under the line\n# 剪枝算法，cutHeight 修剪树枝的高度 minSize集群最小数\nclust = cutreeStatic(sampleTree, cutHeight = 15, minSize = 10)\n# 剪枝结果\ntable(clust)\n# clust 1 contains the samples we want to keep\nkeepSamples = (clust==1)\n# 符合要求的数据\ndatExpr = datExpr0[keepSamples, ]\n# 提取行\nnSamples = nrow(datExpr)\n# 提取列\nnGenes = ncol(datExpr)\n\ntraitData = read.csv(\"./data/ClinicalTraits.csv\");\ndim(traitData)\n#names(traitData)\n# remove columns that hold information we do not need.\n# 删除不需要的列\nallTraits = traitData[, -c(31, 16)];\nallTraits = allTraits[, c(2, 11:36) ];\ndim(allTraits)\nhead(allTraits)\n# names(allTraits)\n\n# 形成一个类似于表达数据的数据框架，以保存临床特征\n# 提取行名\nfemaleSamples = rownames(datExpr)\n# 数据匹配 返回匹配行\ntraitRows = match(femaleSamples, allTraits$Mice);\n# 提取指定要求行\ndatTraits = allTraits[traitRows, -1];\n# 提取行名\nrownames(datTraits) = allTraits[traitRows, 1];\n# 垃圾回收\ncollectGarbage();\n\n# Re-cluster samples\n# 画聚类图\nsampleTree2 = hclust(dist(datExpr), method = \"average\")\n# Convert traits to a color representation: white means low, red means high, grey means missing entry\n# 画表型的热图\n# 将特征转换为颜色表示：白色表示低，红色表示高，灰色表示缺少条目\n# 如果signed为true 以绿色开头代表最大负值，以白色开头代表零附近的值，然后变为红色代表正值\ntraitColors = numbers2colors(datTraits, signed =FALSE);\n# Plot the sample dendrogram and the colors underneath.\n# 绘制出树状图和下面的颜色 \nplotDendroAndColors(sampleTree2, traitColors,groupLabels = names(datTraits),main = \"Sample dendrogram and trait heatmap\")\n\n# Choose a set of soft-thresholding powers\n# 给出候选的β值，c(1:10)表示1到10；seq(from = 12, to=20, by=2)表示从12开始间隔两个数到20\npowers = c(c(1:10), seq(from = 12, to=20, by=2))\npowers\n# Call the network topology analysis function 调用网络拓扑分析函数\n# verbose表示输出结果详细程度\nsft = pickSoftThreshold(datExpr, powerVector = powers, verbose = 0);\n\n# sft这中保存了每个powers值计算出来的网络特征,其中powerEstimate就是最佳power值，fitIndices保存了每个power对应的网络的特征。\nstr(sft)\n\n# Plot the results 结果绘图\n# 设置窗格大小\n#sizeGrWindow(9, 5)\n# 设置图的显示一行两列\n# par(mfrow = c(1,2));\ncex1 = 0.9;\n# Scale-free topology fit index as a function of the soft-thresholding power\n# 生成阈值和网络的特征之间的关系函数\nplot(sft$fitIndices[,1], -sign(sft$fitIndices[,3])*sft$fitIndices[,2],\nxlab=\"Soft Threshold (power)\",ylab=\"Scale Free Topology Model Fit,signed R^2\",type=\"n\",\nmain = paste(\"Scale independence\"))\ntext(sft$fitIndices[,1], -sign(sft$fitIndices[,3])*sft$fitIndices[,2],\nlabels=powers,cex=cex1,col=\"red\");\n# this line corresponds to using an R^2 cut-off of h\nabline(h=0.90,col=\"red\")\n\n# sft$fitIndices 保存了每个power构建的相关性网络中的连接度的统计值，k就是连接度值，每个power值提供了max, median, max3种连接度的统计量\n# 对连接度的均值进行可视化\n# Mean connectivity as a function of the soft-thresholding power\nplot(sft$fitIndices[,1], sft$fitIndices[,5],\nxlab=\"Soft Threshold (power)\",ylab=\"Mean Connectivity\", type=\"n\",\nmain = paste(\"Mean connectivity\"))\ntext(sft$fitIndices[,1], sft$fitIndices[,5], labels=powers, cex=cex1,col=\"red\")\n\n# datExpr表达数据，TOMType拓扑重叠矩阵计算方式，minModuleSize用于模块检测的最小模块尺寸,\n# reassignThreshold 是否在模块之间重新分配基因的p值比率阈值，mergeCutHeight 树状图切割高度\n# numericLabels 返回的模块应该用颜色（FALSE）还是数字（TRUE）标记,pamRespectsDendro树状图相关参数\n# saveTOMs 字符串的向量，saveTOMFileBase 包含包含共识拓扑重叠文件的文件名库的字符串\nnet = blockwiseModules(datExpr, power = sft$powerEstimate,TOMType = \"unsigned\", minModuleSize = 30,reassignThreshold = 0, \n mergeCutHeight = 0.25,numericLabels = TRUE, pamRespectsDendro = FALSE,saveTOMs = TRUE,\n saveTOMFileBase = \"femaleMouseTOM\",verbose = 3)\n\n table(net$colors)\n\n# open a graphics window\n# sizeGrWindow(12, 9)\n# Convert labels to colors for plotting\n# 将标签转化为绘图颜色\nmergedColors = labels2colors(net$colors)\n# Plot the dendrogram and the module colors underneath\n# 绘制树状图和下面的模块颜色\n# dendroLabels树状图标签。设置为FALSE完全禁用树状图标签；设置为NULL使用的行标签datExpr\n# addGuide是否应在树状图中添加垂直的“指导线”？线条使识别单个样本的颜色代码更加容易。\nplotDendroAndColors(net$dendrograms[[1]], mergedColors[net$blockGenes[[1]]],\"Module colors\",\n dendroLabels = FALSE, hang = 0.03,addGuide = TRUE, guideHang = 0.05)\n\nmoduleLabels = net$colors\nmoduleColors = labels2colors(net$colors)\nMEs = net$MEs;\ngeneTree = net$dendrograms[[1]];\nsave(MEs, moduleLabels, moduleColors, geneTree,\nfile = \"FemaleLiver-02-networkConstruction-auto.RData\")\n\n# Define numbers of genes and samples\n# 获得基因数和样本数\nnGenes = ncol(datExpr);\nnSamples = nrow(datExpr);\n\n# Recalculate MEs with color labels\n# 用彩色标签重新计算MEs\n# 在给定的单个数据集中计算模块的模块本征基因\nMEs0 = moduleEigengenes(datExpr, moduleColors)$eigengenes\n# 对给定的（特征）向量进行重新排序，以使相似的向量（通过相关性度量）彼此相邻\nMEs = orderMEs(MEs0)\n\n# 计算module的ME值与表型的相关系数\nmoduleTraitCor = cor(MEs, datTraits, use = \"p\");\nmoduleTraitPvalue = corPvalueStudent(moduleTraitCor, nSamples);\n\nnames(MEs)\n\n# sizeGrWindow(10,6)\n# 显示相关性及其p值\ntextMatrix = paste(signif(moduleTraitCor, 2), \"\\n(\",signif(moduleTraitPvalue, 1), \")\", sep = \"\");\ndim(textMatrix) = dim(moduleTraitCor)\npar(mar = c(6, 8.5, 3, 3));\n# Display the correlation values within a heatmap plot\\\n# ySymbols 当ylabels使用时所使用的其他标签； colorLabels 应该使用颜色标签吗\n# colors 颜色； textMatrix 单元格名字\nlabeledHeatmap(Matrix = moduleTraitCor,xLabels = names(datTraits),yLabels = names(MEs),ySymbols = names(MEs),\n colorLabels = FALSE,colors = greenWhiteRed(50),textMatrix = textMatrix,setStdMargins = FALSE,\n cex.text = 0.4,zlim = c(-1,1),\nmain = paste(\"Module-trait relationships\"))\n\nsizeGrWindow(10,6)\n# Will display correlations and their p-values\n\ndim(textMatrix) = dim(moduleTraitCor)\npar(mar = c(6, 8.5, 3, 3));\n# Display the correlation values within a heatmap plot\nlabeledHeatmap(Matrix = moduleTraitCor,\nxLabels = names(datTraits),\nyLabels = names(MEs),\nySymbols = names(MEs),\ncolorLabels = FALSE,\ncolors = greenWhiteRed(50),\ntextMatrix = textMatrix,\nsetStdMargins = FALSE,\ncex.text = 0.5,\nzlim = c(-1,1),\nmain = paste(\"Module-trait relationships\"))\n\n# Define variable weight containing the weight column of datTrait\n# 定义包含数据特征权重列的变量权重\nweight = as.data.frame(datTraits$weight_g);\nnames(weight) = \"weight\"\ngeneModuleMembership = as.data.frame(cor(datExpr, MEs, use = \"p\"));\n# 模块的名称(颜色) substring提取文本从第3个字母开始\nmodNames = substring(names(MEs), 3)\n# 基因和模块的相关系数\ngeneModuleMembership = as.data.frame(cor(datExpr, MEs, use = \"p\"));\nMMPvalue = as.data.frame(corPvalueStudent(as.matrix(geneModuleMembership), nSamples));\nnames(geneModuleMembership) = paste(\"MM\", modNames, sep=\"\");\nnames(MMPvalue) = paste(\"p.MM\", modNames, sep=\"\");\n\n#gene和性状的关系\ngeneTraitSignificance = as.data.frame(cor(datExpr, weight, use = \"p\"));\nGSPvalue = as.data.frame(corPvalueStudent(as.matrix(geneTraitSignificance), nSamples));\nnames(geneTraitSignificance) = paste(\"GS.\", names(weight), sep=\"\");\nnames(GSPvalue) = paste(\"p.GS.\", names(weight), sep=\"\");\n\n# 模型颜色\nmodule = \"brown\"\n# 匹配列\ncolumn = match(module, modNames);\nmoduleGenes = moduleColors==module;\n#sizeGrWindow(7, 7);\npar(mfrow = c(1,1));\n# 画散点图\nverboseScatterplot(abs(geneModuleMembership[moduleGenes, column]),\n abs(geneTraitSignificance[moduleGenes, 1]),\n xlab = paste(\"Module Membership in\", module, \"module\"),\n ylab = \"Gene significance for body weight\",\n main = paste(\"Module membership vs. gene significance\\n\"),\n cex.main = 1.2, cex.lab = 1.2, cex.axis = 1.2, col = module)\n\n# 提取表带数据样本名称\n# names(datExpr);\n# 指定颜色数据名称\n# names(datExpr)[moduleColors==\"brown\"]\n\n# 基因注释数据\nannot = read.csv(file = \"./data/GeneAnnotation.csv\");\ndim(annot)\nnames(annot)\nprobes = names(datExpr)\nprobes2annot = match(probes, annot$substanceBXH)\n# The following is the number or probes without annotation:\nsum(is.na(probes2annot))\n\n# Create the starting data frame\ngeneInfo0 = data.frame(substanceBXH = probes,\ngeneSymbol = annot$gene_symbol[probes2annot],\nLocusLinkID = annot$LocusLinkID[probes2annot],\nmoduleColor = moduleColors,\ngeneTraitSignificance,\nGSPvalue)\n# Order modules by their significance for weight\nmodOrder = order(-abs(cor(MEs, weight, use = \"p\")));\n# Add module membership information in the chosen order\nfor (mod in 1:ncol(geneModuleMembership))\n{\n oldNames = names(geneInfo0)\n geneInfo0 = data.frame(geneInfo0, geneModuleMembership[, modOrder[mod]],\n MMPvalue[, modOrder[mod]]);\n names(geneInfo0) = c(oldNames, paste(\"MM.\", modNames[modOrder[mod]], sep=\"\"),\n paste(\"p.MM.\", modNames[modOrder[mod]], sep=\"\"))\n}\n# Order the genes in the geneInfo variable first by module color, then by geneTraitSignificance\ngeneOrder = order(geneInfo0$moduleColor, -abs(geneInfo0$GS.weight));\ngeneInfo = geneInfo0[geneOrder, ]\nwrite.csv(geneInfo, file = \"geneInfo.csv\")\n\n# Calculate topological overlap anew: this could be done more efficiently by saving the TOM\n# calculated during module detection, but let us do it again here.\n# 重新计算拓扑重叠：通过保存TOM可以更有效地完成此操作\n# 是在模块检测期间计算的，但让我们在这里再次进行。\ndissTOM = 1-TOMsimilarityFromExpr(datExpr, power = 6);\n# Transform dissTOM with a power to make moderately strong connections more visible in the heatmap\n# 变换dissTOM\nplotTOM = dissTOM^7;\n# Set diagonal to NA for a nicer plot\ndiag(plotTOM) = NA;\n# Call the plot function\n# sizeGrWindow(9,9)\n# 基因的聚类树聚类时的距离为1-TOM值结合基因间的距离，即1-TOM值，用热图展示\n# TOMplot(plotTOM, geneTree, moduleColors, main = \"Network heatmap plot, all genes\")\n\nnSelect = 400\n# For reproducibility, we set the random seed\nset.seed(10);\nselect = sample(nGenes, size = nSelect);\nselectTOM = dissTOM[select, select];\n# There’s no simple way of restricting a clustering tree to a subset of genes, so we must re-cluster.\n# 重新画聚类图\nselectTree = hclust(as.dist(selectTOM), method = \"average\")\nselectColors = moduleColors[select];\n# Open a graphical window\n# sizeGrWindow(9,9)\n# Taking the dissimilarity to a power, say 10, makes the plot more informative by effectively changing\n# the color palette; setting the diagonal to NA also improves the clarity of the plot\nplotDiss = selectTOM^7;\ndiag(plotDiss) = NA;\nTOMplot(plotDiss, selectTree, selectColors, main = \"Network heatmap plot, selected genes\")\n\n# Recalculate module eigengenes\n# 重新计算基因特征值\nMEs = moduleEigengenes(datExpr, moduleColors)$eigengenes\n# Isolate weight from the clinical traits\nweight = as.data.frame(datTraits$weight_g);\nnames(weight) = \"weight\"\n# Add the weight to existing module eigengenes\nMET = orderMEs(cbind(MEs, weight))\n# Plot the relationships among the eigengenes and the trait\n#sizeGrWindow(5,7.5);\npar(cex = 0.9)\n# 画树形图\n# marDendro给出树状图的边距设置，marHeatmap热图边距设置\nplotEigengeneNetworks(MET, \"\", marDendro = c(0,4,1,2), marHeatmap = c(3,4,1,2), cex.lab = 0.8, xLabelsAngle= 90)\n\n# Plot the dendrogram\n# sizeGrWindow(6,6);\npar(cex = 1.0)\nplotEigengeneNetworks(MET, \"Eigengene dendrogram\", marDendro = c(0,4,2,0),\nplotHeatmaps = FALSE)\n# Plot the heatmap matrix (note: this plot will overwrite the dendrogram plot)\npar(cex = 1.0)\nplotEigengeneNetworks(MET, \"Eigengene adjacency heatmap\", marHeatmap = c(3,4,2,2),plotDendrograms = FALSE, xLabelsAngle = 90)\n", "meta": {"hexsha": "b32dec4c5f3a592431fdb23dc782e683034e818e", "size": 12723, "ext": "r", "lang": "R", "max_stars_repo_path": "WGCNA/wgcna_tutorial.r", "max_stars_repo_name": "luohenyueji/R-Study-Notes", "max_stars_repo_head_hexsha": "e6021d86a15f46a981c09c5e2c1b6766c6470462", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 4, "max_stars_repo_stars_event_min_datetime": "2020-10-19T09:43:42.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-19T08:58:33.000Z", "max_issues_repo_path": "WGCNA/wgcna_tutorial.r", "max_issues_repo_name": "qiuli-github/R-Study-Notes", "max_issues_repo_head_hexsha": "eb2fb7d06eabe45622f378ca2d07fd50ac2e21fb", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "WGCNA/wgcna_tutorial.r", "max_forks_repo_name": "qiuli-github/R-Study-Notes", "max_forks_repo_head_hexsha": "eb2fb7d06eabe45622f378ca2d07fd50ac2e21fb", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 5, "max_forks_repo_forks_event_min_datetime": "2020-06-03T07:02:42.000Z", "max_forks_repo_forks_event_max_datetime": "2021-10-31T06:07:55.000Z", "avg_line_length": 36.2478632479, "max_line_length": 125, "alphanum_fraction": 0.7381120805, "num_tokens": 4658, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7662936484231889, "lm_q2_score": 0.6757646010190476, "lm_q1q2_score": 0.5178341215901265}} {"text": "# Analyze Data\n\n# Requires loading/entering/computing\n# MOD \n# -the model specification matrix \n# -based on RAM specification in \"sem\" package\n# -create mod from path diagram (see below)\n# Data\n# -the covariance matrix of the observed variables\n# -variable order should be consistent with MOD\n\n###########################\n##Load Required Packages ##\n###########################\n\nrequire('MASS') #Multivariate Normal Distributions\nrequire('matrixcalc') #Special Matrices\nrequire('numDeriv') #Needed to estimate SEs\nsource('scvm_functions.r') #Custom functions to fit CV models\n\n#######################\n## LOAD DATA & MODEL ##\n#######################\n\n#Enter data file name here:\nY = read.table('cv_sample_data.dat', header=TRUE)\n\n#Specify MOD file here: \nMOD <- read.delim(\"MOD.dat\") ## Read in model here\n\n################################\n## Transform and Check Inputs ##\n################################\n\nY = as.matrix(Y)\nhead(Y)\nhead(MOD)\n\n###################################\n## Count and Selection Variables ##\n###################################\n\n#sample size\nN = nrow(Y)\n\n#total number of variables \np = max(c(MOD$from,MOD$to))\ncat(paste('Total Number of Variables: ',p))\n# ***Assumes no unused variables\n\n#number of observed variables\npOBS = ncol(Y) \ncat(paste('Number of Observed Variables: ', pOBS))\n# ***Assumes variables 1:pOBS are all observed and included in model\n\n#Number of unique covariances\npSTAR = pOBS*(pOBS+1)/2\ncat(paste('Number of Unique Covariance Terms: ', pSTAR))\n\n#Select/Identify Paths and Vars\nsel.ip <- (MOD$code == 1)#paths from ivs\nsel.dp <- (MOD$code == 2)#paths between dvs\nsel.cov <- (MOD$code == 3)#vars and covs of ivs\n\n#Identify and count (strict) IVs \nlistIV = unique(MOD$from[sel.ip])\npIV = length(listIV) #includes errors\ncat(paste('Number of Strictly Independent Variables: ', pIV))\n\n#Identify and count (all) DVs\nlistDV = unique(MOD$to[MOD$code != 3])\npDV = length(listDV)\ncat(paste('Number of Dependent Variables: ', pDV))\n\n#Select/Identify Free Parameters\nsel.fixed <- (MOD$index == 0)\nsel.free <- !(sel.fixed)\n\n#number of parameters\nqTOT = nrow(MOD)\nqFREE = sum(sel.free)\ncat(paste('Number of Free Parameter: ', qFREE))\n# *** Assumes no constraints on parameters\n\n#df for chi-squared\ndf = pSTAR - qFREE\ncat(paste('Number of Degrees of Freedom: ', df))\n\n############################\n## Some Useful Statistics ##\n############################\n\n#The mean\nM = colMeans(Y)\n#Warning: This doesn't return a transposable vector! \nDmInv = diag(1/M)\ncat('Sample Means:')\nprint(M)\n\n#Sample Covariance Matrix\nS = var(Y)\ncat('Sample Covariance Matrix:')\nprint(S)\n\n#Sample CV Matrix\nPsyHat = DmInv %*% S %*% DmInv\ncat('Sample Coefficient of Variation (CV) Matrix:')\nprint(PsyHat)\n\n##########################\n## Compute Start Values ##\n##########################\nMOD = getstart(MOD,S)\n\n#############################################\n## Define Matrices for Bentler-Weeks Model ##\n#############################################\n\n#Get Matrix G to Select Observed Variables\n G = matrix(0,pOBS,p)\n G[cbind(1:pOBS,1:pOBS)] <- 1\n # ***Assumes variables 1:pOBS are observed and included\n\n#Get Matrix B to relate DVs with DVs (contains \"beta and known 0s\")\n B = matrix(0,p,p)\n B[cbind(MOD$to[sel.dp],MOD$from[sel.dp])]=MOD$value[sel.dp]\n\n#Define Reindex Function\n #function to recode IVs for Gam, Phi\n reindex <- function(x) {\n #initialize\n xnew = x\n #get ordered list of unique xvals\n sux = sort(unique(x))\n #loop through x\n for(i in 1:length(x)) {\n #loop over unique values\n for(k in 1:length(sux)) {\n if(x[i] == sux[k])\n xnew[i] = k\n }\n }\n xnew\n }\n\n#Get Matrix Gamma to relate IVs with DVs\n #Initialize\n Gam = matrix(0,p,pIV)\n #List IVs with New Unique Indices\n ivfrom = reindex(MOD$from[sel.ip])\n #Calculate Gamma\n Gam[cbind(MOD$to[sel.ip],ivfrom)] = MOD$value[sel.ip]\n Gam[cbind(MOD$from[sel.ip],ivfrom)] = MOD$value[sel.ip]\n \n#Get CovMat Phi containing covs of IVs\n #Initialize\n Phi = diag(1,pIV,pIV)\n #Find Covariances\n these = c(unique(MOD$from[sel.ip]),MOD$to[sel.cov],MOD$from[sel.cov])\n #List Covariances with New Unique Indices\n Cindex = reindex(these)\n Cindex = Cindex[-(1:pIV)]\n Cindex = matrix(Cindex,ncol=2,byrow=F)\n #Put Covariances into Phi Matrix\n Phi[Cindex] = MOD$value[sel.cov]\n Phi = Phi + t(Phi) - diag(diag(Phi))\n\n###########################################\n## Get Sampling Variances for ADF Method ##\n###########################################\n\n#################\n## SigmaHatPsy ##\n#################\n\n#Get SigmaPsyHat for CV-SEM Method\n #Identity\n Ip = diag(pOBS)\n #Kronecker Product of CV and Ip\n PsyHatI = kronecker(PsyHat,Ip)\n #Duplication and Elimination Matrices\n Dp = D.matrix(pOBS)\n Hp = solve(t(Dp) %*% Dp) %*% t(Dp)\n #Np (Funky Product of Dup Matrices)\n Np = Dp %*% solve( t(Dp) %*% Dp ) %*% t(Dp)\n #Lp\n Lp=matrix(0,pOBS^2,pOBS)\n for(i in 1:pOBS) {\n Lp = Lp + kronecker(Ip[,i],Ip[,i]) %*% t(Ip[,i])\n }\n #Transform Means Vector\n kronDmInv = kronecker(DmInv,DmInv)\n #Estimate Omegas [Prospectus (3.11) and (3.12)]\n #Get Sums of Products for OmegaHat12 and OmegaHat22\n sumthing12 = matrix(0,pOBS,pOBS^2)\n sumthing22 = matrix(0,pOBS^2,pOBS^2)\n for(i in 1:N) {\n #Precursors for Omega12\n ui = kronecker(Y[i,]-M,Y[i,] - M)\n prodthing12 = (Y[i,] - M) %*% t(ui)\n sumthing12 = sumthing12 + prodthing12\n #Precursors for Omega22\n prodthing22 = ui %*% t(ui)\n sumthing22 = sumthing22 + prodthing22\n }\n #Get OmegaHat12\n OmegaHat12 = N/((N-1)*(N-2))* sumthing12\n OmegaHat12str =DmInv %*% OmegaHat12 %*% kronDmInv\n #Get OmegaHat22\n OmegaHat22 = (sumthing22 - (N-2)*vec(S)%*%t(vec(S)) )/(N-pSTAR-1)\n OmegaHat22str = kronDmInv %*% OmegaHat22 %*% kronDmInv\n #Variance of CV Sampling Distribution [Prospectus (3.15)]\n SigmaHatPsy = (2*Np) %*% PsyHatI %*% Lp %*% PsyHat %*% t(Lp) %*% \n PsyHatI %*% (2*Np) +\n -t(OmegaHat12str) %*% t(Lp) %*% PsyHatI %*% (2*Np) +\n -(2*Np) %*% PsyHatI %*% Lp %*% OmegaHat12str + OmegaHat22str\n SHPinv = solve(Hp %*% SigmaHatPsy %*% t(Hp))\n #Variance of CV Sampling Distribution w/Normality [Prospectus (3.16)]\n SigmaHatPsyN = 2*Np %*% PsyHatI %*% Lp %*% PsyHat %*% t(Lp) %*% \n PsyHatI %*% (2*Np) +\n 2*Np %*% kronecker(PsyHat,PsyHat)\n SHPNinv = solve(Hp %*% SigmaHatPsyN %*% t(Hp))\n\n\n#########################################\n## Fit CV Models and Examine Estimates ##\n#########################################\n\n#########################\n## ADF - CV (Arbitrary)##\n#########################\n\nMETH = 'ADF Estimation with Arbitrary Distribution'\n#Fit Model Using ADFcv procedure\n theta = MOD$value[sel.free]\n rADFcv = plzcon(theta,fADFcv)\n rADFcv$METH = METH\n#SEs of ADFcv Estimates\n DeltaADFcv = getDelta(rADFcv,SHPinv)\n rADFcv$SEs = sqrt(diag(DeltaADFcv)/N)\n rADFcv=compsings(rADFcv)\n#Results\n printRES(rADFcv)\n cat('Sample CV Matrix:')\n print(PsyHat)\n cat('Reproduced CV Mat:')\n print(SIGMAof(rADFcv$pars))\n cat('Covariance Residuals:')\n sdres_ADF = (vech(PsyHat) - vech(SIGMAof(rADFcv$pars)))/sd(vech(PsyHat))\n print(PsyHat - SIGMAof(rADFcv$pars))\n cat('Covariance Residuals as Percent Difference:')\n print((PsyHat - SIGMAof(rADFcv$pars))*100/PsyHat)\n\n\n#####################\n## ADF - CV Normal ##\n#####################\n\nMETH = 'ADF Estimation with Normal Distribution'\n#Fit Model Using ADFcvn procedure\n theta = MOD$value[sel.free]\n rADFcvn = plzcon(theta,fADFcvn)\n rADFcvn$METH = METH\n#SEs of ADF Estimates\n DeltaADFcvn = getDelta(rADFcvn,SHPNinv)\n rADFcvn$SEs = sqrt(diag(DeltaADFcvn)/N) \n rADFcvn=compsings(rADFcvn)\n#Results\n printRES(rADFcvn)\n cat('Sample CV Matrix:')\n print(PsyHat)\n cat('Reproduced CV Mat:')\n print(SIGMAof(rADFcvn$pars))\n cat('Covariance Residuals:')\n sdres_ADFN = (vech(PsyHat) - vech(SIGMAof(rADFcvn$pars)))/sd(vech(PsyHat))\n print(PsyHat - SIGMAof(rADFcvn$pars))\n cat('Covariance Residuals as Percent Difference:')\n print((PsyHat - SIGMAof(rADFcvn$pars))*100/PsyHat)\n\n##############\n## GLS - CV ##\n##############\n\nMETH = 'GLS Estimation'\n#Fit Model Using GLS procedure\n theta = MOD$value[sel.free]\n rGLScv = plzcon(theta,fGLScv)\n rGLScv$METH = METH\n#SEs of GLS Estimates\n DeltaGLScv = getDelta(rGLScv,solve(PsyHat))\n rGLScv$SEs = sqrt(diag(DeltaGLScv)/N)\n rGLScv=compsings(rGLScv)\n#Results\n printRES(rGLScv)\n cat('Sample CV Mat:')\n print(PsyHat)\n cat('Reproduced CV Mat:')\n print(SIGMAof(rGLScv$pars))\n cat('Covariance Residuals:')\n sdres_GLS = (vech(PsyHat) - vech(SIGMAof(rGLScv$pars)))/sd(vech(PsyHat))\n print(PsyHat - SIGMAof(rGLScv$pars))\n cat('Covariance Residuals as Percent Difference:')\n print((PsyHat - SIGMAof(rGLScv$pars))*100/PsyHat)\n\n#############\n## ML - CV ##\n#############\n\nMETH = 'Maximum Likelihood Estimation'\n#Fit Model Using ML procedure\n theta = MOD$value[sel.free]\n rMLcv = plzcon(theta,fMLcv)\n rMLcv$METH = METH\n#SEs of ML Estimates\n DeltaMLcv = getDelta(rMLcv,solve(SIGMAof(rMLcv$pars)))\n rMLcv$SEs = sqrt(diag(DeltaMLcv)/N)\n rMLcv=compsings(rMLcv)\n#Result\n printRES(rMLcv)\n cat('Sample CV Mat:')\n print(PsyHat)\n cat('Reproduced CV Mat:')\n print(SIGMAof(rMLcv$pars))\n cat('Covariance Residuals:')\n sdres_ML = (vech(PsyHat) - vech(SIGMAof(rMLcv$pars)))/sd(vech(PsyHat))\n print(PsyHat - SIGMAof(rMLcv$pars))\n cat('Covariance Residuals as Percent Difference:')\n print((PsyHat - SIGMAof(rMLcv$pars))*100/PsyHat)\n\n#Plot Histograms of Standardized Residuals\npar(mfrow = c(2,2),\n mar=c(3, 4, 1.5, 1.5)+0.1,\n cex=0.9)\nxtitle = \"Standardized Residual\"\nhist(sdres_ADF,\n ylab=NULL,xlab=NULL,main=NULL,\n ylim=c(0,8),xlim=c(-0.5,0.5),\n col=\"gray\")\nmtext(\"AGLS\",3,line=-1.5,cex=0.9)\nmtext(xtitle,1,line=2,cex=0.9)\nmtext(\"Frequency\",2,line=2,cex=0.9)\n\nhist(sdres_ADFN,\n ylab=NULL,xlab=NULL,main=NULL,\n ylim=c(0,8),xlim=c(-0.5,0.5),\n col=\"gray\")\nmtext(\"NGLS\",3,line=-1.5,cex=0.9)\nmtext(xtitle,1,line=2,cex=0.9)\nmtext(\"Frequency\",2,line=2,cex=0.9)\n\nhist(sdres_GLS,\n ylab=NULL,xlab=NULL,main=NULL,\n ylim=c(0,8),xlim=c(-0.5,0.5),\n col=\"gray\")\nmtext(\"MGLS\",3,line=-1.5,cex=0.9)\nmtext(xtitle,1,line=2,cex=0.9)\nmtext(\"Frequency\",2,line=2,cex=0.9)\n\nhist(sdres_ML,\n ylab=NULL,xlab=NULL,main=NULL,\n ylim=c(0,8),xlim=c(-0.5,0.5),\n col=\"gray\")\nmtext(\"MRLS\",3,line=-1.5,cex=0.9)\nmtext(xtitle,1,line=2,cex=0.9)\nmtext(\"Frequency\",2,line=2,cex=0.9)\n\nParameter_Table = rbind(rADFcv$pars, rADFcv$SEs,\n rADFcvn$pars,rADFcvn$SEs,\n rGLScv$pars, rGLScv$SEs,\n rMLcv$pars, rMLcv$SEs)\nrow.names(Parameter_Table) <- c('ADF Param Est',\n 'ADF SEs',\n 'Normal ADF Param Est',\n 'Normal ADF SEs',\n 'GLS Param Est',\n 'GLS SEs',\n 'ML Param Est',\n 'ML SEs')\nView(Parameter_Table)\n\nCV_Matrix_Table = t(cbind(vech(PsyHat), \n vech(SIGMAof(rADFcv$pars)), \n vech(SIGMAof(rADFcvn$pars)), \n vech(SIGMAof(rGLScv$pars)),\n vech(SIGMAof(rMLcv$pars))))\nrow.names(CV_Matrix_Table) = c(\n 'Sample CV Matrix', \n 'ADF Modeled CV Matrix',\n 'Normal ADF Modeled CV Matrix',\n 'GLS Modeled CV Matrix',\n 'ML Modeled CV Matrix')\nView(CV_Matrix_Table)\n", "meta": {"hexsha": "dbe8adaf64cc51033f995ab8ae8135c7bcc9243f", "size": 11562, "ext": "r", "lang": "R", "max_stars_repo_path": "fitCVmod.r", "max_stars_repo_name": "alextrickey/StructuralCVModels", "max_stars_repo_head_hexsha": "3bc5781bde0e8243fe38115133e4f7b4bc1949d0", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "fitCVmod.r", "max_issues_repo_name": "alextrickey/StructuralCVModels", "max_issues_repo_head_hexsha": "3bc5781bde0e8243fe38115133e4f7b4bc1949d0", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "fitCVmod.r", "max_forks_repo_name": "alextrickey/StructuralCVModels", "max_forks_repo_head_hexsha": "3bc5781bde0e8243fe38115133e4f7b4bc1949d0", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 29.4198473282, "max_line_length": 76, "alphanum_fraction": 0.5953987199, "num_tokens": 3760, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8080672043084051, "lm_q2_score": 0.640635854839898, "lm_q1q2_score": 0.5176768242002017}} {"text": "#' all the below code is cannibalized from aRbor\n\n#' this code attempts to detect whether data is discrete or continuous for use in calculating phylogenetic signal\n#' it uses a rather heuristic method to determine if something is discrete, so be careful!\n\ndetectCharacterType<-function(dat) {\n\tif(is.factor(dat)) {\n\t\t\tcharType<-\"discrete\"\n\t} else if(nlevels(as.factor(dat))/length(dat) < 0.1) {\n\t\t\twarning(\"Guessing that this is a discrete character based on repeated values\")\n\t\t\tcharType<-\"discrete\"\n\t} else {\n\t\t\tcharType<-\"continuous\"\n\t}\n\treturn(charType)\n}\t# needless to say, this is not yet robust\n\n#' Function for calculating phylogenetic signal of discrete and continuous traits\n#' \n#' This function allows testing either Blomberg (for continuous), Pagel lambda (for both), or \"garbage test\" (for discrete)\n\nphysigArbor<-function(phy, dat, charType=\"fromData\", signalTest=\"pagelLambda\", discreteModelType=\"ER\") {\n\tctype = match.arg(charType, c(\"fromData\", \"discrete\", \"continuous\"))\n\n\tif(ctype==\"fromData\") # then try to figure it out\n\t\tctype<-detectCharacterType(dat)\n\t\n\tsignalTest = match.arg(signalTest, c(\"pagelLambda\", \"Blomberg\", \"garbageTest\"))\n\tdiscreteModelType = match.arg(discreteModelType, c(\"ER\", \"SYM\", \"ARD\"))\n\n\t\n\tif(ctype==\"discrete\") {\n\t\t# this changes the discrete data to 1:n and remembers the original charStates\n\t\tdat<-as.factor(dat)\n\t\tcharStates<-levels(dat)\n\t\tk<-nlevels(dat)\n\t\t\n\t\tndat<-as.numeric(dat)\n\t\tnames(ndat)<-names(dat)\n\t\t\n\t\tif(signalTest==\"pagelLambda\") {\n\t\t\tres<-discreteLambdaTest(phy, ndat, k, discreteModelType)\n\t\t} else if(signalTest==\"garbageTest\") {\n\t\t\tres<-discreteGarbageTest(phy, ndat, k, discreteModelType)\n\t\t} else if(signalTest==\"Blomberg\") {\n\t\t\tstop(\"Blomberg K should not be used for discrete characters\")\n\t\t}\n\t}\n\t\n\tif(ctype==\"continuous\") {\n\t\tif(signalTest==\"pagelLambda\") {\n\t\t\tres<-continuousLambdaTest(phy, dat)\n\t\t} else if(signalTest==\"garbageTest\") {\n\t\t\tres<-continuousGarbageTest(phy, dat)\n\t\t} else if(signalTest==\"Blomberg\") {\n\t\t\tres<-continuousBlombergTest(phy, dat)\n\t\t}\n\t}\n\t\n\treturn(res)\n\n}\n\ndiscreteLambdaTest<-function(phy, ndat, k, discreteModelType) {\n\tm1<-fitDiscrete(phy, ndat, model=discreteModelType)\n\tm2<-fitDiscrete(phy, ndat, model=discreteModelType, transform=\"lambda\")\n\t\n\tchisqTestStat <- 2 * (m2$opt$lnL-m1$opt$lnL)\n\tchisqPVal <- pchisq(chisqTestStat, 1, lower.tail=F)\n\t\n\taicScores<-c(m1$opt$aicc, m2$opt$aicc)\n\tnames(aicScores)<-c(\"Mk\", \"Mk+lambda\")\n\t\n\tres<-list(chisqTestStat= chisqTestStat, chisqPVal= chisqPVal, aicScores= aicScores)\n\treturn(res)\n}\n\ndiscreteGarbageTest<-function(phy, ndat, k, discreteModelType) {\n\tm1<-fitDiscrete(phy, ndat, model=discreteModelType)\n\tm2<-fitDiscreteGarbageModel(phy, ndat)\n\t\n\taiccScores<-c(m1$opt$aicc, m2$aicc)\n\tnames(aiccScores)<-c(\"Mk\", \"Garbage\")\n\t\n\tres<-list(aiccScores= aiccScores)\n\treturn(res)\n}\n\nfitDiscreteGarbageModel<-function(phy, ndat) {\n\ttt <- table(ndat)\n\tprob <- tt/sum(tt)\n\tlnL <- sum(tt * log(prob))\n\tk <- length(tt)-1\n\tn <- length(ndat)\n\taic <- -2 * lnL + 2 * k\n\taicc <- aic + 2 * k * (k+1) / (n - k - 1)\n\tres<-list(prob=prob, lnL=lnL, k=k, aic=aic, aicc=aicc)\n\treturn(res)\n}\n\n\ncontinuousLambdaTest<-function(phy, dat) {\n\tm1<-fitContinuous(phy, dat, model=\"BM\")\n\tm2<-fitContinuous(phy, dat, model=\"lambda\")\n\t\n\tchisqTestStat <- 2 * (m2$opt$lnL-m1$opt$lnL)\n\tchisqPVal <- pchisq(chisqTestStat, 1, lower.tail=F)\n\t\n\taicScores<-c(m1$opt$aicc, m2$opt$aicc)\n\tnames(aicScores)<-c(\"BM\", \"BM+lambda\")\n\t\n\tres<-list(chisqTestStat= chisqTestStat, chisqPVal= chisqPVal, aicScores= aicScores)\n\treturn(res)\n}\n\n\ncontinuousGarbageTest<-function(phy, dat) {\n\tm1<-fitContinuous(phy, dat)\n\tm2<-fitContinuous(phy, dat, model=\"white\")\n\t\n\taiccScores<-c(m1$opt$aicc, m2$opt$aicc)\n\tnames(aiccScores)<-c(\"Mk\", \"WhiteNoise\")\n\t\n\tres<-list(aiccScores= aiccScores)\n\treturn(res)\n}\n\ncontinuousBlombergTest<-function(phy, dat) {\n\tres<-phylosignal(dat, phy)\n\treturn(res)\n}", "meta": {"hexsha": "e632273af3f5ff8f8114c8628ed2e9b4456b2ef9", "size": 3871, "ext": "r", "lang": "R", "max_stars_repo_path": "rAnalysis/phyloSignalARbor.r", "max_stars_repo_name": "lukejharmon/traitathon", "max_stars_repo_head_hexsha": "247e1691859141e2564974a8771d1efa3a8e7aed", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2016-02-20T18:08:05.000Z", "max_stars_repo_stars_event_max_datetime": "2016-07-14T01:41:01.000Z", "max_issues_repo_path": "rAnalysis/phyloSignalARbor.r", "max_issues_repo_name": "lukejharmon/traitathon", "max_issues_repo_head_hexsha": "247e1691859141e2564974a8771d1efa3a8e7aed", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2015-07-05T15:00:30.000Z", "max_issues_repo_issues_event_max_datetime": "2015-07-14T03:28:03.000Z", "max_forks_repo_path": "rAnalysis/phyloSignalARbor.r", "max_forks_repo_name": "lukejharmon/traitathon", "max_forks_repo_head_hexsha": "247e1691859141e2564974a8771d1efa3a8e7aed", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 29.5496183206, "max_line_length": 123, "alphanum_fraction": 0.70679411, "num_tokens": 1262, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8267117940706734, "lm_q2_score": 0.6261241772283034, "lm_q1q2_score": 0.5176242418674349}} {"text": "\n## * Utilities\n\nupper1st <- function(x) {\n if (is.factor(x))\n x0 <- levels(x) else x0 <- as.character(x)\n nc <- nchar(x0)\n x0[nc > 1] <- sprintf('%s%s', toupper(substr(x0[nc > 1], 1, 1)), substr(x0[nc > 1], 2, nchar(x0[nc > 1])))\n x0[nc == 1] <- toupper(x0[nc == 1])\n if (is.factor(x))\n levels(x) <- x0 else x <- x0\n return(x)\n}\n\nslugify <- function(x) {\n gsub(' +', '_', gsub('\\'', '', tolower(x)))\n}\n\n## * Statistical\n\nestBetaParams <- function(mu, var) {\n ## Estimate alpha and beta of beta distribution from mean and variance\n ## from http://stats.stackexchange.com/questions/12232/calculating-the-parameters-of-a-beta-distribution-using-the-mean-and-variance\n alpha <- ((1 - mu) / var - 1 / mu) * mu ^ 2\n beta <- alpha * (1 / mu - 1)\n return(params = list(alpha = alpha, beta = beta))\n}\n\n\n## * For PST\n\n## ** Rmax formula from Niel & Lebreton 2008\nfunNL<- function(x,a,s) (exp((a+s/(x-s))^-1)-x)^2\nlmax_nl <- function(a,s) return(optimise(funNL, c(1,2), a=a, s=s, tol=1e-10)$minimum)\nLmax_nl <- function(a,s,usemc=F) {\n if (usemc) {\n library(parallel)\n return(mcmapply(lmax_nl, a, s))\n } else return(mapply(lmax_nl, a, s))\n}\nRmax_NL <- function(s,a,...) return(Lmax_nl(a,s,...)-1)\n\n\n## * For PST\n\n## formula from Neil & Lebreton 2008\nfunNL<- function(x,a,s) (exp((a+s/(x-s))^-1)-x)^2\nlmax_nl <- function(a,s) return(optimise(funNL, c(1,2), a=a, s=s, tol=1e-10)$minimum)\nLmax_nl <- function(a,s,usemc=F) {\n if (usemc) {\n library(parallel)\n return(mcmapply(lmax_nl, a, s))\n } else return(mapply(lmax_nl, a, s))\n}\nRmax_NL <- function(s,a,...) return(Lmax_nl(a,s,...)-1)\n\n\n ## Survival variation\nsurv_var_norm_mean_se <- function(n=1000, smean, sse)\n {\n se_beta <- sse/(smean*(1-smean)) ## calculate se(beta(S)) from se(S), uses delta method\n logit_s <- rnorm(n=n, mean=log(smean/(1-smean)), sd=se_beta) ## apply normal variation to logit(s)\n surv <- exp(logit_s)/(1+exp(logit_s)) ## back-transform\n return(surv)\n }\n## surv_var_norm_mean_se1 <- Vectorize(surv_var_norm_mean_se0, SIMPLIFY=F)\n## surv_var_norm_mean_se <- function(n=1000, smean, sse)\n## do.call('rbind', surv_var_norm_mean_se1(n=1000, smean, sse))\n\nsurv_ci0 <- function(smean, sse)\n {\n se_beta <- sse/(smean*(1-smean)) ## calculate se(beta(S)) from se(S)\n logit_s <- log(smean/(1-smean))\n ll <- logit_s-1.96*se_beta\n ul <- logit_s+1.96*se_beta\n ll <- exp(ll) / (1 + exp(ll))\n ul <- exp(ul) / (1 + exp(ul))\n return(data.frame(ll=ll, ul=ul))\n }\nsurv_ci1 <- Vectorize(surv_ci0, SIMPLIFY=F)\nsurv_ci <- function(smean, sse) do.call('rbind', surv_ci1(smean, sse))\n\nsurv_meansd_from_ci0 <- function(lcl, ucl)\n {\n logitlcl <- log(lcl/(1-lcl))\n logitucl <- log(ucl/(1-ucl))\n sdlogit <- (logitucl-logitlcl) / (2*1.96)\n meanlogit <- mean(c(logitlcl,logitucl))\n mean <- exp(meanlogit)/(1+exp(meanlogit))\n sd <- sdlogit*(mean*(1-mean))\n return(data.frame(mean=mean, sd=sd))\n }\nsurv_meansd_from_ci1 <- Vectorize(surv_meansd_from_ci0, SIMPLIFY=F)\nsurv_meansd_from_ci <- function(lcl, ucl) do.call('rbind', surv_meansd_from_ci1(lcl, ucl))\n\n\nntot2 <- function(Nadtot, surv, afr, nsims=4000) {\n if (length(Nadtot)==1 | is.null(names(Nadtot))) {\n n <- Nadtot\n } else { ## point estimate or vector of samples\n\tif (all(names(Nadtot)==c('mean','se'))) { ## mean-se\n\t n <- surv_var_norm_mean_se(nsims, smean=Nadtot['mean'], sse=Nadtot['se'])\n } else {\n if (prod(names(Nadtot)==c('min','max'))) { ## min-max\n n <- runif(nsims, min=Nadtot['min'], max=Nadtot['max'])\n }\n }\n }\n if (length(surv)==1 | is.null(names(surv))) {\n s <- surv\n } else {\n\tif (prod(names(surv)==c('mean','se'))) {\n\t s <- surv_var_norm_mean_se(nsims, smean=surv['mean'], sse=surv['se'])\n } else {\n if (prod(names(surv)==c('min','max'))) {\n s <- runif(nsims, min=surv['min'], max=surv['max'])\n }\n }\n }\n if (length(afr)==1 | is.null(names(afr))) {\n a <- afr\n } else {\n\tif (prod(names(afr)==c('mean','se'))) {\n\t a <- rnorm(nsims, mean=afr['mean'], sd=afr['se'])\n } else {\n if (prod(names(afr)==c('min','max'))) {\n a <- runif(nsims, min=afr['min'], max=afr['max'])\n }\n }\n }\n\n ## maximum age\n m=1+log(0.02)/log(s)\n\n ## Indiv/adult ratio\n r = s^(1-a)\n\n ## Total population\n ntot = n*r\n\n if (min(ntot)<0)\n print('===================================>>> WARNING!!! Minimum of Ntot is negative!')\n\n return(list(all=ntot, min=quantile(ntot,0.2), s=s, a=a, r=r, m=m))\n ## for min: take the lower 60 percentile, following Wade\n}\n\n\n\n## * South Pole-centered map of grid values\n\nmap_grid_values <- function(grid_values, legend.title='', colourscale.trans='log1p', zeros_to_na=T,\n leg.pos = c(0.95, 0.15),\n pal=c(\"#E3F2FD\",\"#BBDEFB\",\"#90CAF9\",\"#64B5F6\",\"#42A5F5\",\"#2196F3\",\"#1E88E5\",\"#1976D2\",\"#1565C0\",\"#0D47A1\"),\n bg.col = 'grey95', graticule.col = '#B0B0B0', \n highlighted.lat = NA, highlighted.lat.col = 'grey', min0=T, grid = grid_noland, ...) {\n\n suppressPackageStartupMessages({\n library(data.table)\n library(ggplot2)\n library(colorspace)\n library(sf)\n })\n if (ncol(grid_values) != 2)\n stop('`grid_values` needs to have two columns only: grid ids and values')\n\n if (any(duplicated(grid_values$grid_id)))\n stop('Duplicated `grid_id` values')\n\n grid_values <- copy(grid_values)\n setDT(grid_values)\n setnames(grid_values, c('grid_id', 'value'))\n \n values <- grid_values[match(grid$grid_id, grid_id), value]\n if (zeros_to_na) {\n values[values == 0] <- NA\n }\n grid$value <- values\n\n logtrans <- grepl('log', colourscale.trans)\n if (logtrans) {\n \n mx <- max(pretty(c(0, max(grid$value, na.rm=T))))\n z <- round(mx / (10^(-10:10)))\n z <- (-10:10)[min(which(z > 0 & nchar(z) == 1))]\n brks <- c(1, 3) * rep(10^((z-1):z), each=2)\n\n } else {\n brks <- pretty(grid$value) \n }\n if (min0) {\n brks <- unique(c(0, brks))\n }\n\n bb <- st_bbox(grid)\n world <- suppressMessages(st_crop(oneworld, bb))\n world <- st_segmentize(world, 100000)\n \n g <- ggplot() +\n geom_sf(data=grid, aes(fill = value), colour = graticule.col, size = 0.1) +\n scale_fill_gradientn(colours = pal, name = legend.title,\n na.value='white', limits = c(0, max(brks)),\n trans = colourscale.trans, breaks = brks)\n\n if (!is.na(highlighted.lat)) {\n hlat <- st_as_sf(as(SpatialLines(list(Lines(list(Line(cbind(x = seq(0, 360, 1),\n y = rep(highlighted.lat, 361)))), ID=1)),\n proj4string=CRS(\"+init=epsg:4326 +over\")), 'SpatialLines'))\n g <- g + geom_sf(dat = hlat, colour = 'grey')\n }\n\n g <- g + \n ## ** world\n geom_sf(data = world, fill = '#AAAAAA', colour = NA, size = 0.1) +\n theme_void() +\n theme(plot.margin = unit(rep(3, 4), 'mm'),\n legend.position=leg.pos,\n legend.margin = margin(5, 2, 2, 2, unit='mm'),\n legend.background=element_rect(fill = '#FFFFFF44'),\n legend.title.align = 0,\n legend.text.align = 0,\n legend.key.width = unit(0.5, 'cm'),\n plot.background = element_rect(fill = bg.col)) +\n coord_sf(crs=\"+proj=ortho +y_0=0 +lon_0=178 +lat_0=-90.0\", expand=F, datum=NA)\n\n return(g)\n}\n\n\n\n", "meta": {"hexsha": "a2305be63fbffdee3291d3d9a2c099d05ef67c01", "size": 7754, "ext": "r", "lang": "R", "max_stars_repo_path": "functions.r", "max_stars_repo_name": "seabird-risk-assessment/abnj-seabird-bycatch-analysis", "max_stars_repo_head_hexsha": "fdb52ca62a9a3a518a60de9c33a7358e5cf43667", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "functions.r", "max_issues_repo_name": "seabird-risk-assessment/abnj-seabird-bycatch-analysis", "max_issues_repo_head_hexsha": "fdb52ca62a9a3a518a60de9c33a7358e5cf43667", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "functions.r", "max_forks_repo_name": "seabird-risk-assessment/abnj-seabird-bycatch-analysis", "max_forks_repo_head_hexsha": "fdb52ca62a9a3a518a60de9c33a7358e5cf43667", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 33.5670995671, "max_line_length": 136, "alphanum_fraction": 0.5488779985, "num_tokens": 2427, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7606506635289836, "lm_q2_score": 0.679178692681616, "lm_q1q2_score": 0.5166177232430188}} {"text": "pGumbel <- function (q, mu = 0, sigma = 1){\n stopifnot(sigma > 0)\n exp(-exp(-((q - mu)/sigma)))\n}\n\n\n#' Take a random draw from a variable variable\n#' @param obj Model object of class `lm`, `glm`, `polr` or `multinom`\n#' @param data Data frame giving variables used to estimate model\n#' @param ret For return a draw from the variable or return the probabilities.\n#' @param incl_b_var Logical indicating whether the sampling variable of\n#' the coefficients should be incorporated into the simulation.\n#' @param ... Other arguments to be passed down, currently not implemented.\n#'\n#' @importFrom MASS mvrnorm\n#' @importFrom stats model.matrix coef vcov rmultinom rbinom rnorm plogis family formula pcauchy pnorm\n#' @export\ndraw_val <- function(obj,\n data,\n ret=c(\"draw\", \"expected\"),\n incl_b_var = TRUE,\n ...){\n ret <- match.arg(ret)\n X <- model.matrix(formula(obj), data)\n if(inherits(obj, \"multinom\")){\n b <- c(t(coef(obj)))\n if(incl_b_var){\n B <- mvrnorm(1, b, vcov(obj))\n B <- cbind(0, matrix(B, ncol=(length(obj$lev)-1)))\n }else{\n B <- cbind(0, t(coef(obj)))\n }\n xb <- X %*% B\n prob <- prop.table(exp(xb), 1)\n tmp <- sapply(1:nrow(prob), function(i)rmultinom(1, 1, prob[i,]) )\n out <- factor(apply(tmp, 2, which.max),\n levels=1:length(obj$lev),\n labels=obj$lev)\n\n }\n if(inherits(obj, \"polr\")){\n cdf <- switch(obj$method,\n logistic = plogis,\n probit = pnorm,\n cloglog = pGumbel,\n cauchit = pcauchy)\n\n X <- X[,-1]\n b <- c(coef(obj), obj$zeta)\n if(incl_b_var){\n B <- mvrnorm(1, b, vcov(obj))\n Z <- B[grep(\"\\\\|\", names(B))]\n B <- B[-grep(\"\\\\|\", names(B))]\n }else{\n B <- coef(obj)\n Z <- obj$zeta\n }\n xb <- X %*% B\n Z <- c(-Inf, Z, Inf)\n qs <- sapply(Z, function(z)cdf(z-xb))\n prob <- sapply(2:ncol(qs), function(i)qs[,i] - qs[,(i-1)])\n tmp <- sapply(1:nrow(prob), function(i)rmultinom(1, 1, prob[i,]) )\n out <- factor(apply(tmp, 2, which.max),\n levels=1:length(obj$lev),\n labels=obj$lev)\n }\n if(inherits(obj, \"glm\")){\n if(family(obj)$family == \"binomial\"){\n if(incl_b_var){\n b <- mvrnorm(1, coef(obj),vcov(obj))\n }else{\n b <- coef(obj)\n }\n prob <- family(obj)$linkinv(X %*% b)\n out <- rbinom(length(prob), 1, prob)\n }else{\n stop(paste0(\"GLM family \", family(obj)$family, \" not currently supported.\\n\"))\n }\n }\n if(inherits(obj, \"lm\") & !inherits(obj, \"glm\")){\n if(incl_b_var){\n b <- mvrnorm(1, coef(obj),vcov(obj))\n }else{\n b <- coef(obj)\n }\n prob <- X %*% b\n out <- rnorm(nrow(X), prob, summary(obj)$sigma)\n }\n if(ret == \"draw\"){\n out\n }else{\n prob\n }\n}\n\n#' Find type of variable\n#'\n#' Find the variable type to identify the appropriate model for the variable\n#'\n#' @param x a vector of values to be evaluated\n#' @param ... other arguments to be passed down, currently not implemented.\n#'\n#' @importFrom stats na.omit\n#' @export\nfind_type <- function(x, ...){\n if(is.numeric(x) & length(unique(na.omit(x))) > 2){\n type <- \"lm\"\n }\n if(length(unique(na.omit(x))) == 2 & !is.character(x)){\n type <- \"glm\"\n }\n if(is.factor(x)){\n if(is.ordered(x)){\n type <- \"polr\"\n }else{\n type <- \"multinom\"\n }\n }\n type\n}\n\n\n#' Specify a Vote Path Analysis\n#'\n#' The vote path model estimates independent models that would otherwise use a block-recursive\n#' structure. Each model in each block is fit independently using the appropriate error distribution.\n#'\n#' @param blocks A list in which each element contains a vector of variable names for variables that will represent\n#' that block's effects. The list may be named or not.\n#' @param models If `NULL`, the algorithm will use `find_type()` to find the appropriate model. This assumes that variables\n#' that should be estimated using ordinal regression models are ordered factors. Otherwise, models can be a named vector of\n#' model types for some or all of the variables included in the model.\n#' @param data A data frame containing all of the variables in the `blocks` list.\n#' @param ... Other arguments to be passed down.\n#'\n#' @importFrom stats lm glm family binomial reformulate\n#' @importFrom MASS polr\n#' @importFrom nnet multinom\n#'\n#' @export\nvote_path <- function(blocks,\n models = NULL,\n data, ...){\n\n types <- sapply(c(unlist(blocks)), function(nm)find_type(data[[nm]]))\n allvars <- c(unlist(blocks))\n dv <- allvars[length(allvars)]\n if(!is.null(models)){\n types[match(names(models), names(types))] <- models\n }\n res <- vector(mode=\"list\", length=length(blocks)-1)\n for(i in 2:(length(blocks)-1)){\n mlist <- vector(mode=\"list\", length=length(blocks[[i]]))\n for(j in 1:length(blocks[[i]])){\n form <- reformulate(blocks[[(i-1)]], response=blocks[[i]][j])\n arglist <- list(formula = form, data=data)\n if(types[blocks[[i]][j]] %in% c(\"polr\", \"multinom\")){\n arglist$Hess <- TRUE\n }\n if(types[blocks[[i]][j]] == \"multinom\"){\n arglist$maxit <- 250\n }\n if(types[blocks[[i]][j]] == \"glm\"){\n arglist$family <- binomial\n }\n mlist[[j]] <- do.call(types[blocks[[i]][j]], arglist)\n }\n res[[(i-1)]] <- mlist\n }\n fullform <- reformulate(allvars[-length(allvars)],\n response=dv)\n arglist <- list(formula = fullform,\n data=data)\n if(types[dv] %in% c(\"polr\", \"multinom\")){\n arglist$Hess <- TRUE\n }\n if(types[dv] == \"multinom\"){\n arglist$maxit <- 250\n }\n if(types[dv] == \"glm\"){\n arglist$family <- binomial\n }\n res[[(length(blocks)-1)]] <- do.call(types[dv], arglist)\n out <- list(models = res, blocks = blocks)\n class(out) <- \"votepath\"\n out\n}\n\n\n#' Simulate Variable Effect\n#'\n#' Simulate a variable's effect in a vote path analysis.\n#'\n#' @param obj An object of class `votepath`\n#' @param data A data frame that contains all of the variables form the analysis.\n#' @param varname The name of a variable whose effect will be evaluated.\n#' @param diffchange The amount to change `varname`. Changes will be `x-.5*diffchange` and `x+.5*diffchange`.\n#' For categorical variables, the first and last categories will be chosen. The `diffchange` parameter\n#' defines the change so long as `vals=NULL`. If `vals` is not `NULL`, then those values will be used for everyone.\n#' @param vals A vector of length 2 giving the values that will be used to evaluate the effect size. This will override\n#' `diffchange`. The values must be of the same class as the variable being changed. For example, if the variable being\n#' changed is a factor, the `vals` vector also has to be a factor with the same levels as the variable in `varname`.\n#' @param b_var Logical indicating whether sampling variability on the coefficients should be incorporated in the simulation.\n#' @param R Number of simulations to be conducted.\n#' @param lastMod Should the prediction from the last model be a draw or a the expected value?\n#' @param ... Other arguments to be passed down.\n#'\n#' @importFrom progress progress_bar\n#' @importFrom stats sd\n#' @export\n#'\n#'\n#'\nsim_effect <- function(obj,\n data,\n varname,\n diffchange=c(\"unit\", \"sd\"),\n vals=NULL,\n b_var = TRUE,\n R=100,\n lastMod = c(\"expected\", \"draw\"),\n ...){\n mods <- obj$models\n blocks <- obj$blocks\n lastMod <- match.arg(lastMod)\n dv <- blocks[[length(blocks)]]\n out_i <- out_d <- out_t <- out_br <- NULL\n pb <- progress_bar$new(total = R)\n if(!is.null(vals) & length(vals) != 2)stop(\"vals must be a vector of length 2\\n\")\n if(!is.null(vals) & inherits(data[[varname]], \"factor\") & !inherits(vals, \"factor\"))stop(\"vals must have the same class as varname\\n\")\n which_block <- min(which(sapply(blocks, function(x)max(varname == x)) == 1))\n if(is.factor(data[[varname]]) & is.null(vals)){\n levs <- levels(data[[varname]])\n vals <- factor(c(1,length(levs)),\n levels=1:length(levs),\n labels=levs)\n }\n med <- vector(mode=\"list\", length=length((which_block+1):(length(blocks)-1)))\n brvars <- c(unlist(blocks[1:which_block]))\n br_form <- reformulate(brvars, response=dv)\n dv_type <- find_type(data[[dv]])\n br_args <- list(formula = br_form, data=data)\n if(dv_type %in% c(\"polr\", \"multinom\")){\n br_args$Hess <- TRUE\n }\n if(dv_type == \"multinom\"){\n br_args$maxit <- 250\n }\n if(dv_type == \"glm\"){\n br_args$family <- binomial\n }\n br_mod <- do.call(dv_type, br_args)\n\n\n for(r in 1:R){\n new_0 <- new_1 <- br_0 <- br_1 <- data\n if(is.null(vals)){\n delta <- ifelse(diffchange == \"sd\", sd(data[[varname]], na.rm=TRUE), 1)\n new_0[[varname]] <- br_0[[varname]] <- new_0[[varname]] - .5*delta\n new_1[[varname]] <- br_1[[varname]] <- new_1[[varname]] + .5*delta\n }else{\n new_0[[varname]] <- br_0[[varname]] <-vals[1]\n new_1[[varname]] <- br_1[[varname]] <-vals[2]\n }\n md0 <- md1 <- data\n k <- 1\n for(i in (which_block+1):(length(blocks)-1)){\n for(j in 1:length(blocks[[i]])){\n new_0[[blocks[[i]][j]]] <- md0[[blocks[[i]][j]]] <- draw_val(mods[[(i-1)]][[j]], new_0, incl_b_var = b_var)\n new_1[[blocks[[i]][j]]] <- md1[[blocks[[i]][j]]] <- draw_val(mods[[(i-1)]][[j]], new_1, incl_b_var = b_var)\n }\n tmp0 <- draw_val(mods[[length(mods)]], md0, ret=lastMod, incl_b_var = b_var)\n tmp1 <- draw_val(mods[[length(mods)]], md1, ret=lastMod, incl_b_var = b_var)\n if(!is.matrix(tmp0))tmp0 <- matrix(tmp0, ncol=1)\n if(!is.matrix(tmp1))tmp1 <- matrix(tmp1, ncol=1)\n med[[k]] <- rbind(med[[k]], colMeans(tmp1-tmp0))\n k <- k+1\n }\n\n res0 <- draw_val(mods[[length(mods)]], new_0, ret=lastMod, incl_b_var = b_var)\n res1 <- draw_val(mods[[length(mods)]], new_1, ret=lastMod, incl_b_var = b_var)\n if(!is.matrix(res0))res0 <- matrix(res0, ncol=1)\n if(!is.matrix(res1))res1 <- matrix(res1, ncol=1)\n total_effect <- res1-res0\n\n d0 <- draw_val(mods[[length(mods)]], br_0, ret=lastMod, incl_b_var = b_var)\n d1 <- draw_val(mods[[length(mods)]], br_1, ret=lastMod, incl_b_var = b_var)\n if(!is.matrix(d0))d0 <- matrix(d0, ncol=1)\n if(!is.matrix(d1))d1 <- matrix(d1, ncol=1)\n\n direct_effect <- d1-d0\n\n indirect_effect <- total_effect - direct_effect\n\n be0 <- draw_val(br_mod, br_0, ret=lastMod, incl_b_var = b_var)\n be1 <- draw_val(br_mod, br_1, ret=lastMod, incl_b_var = b_var)\n if(!is.matrix(be0))be0 <- matrix(be0, ncol=1)\n if(!is.matrix(be1))be1 <- matrix(be1, ncol=1)\n\n out_br <- rbind(out_br, colMeans(be1-be0))\n\n out_t <- rbind(out_t, colMeans(total_effect))\n out_d <- rbind(out_d, colMeans(direct_effect))\n out_i <- rbind(out_i, colMeans(indirect_effect))\n pb$tick()\n }\n res <- list(total = out_t, direct= out_d, indirect=out_i, br = out_br, mediated=med)\n}\n\n", "meta": {"hexsha": "90dbd1df05711b34827fdfb6977cc74e8c709173", "size": 11049, "ext": "r", "lang": "R", "max_stars_repo_path": "R/votepath.r", "max_stars_repo_name": "davidaarmstrong/votepath", "max_stars_repo_head_hexsha": "2e22cb96670ef38bf07a0579073d69421859bb14", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/votepath.r", "max_issues_repo_name": "davidaarmstrong/votepath", "max_issues_repo_head_hexsha": "2e22cb96670ef38bf07a0579073d69421859bb14", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/votepath.r", "max_forks_repo_name": "davidaarmstrong/votepath", "max_forks_repo_head_hexsha": "2e22cb96670ef38bf07a0579073d69421859bb14", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.6419354839, "max_line_length": 136, "alphanum_fraction": 0.6024074577, "num_tokens": 3293, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8397339596505965, "lm_q2_score": 0.6150878696277513, "lm_q1q2_score": 0.5165101722955615}} {"text": "######################### SCRIPT 0: SUPPORTING FUNCTIONS #########################\n\nrunModel = function(indexu,indexk,NITERATION)\n{\n dataset=list(\n N=dim(data0)[1],\n Kk=length(indexk),\n nk=data0[,indexk],\n Ku=length(indexu),\n nu=data0[,indexu],\n Sk=knownpop$real_data[indexk],\n Su=rep(NA,length(indexu)))\n \n initialisation=list(lambda=0.1)\n jagmod=jags.model(textConnection(model1),data=dataset,inits=initialisation,n.chains=2)\n update(jagmod, n.iter=5000, progress.bar=\"text\")\n posterior = coda.samples(jagmod, c(\"alpha\",\"lambda\",\"tau\",\"Su\"),n.iter=NITERATION,progress.bar=\"text\",thin=10)\n dicsamples = dic.samples(jagmod,type = \"pD\",n.iter=20000,thin=10)\n results = list(indexk=indexk,indexu=indexu,dataset = dataset,posterior=posterior,dicsamples=dicsamples)\n return(results)\n}\n\n\nrunModel_te = function(indexu,indexk,NITERATION,x)\n{\n dataset=list(\n N=dim(data0)[1],\n Kk=length(indexk),\n nk=data0[,indexk],\n Ku=length(indexu),\n nu=data0[,indexu],\n Sk=knownpop$real_data[indexk],\n Su=rep(NA,length(indexu)),\n x=x)\n \n initialisation=list(lambda=0.1)\n jagmod=jags.model(textConnection(model2),data=dataset,inits=initialisation,n.chains=2)\n update(jagmod, n.iter=5000, progress.bar=\"text\")\n posterior = coda.samples(jagmod, c(\"alpha\",\"lambda\",'beta','sigma',\"tau\",\"Su\"),n.iter=NITERATION,progress.bar=\"text\",thin=10)\n dicsamples = dic.samples(jagmod,type = \"pD\",n.iter=20000,thin=10)\n \n results = list(indexk=indexk,indexu=indexu,dataset = dataset,posterior=posterior,dicsamples=dicsamples)\n \n return(results)\n}\n\n\nrunModel_te_be = function(indexu,indexk,NITERATION,x)\n{\n dataset=list(\n N=dim(data0)[1],\n Kk=length(indexk),\n nk=data0[,indexk],\n Ku=length(indexu),\n nu=data0[,indexu],\n Sk=knownpop$real_data[indexk],\n Su=rep(NA,length(indexu)),\n x=x,\n age=demo$age-40.5,\n sex=demo$sex-0.49,\n malay=demo$malay-0.15,\n indian=demo$indian-0.07)\n \n initialisation=list(lambda=0.1)\n jagmod=jags.model(textConnection(model3),data=dataset,inits=initialisation,n.chains=2)\n update(jagmod, n.iter=5000, progress.bar=\"text\")\n posterior = coda.samples(jagmod, c(\"alpha\",\"lambda\",'beta','sigma','sigmab',\"tau\",\"Su\",\"b1u\",\"b2u\",\"b3u\",'b4u',\"b1k\",\"b2k\",'b3k','b4k'),n.iter=NITERATION,progress.bar=\"text\",thin=10)\n dicsamples = dic.samples(jagmod,type = \"pD\",n.iter=20000,thin=10)\n \n results = list(indexk=indexk,indexu=indexu,dataset = dataset,posterior=posterior,dicsamples=dicsamples)\n return(results)\n}\n\n\n### Models\n\nmodel1 = 'model {\n\nfor(i in 1:N)\n\n{\n \n for(k in 1:Ku)\n \n {\n \n nu[i,k] ~ dpois(lambda*alpha[i]*Su[k])\n \n }\n \n for(k in 1:Kk)\n \n {\n \n nk[i,k] ~ dpois(lambda*alpha[i]*Sk[k])\n \n }\n \n alpha[i]~dlnorm(0,tau)\n \n}\n\nfor(k in 1:Ku)\n\n{\n \n Su[k]~dunif(0,2500000)\n \n}\n\nfor(k in 1:Kk)\n\n{\n \n Sk[k]~dunif(0,2500000)\n \n}\n\nlambda ~ dunif(0,10)\n\ntau ~ dunif(0,10)\n\n}\n'\n\nmodel2 ='model {\n\nfor(i in 1:N)\n\n{\n \n for(k in 1:Ku)\n \n {\n \n nu[i,k] ~ dpois(lambda*alpha[i]*exp(beta[k]*x[i,k])*Su[k])\n \n }\n \n for(k in 1:Kk)\n \n {\n \n nk[i,k] ~ dpois(lambda*alpha[i]*Sk[k])\n \n }\n \n alpha[i]~dlnorm(0,tau)\n \n}\n\nfor(k in 1:Ku)\n\n{\n \n Su[k]~dunif(0,2500000)\n \n}\n\nfor(k in 1:Kk)\n\n{\n \n Sk[k]~dunif(0,2500000)\n \n}\n\nfor(k in 1:Ku)\n\n{\n \n beta[k]~dnorm(0,(1/sigma)^2)\n \n}\n\nlambda ~ dunif(0,10)\n\ntau ~ dunif(0,10)\n\nsigma ~ dgamma(1,0.01)\n}\n'\n\nmodel3 = 'model {\n \nfor(i in 1:N)\n\n{\n \n for(k in 1:Ku)\n \n {\n \n nu[i,k] ~ dpois(lambda*alpha[i]*exp(beta[k]*x[i,k])*exp(b1u[k]*age[i])*exp(b2u[k]*sex[i])*exp(b3u[k]*malay[i])*exp(b4u[k]*indian[i])*Su[k])\n \n }\n \n for(k in 1:Kk)\n \n {\n \n nk[i,k] ~ dpois(lambda*alpha[i]*exp(b1k[k]*age[i])*exp(b2k[k]*sex[i])*exp(b3k[k]*malay[i])*exp(b4k[k]*indian[i])*Sk[k])\n \n }\n \n alpha[i]~dlnorm(0,tau)\n \n}\n\nfor(k in 1:Ku)\n\n{\n \n Su[k]~dunif(0,2500000)\n \n}\n\nfor(k in 1:Kk)\n\n{\n \n Sk[k]~dunif(0,2500000)\n b1k[k]~dnorm(0,(1/sigmab)^2)\n b2k[k]~dnorm(0,(1/sigmab)^2)\n b3k[k]~dnorm(0,(1/sigmab)^2)\n b4k[k]~dnorm(0,(1/sigmab)^2)\n}\n\nfor(k in 1:Ku)\n\n{\n \n beta[k]~dnorm(0,(1/sigma)^2)\n b1u[k]~dnorm(0,(1/sigmab)^2)\n b2u[k]~dnorm(0,(1/sigmab)^2)\n b3u[k]~dnorm(0,(1/sigmab)^2)\n b4u[k]~dnorm(0,(1/sigmab)^2)\n}\n\nlambda ~ dunif(0,10)\n\ntau ~ dunif(0,10)\n\nsigma ~ dgamma(1,0.01)\nsigmab ~ dgamma(1,0.01)\n}\n'", "meta": {"hexsha": "858cc83feefda2abcb824f7772df270e080e8edf", "size": 4269, "ext": "r", "lang": "R", "max_stars_repo_path": "codes/supporting_functions.r", "max_stars_repo_name": "kieshaprem/nsum", "max_stars_repo_head_hexsha": "c31bf7f0c779b5557faa162e6a5c8b609d1fe83d", "max_stars_repo_licenses": ["CC0-1.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "codes/supporting_functions.r", "max_issues_repo_name": "kieshaprem/nsum", "max_issues_repo_head_hexsha": "c31bf7f0c779b5557faa162e6a5c8b609d1fe83d", "max_issues_repo_licenses": ["CC0-1.0"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2020-03-14T18:53:53.000Z", "max_issues_repo_issues_event_max_datetime": "2020-03-14T18:53:53.000Z", "max_forks_repo_path": "codes/supporting_functions.r", "max_forks_repo_name": "kieshaprem/nsum", "max_forks_repo_head_hexsha": "c31bf7f0c779b5557faa162e6a5c8b609d1fe83d", "max_forks_repo_licenses": ["CC0-1.0"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-05-15T07:13:49.000Z", "max_forks_repo_forks_event_max_datetime": "2020-05-15T07:13:49.000Z", "avg_line_length": 17.2834008097, "max_line_length": 184, "alphanum_fraction": 0.6104474116, "num_tokens": 1653, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8289388083214156, "lm_q2_score": 0.6224593312018546, "lm_q1q2_score": 0.5159806962350107}} {"text": "#\n# Secure Digitalisation Monte Carlo Simulation Script\n#\n# This script runs a Monto Carlo simulation using breach likelihood and cost\n# figures derived from the Cyber Security Breaches Survey 2020 (CSBS).\n#\tThis script is an unfinished prototype, and has since been superseded by\n# `montecarlo.py`.\n#\n# Acknowledgements: Dr Dan Prince & Dr Chris Sherlock\n#\n\nmasses = c(0.54, 0.1058, 0.1012, 0.0966, 0.069, 0.0368, 0.0414)\nboundaries = c(1, 2, 8, 18, 80, 400, 8000)\n\nFs = cumsum(masses)\nplot(log(boundaries), log(1 - Fs))\n\nxs = log(boundaries)\nys = log(1 - Fs)\nfit = lm(ys ~ xs)\nsummary(fit)\n\nalogb = fit$coeff[1]\na = -fit$coeff[2]\nb = exp(alogb/a)\nprint(a)\nprint(b)\n\nn = 10000\n\nus = runif(n)\nxs = b / (1 - us)^(1 / a)\nprint()\np0 = mean(xs < boundaries[1])\np1 = mean(xs < boundaries[2]) - p0\np2 = mean(xs < boundaries[3]) - p0 - p1\np3 = mean(xs < boundaries[4]) - p0 - p1 - p2\np4 = mean(xs < boundaries[5]) - p0 - p1 - p2 - p3\np5 = mean(xs < boundaries[6]) - p0 - p1 - p2 - p3 - p4\nps = c(p0, p1, p2, p3, p4, p5, 1 - (p0 + p1 + p2 + p3 + p4 + p5))\n\nprint(ps)\nprint(masses)\n\nnattacks = floor(xs)\nhist(log10(nattacks),\n main = \"Histogram of Number of Attacks/Breaches Over 12 Months\",\n xlab = expression(\"Number of Attacks (log\"[10]*\")\"),\n ylab = \"Frequency\",\n breaks = 0:12)\n\n# Plots the distribution for the average cost of breach(es) over 12 months\n\nmean = 3230\nmedian = 274\n\nlogstd = sqrt(2 * (log(mean) - if (median == 0) 0 else log(median)))\nstd = exp(1)^logstd\n\ncurve(dlnorm(x, log(mean), log(std)), from=1, to=5000,\n main = \"Average annual breach cost distribution\",\n xlab = 'Cost (£)',\n ylab = 'Density',\n lwd = 2)\n\n# Runs the MonteCarlo simulation\n\nsimulateCosts <- function(n) {\n return(if (n >= 1) sum(rlnorm(n, loc, shape)) else 0)\n}\n\nn = 10000\n\nloc <- log(mean^2 / sqrt(std^2 + mean^2))\nshape <- sqrt(log(1 + (std^2 / mean^2)))\n\nnumAttacks <- sample(log10(nattacks), n)\nresults <- sapply(numAttacks, simulateCosts)\n\nhist(results,\n main = \"Histogram of Total Costs Over 12 Months (Monte Carlo sim)\",\n xlab = \"Total cost (£)\")\n\nd <- density(results)\nplot(d,\n main=\"Density of Total Costs Over 12 Months (Monte Carlo sim)\", \n xlab=expression(\"Total Cost (£)\"),\n ylab=\"Density\")\n\n# Get loss exceedance\n# TODO: needs to be prettier, but `evaluate::loss_exceedance_curve()` is broken\n\nmaxValue = 2500\nnumOver <- length(results[results > maxValue])\nrisk = numOver/n\n\nplot(d,\n main=\"Loss Exceedance (Monte Carlo sim)\", \n xlab=expression(\"Total Cost (£)\"),\n ylab=\"Density\")\n\nabline(v = maxValue, col=\"red\", lwd=3, lty=2)\ntext(3000, 4e-04, labels=paste(floor(risk*100), \"% chance of ≥£\", maxValue, \" losses\"), adj=c(0, 0.5))\n", "meta": {"hexsha": "e60e154595925bbd4ffe1561dc24234827c05cb5", "size": 2696, "ext": "r", "lang": "R", "max_stars_repo_path": "src/montecarlo.r", "max_stars_repo_name": "Rumperuu/Threat-Intelligence-Service", "max_stars_repo_head_hexsha": "c72e312c9b2ad7acc0f3b564f735944b437c298b", "max_stars_repo_licenses": ["CNRI-Python"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/montecarlo.r", "max_issues_repo_name": "Rumperuu/Threat-Intelligence-Service", "max_issues_repo_head_hexsha": "c72e312c9b2ad7acc0f3b564f735944b437c298b", "max_issues_repo_licenses": ["CNRI-Python"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/montecarlo.r", "max_forks_repo_name": "Rumperuu/Threat-Intelligence-Service", "max_forks_repo_head_hexsha": "c72e312c9b2ad7acc0f3b564f735944b437c298b", "max_forks_repo_licenses": ["CNRI-Python"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 25.9230769231, "max_line_length": 102, "alphanum_fraction": 0.640578635, "num_tokens": 915, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8539127678225575, "lm_q2_score": 0.6039318337259583, "lm_q1q2_score": 0.5157051037130856}} {"text": "##### Chapter 8: Association Rules -------------------\n# 장바구니 분석의 결과는 연관 규칙의 모음이다.\n # 연관 규칙 : 아이템 집합 사이에 관계에 존재하는 패턴\n # LHS : 규칙을 실행하기 위해 만족돼야 하는 조건\n # RHS : 그 조건을 만족했을 때 기대하는 결과\n# 연관 규칙은 예측이 아니라, 대규모 DB에서 자율적인 지식의 발견을 위해 사용된다.\n# 연관 구칙 학습자는 자율적이기 때문에, 알고리즘이 훈련될 필요가 없다.\n # 즉 데이터가 사전에 레이블될 필요가 없다.\n # 하지만 학습자에 대해 사용자가 직접 확인하는 것 외에는 규칙 학습자의 성능을 측정할 방법이 없다.\n\n# 연관 규칙 학습을 위한 a priori 알고리즘 (아 프리오리, 선험적인)\n # 쇼핑몰에 물건 100개만 있어도 2^100개의 아이템 집합을 평가해야 함\n # 좀 더 똑똑한 학습 알고리즘은 아이템 집합을 전부 평가하지 않는다.\n # 대신 잠재적인 아이템 조합이 실제로 드물게 발견된다는 사실을 이용한다.\n # 아프리오리 알고리즘은 빈번한 아이템집합의 속성에 대해 단순한 사전적 믿음을 이용한다.리\n # 통계 척도인, 아이템 집합의 흥미도를 이용한다.\n # 이는 어떻게 측정할까? 또 학습될 규칙의 수를 줄이기 위해 아프리오리 속성과 어떻게 결합되는가?\n# 장단점\n # 장점\n # 대규모 거래 데이터에 대해 작업 가능\n # 이해하기 쉬운 규칙 생성\n # 데이터 마이닝과 DB에서 예상치 못한 지식 발굴 가능\n # 단점\n # 작은 데이터셋에서는 유용하지 않음\n # 진정한 통찰과 상식을 분리하기 위한 노력이 필요\n # 랜덤 패턴에서 비논리적인 결론을 도출하기 쉽다.\n# 규칙 흥미 측정 : 지지도와 신뢰도\n # 지지도와 신뢰도를 사용해, 보고되는 규칙의 갯수를 제한할 수 있다.\n # 확실하거나 상식적인 규칙만 식별될 떄까지 제한하는 것도 가능하다.\n # support(X) = count(X) / N <- (아이템 집합 X 를 포함하는 거래 건수 / 데이터베이스의 거래 건수)\n # confidence(X -> Y) = support(X,Y)/support(X) (X와 Y를 모두 포함하는 아이템집합의 지지도/X만포함지지도)\n # 신뢰도는 아이템 또는 아이템집합 X의 존재가, 아이템 또는 아이템집합 Y의 존재를 유발하는 거래의 비율\n # confidence(X -> Y) != confidence(Y -> X)\n# a priori 원칙을 이용한 규칙 집합의 구축\n # 빈번한 아이템집합의 모든 부분집합 또한 빈번해야 한다.\n # a priori algorithm은 이 논리를 사용해, 실제 평가하기 전에 잠재적인 연관 규칙을 제외시킨다.\n # 규칙을 생성하는 절차\n # 1. 최소 지지도 임계치를 만족하는 모든 아이템 집합을 식별한다.\n # 2. 이 아이템 집합에서 최소 신리도 임계치를 만족하는 아이템 집합으로 규칙을 생성한다.\n\n\n\n## Example: Identifying Frequently-Purchased Groceries ----\n## Step 2: Exploring and preparing the data ----\n\n# load the grocery data into a sparse matrix\n# 거래 데이터는 좀 더 자유로운 형태로 저장되어 있다.\n # 각 레코드는 고정된 개수의 특징 대신, 한 개부터 여러 개까지 쉼표로 분리된 임의의 개수의 아이템목록으로 이루어짐.\ninstall.packages(\"arules\")\nlibrary(arules)\ngetwd()\nsetwd(\"D:/R_LAB/MlwR/Chapter 08\")\n# 그냥 read.csv로 읽으면 나중에 문제가 생긴다\n # 첫 번째 줄이 네 개의 값을 가졌기 때문에 R은 생성해야 할 네 변수를 생성했다. -> 더 많이 사면 잘린다\n # 거래를 특정 아이템으로 채우는 위치 집합이 아닌,\n # 각 특정 아이템을 포함하거나 포함하지 않는 장바구니로 취급하는 데이터셋이 필요하다.\n# 거래 데이터를 위한 희소 행렬(sparse matrix)생성\n # 희소 행렬은 의미있는 셀에 비해 0이 너무 많기 떄문에, 전체 행렬을 통째로 저장하는 것은 비효율적이다.(170만개)\n # 희소 행렬 구조를 만들기 위해 arules 패키지를 사용하자.\ngroceries <- read.transactions(\"groceries.csv\", sep = \",\")\nsummary(groceries)# 0이 아닌 값 2.6%\n# look at the first five transactions\n # 희소 행렬의 내용을 보려면 inspect 함수를 벡터 연산과 결합해 사용.\ninspect(groceries[1:5])\n\n# examine the frequency of items\nitemFrequency(groceries[, 1:3])\n\n# plot the frequency of items(아이템 빈도 그래프)\nitemFrequencyPlot(groceries, support = 0.1) # 최소 10% 지지도 갖는 8개 아이템\nitemFrequencyPlot(groceries, topN = 20) # 지지도 가장 높은 20개\n\n# a visualization of the sparse matrix for the first five transactions\n # 5행 169열의 행렬. 5건의 거래와 169개의 가능한 아이템을 나타낸다.\nimage(groceries[1:5])\n\n# visualization of a random sample of 100 transactions\nimage(sample(groceries, 100)) # 거래 100건에 대한 희소 행렬 시각화\n\n## Step 3: Training a model on the data ----\nlibrary(arules)\n\n# default settings result in zero rules learned\n# 실행하는 것 자체는 간단하지만, 합리적인 개수와 연관 규칙을 생성하는 지지도, 신뢰도 찾기는 시행착오 필요\n # 파라미터의 수준이 너무 높은 경우\n # 규칙을 찾지 못하거나 너무 포괄적인 규칙만을 찾음(숟가락 옆에 젓가락 놓아봐야 다 안다)\n # 파라미터의 수준이 너무 낮은 경우\n # 규칙이 너무 많아져서 통제 힘듦, 혹은 메모리 부족\narules::apriori(groceries, parameter = # minlen은 요구되는 최소 규칙 아이템\n list(support = 0.1, confidence = 0.8, minlen = 1))우 # default setting -> nothing learned\n\n# set better support and confidence levels to learn more rules\n# 최소 지지도 임계치를 다루는 방법\n # ex)한달 동안 하루 2번 구입되면 60회, 9835회 중 60회는 0.006\n# 최소 신뢰도 임계치를 다루는 방법(일단 0.25로 시작)\n # 너무 낮으면 규칙 너무 많아짐, 너무 높으면 뻔한 규칙\ngroceryrules <- apriori(groceries, parameter = list(support =\n 0.006, confidence = 0.25, minlen = 2))\ngroceryrules # 규칙 463\n\n## Step 4: Evaluating model performance ----\n# summary of grocery association rules\nsummary(groceryrules)\n# 대부분 또는 모든 규칙이 최소 임계치 바로 근처에서 지지도와 신뢰도를 갖는다면\n # 기준을 너무 높게 설정했다는 것을 의미 -> 재조정 필요\n # lift(X -> Y) = confidence(X ->Y)/support(Y)\n # 향상도는 어떤 아이템(집합) X가 구매됐다는 것을 안다면, 다른 아이템(집합)Y가 어떤 확률로 구매될 것인가를\n # Y의 일반적인 구매 확률과 비교해 측정한다.\n # 아이템 순서가 중요한 신뢰도와는 달리, 향상도는 순서가 달라도 동일하다.\n\n# look at the first three rules\ninspect(groceryrules[1:3])\n# 논리적으로 potted plant -> whole milk는 말이 안되지만, 데이터는 이 규칙이 믿을만 하다고 말하고 있다.\n# 일반적인 방법은, 연관 규칙을 받아 다음 세 가지 범주로 규칙을 나누는 것이다.\n # 실행 가능한\n # 사소한(기저귀 -> 분유)\n # 설명하기 어려운(연관성 불명확. 단순히 데이터에 있는 랜덤 패턴일 수도 있다.)\n# 충분한 시간이 있다면 숨겨진 보석 규칙을 위해 모든 규칙을 평가할 수 있다.\n # 하지만 분석가는 규칙이 어떤지까지는 잘 모를 수도 있다.\n # 가장 흥미로운 결과가 상단에 표시되도록, 학습된 규칙을 정렬하고 공유하자.\n\n\n## Step 5: Improving model performance ----\n\n# sorting grocery rules by lift\n # sort()함수로 by 파라미터에 lift를 넣어서, lift가 가장 높은 순서대로 5개를 출력시켰다. \ninspect(sort(groceryrules, by = \"lift\")[1:5])\n\n# finding subsets of rules containing any berry items\n# subset()함수는 거래, 아이템, 규칙의 부분집합을 찾는 방법을 제공한다.\nberryrules <- subset(groceryrules, items %in% \"berries\") # 베리가 들어간 규칙만 부분집합으로 뽑음\ninspect(berryrules) # 규칙 출력. lift 높은 두 규칙은 실행 가능해 보인다.\n# subset() 사용법\n # items %in% 쓰면 좌측, 우측 모두 포함되기만 하면 뽑는다.\n # 좌측에만 해당되게 하려면 item 대신 lhs, 우측은 rhs\n # 연산자 %in%은 아이템 중 최소 하나가 정의된 목록에서 발견돼야만 한다.\n # 두 개 하고 싶으면 %in% c(\"berries\", \"yogurt\")로 넣자.\n # %ain%으로 완전 매칭(베리, 요거트 둘 다 있는 규칙만 찾기)\n # %pin%으로 부분 매칭(%pin% fruit로 그냥 과일이랑 열대 과일 같이 찾기)\n # 부분집합은 support, confidence, lift로 제약 가능\n # 매칭 조건은 %, |, !같은 표준 R 논리 연산자와 결합 가능\n\n\n# writing the rules to a CSV file\nwrite(groceryrules, file = \"groceryrules.csv\",\n sep = \",\", quote = TRUE, row.names = FALSE)\n\n# converting the rule set to a data frame\ngroceryrules_df <- as(groceryrules, \"data.frame\")\nstr(groceryrules_df)\n", "meta": {"hexsha": "364bb6194b80d6088b2252a92a95412e869bd476", "size": 5496, "ext": "r", "lang": "R", "max_stars_repo_path": "MLwR/Chapter 08/MLwR_v2_08(Association Rules).r", "max_stars_repo_name": "BeginnerJay/R_STUDYING", "max_stars_repo_head_hexsha": "4dcca5a781cf550c8c40edc636c247357c95a296", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "MLwR/Chapter 08/MLwR_v2_08(Association Rules).r", "max_issues_repo_name": "BeginnerJay/R_STUDYING", "max_issues_repo_head_hexsha": "4dcca5a781cf550c8c40edc636c247357c95a296", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "MLwR/Chapter 08/MLwR_v2_08(Association Rules).r", "max_forks_repo_name": "BeginnerJay/R_STUDYING", "max_forks_repo_head_hexsha": "4dcca5a781cf550c8c40edc636c247357c95a296", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.3973509934, "max_line_length": 106, "alphanum_fraction": 0.66430131, "num_tokens": 3571, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.904650527388829, "lm_q2_score": 0.5698526514141571, "lm_q1q2_score": 0.5155175016357397}} {"text": "\\name{generate_simulated_data_from_estimated_model}\n\\alias{generate_simulated_data_from_estimated_model}\n%- Also NEED an '\\alias' for EACH other topic documented here.\n\\title{\n Generating simulated data from a fitted model \n}\n\\description{\n This function generates simulated networks from a fitted model and performs estimations on these simulated networks with the same setting used in the original estimation. Each simulated network is generated using parameters of the fitted model, while keeping other aspects of the growth process as faithfully as possible to the original observed network.\n}\n\\usage{\ngenerate_simulated_data_from_estimated_model(net_object, net_stat, result, M = 5)\n}\n%- maybe also 'usage' for other objects documented here.\n\\arguments{\n \\item{net_object}{\n an object of class \\code{PAFit_net} that contains the original network.\n }\n \\item{net_stat}{\n An object of class \\code{PAFit_data} which contains summarized statistics of the original network. This object is created by the function \\code{\\link{get_statistics}}. \n }\n \\item{result}{\n An object of class \\code{Full_PAFit_result} which contains the fitted model obtained by applying the function \\code{\\link{joint_estimate}}. \n }\n\\item{M}{integer. The number of simulated networks. Default value is \\code{5}.} \n}\n\n\\value{\n Outputs a \\code{Simulated_Data_From_Fitted_Model} object, which is a list containing the following fields:\n \\itemize{\n \\item \\code{graph_list}: a list containing \\code{M} simulated graphs.\n \n \\item \\code{stats_list}: a list containing \\code{M} objects of class \\code{PAFit_data}, which are the results of applying \\code{\\link{get_statistics}} on the simulated graphs.\n \\item \\code{result_list}: a list containing \\code{M} objects of class \\code{Full_PAFit_result}, which are the results of applying \\code{\\link{joint_estimate}} on the simulated graphs.\n}\n}\n\n\\author{\n Thong Pham \\email{thongphamthe@gmail.com}\n}\n\\references{\n 1. Pham, T., Sheridan, P. & Shimodaira, H. (2015). PAFit: A Statistical Method for Measuring Preferential Attachment in Temporal Complex Networks. PLoS ONE 10(9): e0137796. (\\doi{10.1371/journal.pone.0137796}).\n \n 2. Pham, T., Sheridan, P. & Shimodaira, H. (2016). Joint Estimation of Preferential Attachment and Node Fitness in Growing Complex Networks. Scientific Reports 6, Article number: 32558. (\\doi{10.1038/srep32558}).\n \n 3. Pham, T., Sheridan, P. & Shimodaira, H. (2020). PAFit: An R Package for the Non-Parametric Estimation of Preferential Attachment and Node Fitness in Temporal Complex Networks. Journal of Statistical Software 92 (3). (\\doi{10.18637/jss.v092.i03}).\n\n4. Inoue, M., Pham, T. & Shimodaira, H. (2020). Joint Estimation of Non-parametric Transitivity and Preferential Attachment Functions in Scientific Co-authorship Networks. Journal of Informetrics 14(3). (\\doi{10.1016/j.joi.2020.101042}).\n}\n\\seealso{\n \\code{\\link{get_statistics}}, \\code{\\link{joint_estimate}}, \\code{\\link{plot_contribution}}\n}\n\n\\examples{\n\\dontrun{\n \n library(\"PAFit\")\n net_object <- generate_net(N = 500, m = 10, s = 10, alpha = 0.5)\n net_stat <- get_statistics(net_object) \n result <- joint_estimate(net_object, net_stat)\n simulated_data <- generate_simulated_data_from_estimated_model(net_object, net_stat, result)\n plot_contribution(simulated_data, result, which_plot = \"PA\")\n plot_contribution(simulated_data, result, which_plot = \"fit\")\n }\n}\n\n\\concept{preferential attachment}\n\\concept{attachment function}\n\\concept{fitness}\n\n", "meta": {"hexsha": "4af6c489e81a604554826751af371a0073e47bea", "size": 3522, "ext": "rd", "lang": "R", "max_stars_repo_path": "man/generate_simulated_data_from_estimated_model.rd", "max_stars_repo_name": "thongphamthe/PAFit", "max_stars_repo_head_hexsha": "e1c3b32e0f886eb0b90a8bcbed0c3719bd28ef44", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": 14, "max_stars_repo_stars_event_min_datetime": "2017-08-22T14:24:33.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-19T18:38:48.000Z", "max_issues_repo_path": "man/generate_simulated_data_from_estimated_model.rd", "max_issues_repo_name": "thongphamthe/PAFit", "max_issues_repo_head_hexsha": "e1c3b32e0f886eb0b90a8bcbed0c3719bd28ef44", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2017-02-22T16:33:00.000Z", "max_issues_repo_issues_event_max_datetime": "2021-08-25T01:01:52.000Z", "max_forks_repo_path": "man/generate_simulated_data_from_estimated_model.rd", "max_forks_repo_name": "thongphamthe/PAFit", "max_forks_repo_head_hexsha": "e1c3b32e0f886eb0b90a8bcbed0c3719bd28ef44", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2017-05-12T04:40:53.000Z", "max_forks_repo_forks_event_max_datetime": "2021-05-05T12:09:33.000Z", "avg_line_length": 50.3142857143, "max_line_length": 356, "alphanum_fraction": 0.7555366269, "num_tokens": 922, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.6992544085240401, "lm_q2_score": 0.7371581568543043, "lm_q1q2_score": 0.5154610909598282}} {"text": "## Function to do Wright-Fisher simulation with selection and a finite sample size\n# using priors on Ne, s, and initial allele frequency\n# Sample at two time points\n\n# i is a dummy argument so that it can be used with sapply and parSapply\n# f1min: the minimum for the initial allele frequency uniform prior\n# f1max:\n# smin: the minimum for the selection coefficient uniform prior\n# smax:\n# c1: the size of the first sample (in # chromosomes)\n# c2: the size of the second sample (in # chromosomes)\n# gen: the number of generations\n# ne: a vector of possible Ne values. Simulation picks one to use.\n# h: dominance effect\n\nwfs <- function(i, f1min=0, f1max=1, smin=-1, smax=1, c1=58, c2=48, gen=20, ne=100, h=0.5){ \n\t# choose parameters for this simulation\n\tf1 <- runif(1, min=f1min, max=f1max) # starting allele frequency\n\ts <- runif(1, min=smin, max=smax) # selection coefficient\n\tif(length(ne)>1) thisne <- round(sample(ne, 1)) # ne has to be integer for binomial sampling\n\tif(length(ne)==1) thisne <- round(ne)\n\t\n#\tprint(paste(f1, s, thisne))\n\t\n\tp <- f1 # current allele frequency\n\twaa <- 1+s # relative fitness of genotype AA\n\twab <- 1+s*h\n\twbb <- 1\n\tfor(i in 1:gen){\n\t\tx <- (waa*p^2 + wab*p*(1-p))/(waa*p^2 + wab*2*p*(1-p) + wbb*(1-p)^2) # probability of sampling allele A, given selection\n\t\tp <- rbinom(1,thisne,x)/thisne\n#\t\tprint(paste(x,p))\n\t}\n\tf2 <- p\n\n\tf1samp <- rbinom(1,c1,f1)/c1 # first sample allele frequency\n\tf2samp <- rbinom(1,c2,f2)/c2 # second sample allele frequency\n\n\t\n\t# return values\n\tout <- c(ne=thisne, f1=f1, s=s, gen=gen, f2=f2, f1samp=f1samp, f2samp=f2samp)\n\treturn(out)\n}", "meta": {"hexsha": "f655605217f4506ff8ff69f2adf7c071daae0102", "size": 1598, "ext": "r", "lang": "R", "max_stars_repo_path": "scripts/wfs.r", "max_stars_repo_name": "pinskylab/codEvol", "max_stars_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "scripts/wfs.r", "max_issues_repo_name": "pinskylab/codEvol", "max_issues_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 5, "max_issues_repo_issues_event_min_datetime": "2020-04-11T11:14:18.000Z", "max_issues_repo_issues_event_max_datetime": "2021-02-21T19:57:31.000Z", "max_forks_repo_path": "scripts/wfs.r", "max_forks_repo_name": "pinskylab/codEvol", "max_forks_repo_head_hexsha": "2710c4e4d9223ce02873938fd7aa7fc80f7dd8ac", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 37.1627906977, "max_line_length": 122, "alphanum_fraction": 0.6902377972, "num_tokens": 527, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.837619947119304, "lm_q2_score": 0.615087848460224, "lm_q1q2_score": 0.5152098511009793}} {"text": "#Analyses for Tutorial 05: Discrete PCMs\r\n\r\n## 1: Read data, tree, and prune/match one to the other\r\nlibrary(geiger) #also loads ape\r\nlibrary(phytools)\r\n\r\n#Here is a large time-dated molecular phylogeny (a chronogram):\r\ntree<-read.tree(\"TutorialData/tree.64.tre\",tree.names=T)\r\nplot(tree, show.tip.label = FALSE)\r\nmydata<-read.csv('TutorialData/DiscreteData.csv', row.names=1, header=TRUE)\r\nmydata[1:10,]\r\n\r\n#Match data with tree\r\ndata.pruned<-treedata(phy=tree,data = mydata, warnings=FALSE)\r\ntree<-data.pruned$phy\r\nmydata<-data.pruned$data\r\n\r\n#Plot data on tree\r\nplot.phylo(tree,show.tip.label = F)\r\ntiplabels(pie = to.matrix(mydata[,1],sort(unique(mydata[,1]))),piecol=c(\"red\", \"black\"),cex=.3, offset=0)\r\ntiplabels(pie = to.matrix(mydata[,3],sort(unique(mydata[,1]))),piecol=c(\"green\", \"orange\"),cex=.3, offset=.2)\r\n\r\n#####1: Single trait analysis: transition rate comparison\r\nlibrary(corHMM)\r\n #Set up initial rate matrices\r\nrmat.er<-rate.mat.maker(rate.cat=1, hrm=FALSE, ntraits=1, nstates=2, model=\"ER\") #equal transition rates\r\nrmat.ard<-rate.mat.maker(rate.cat=1, hrm=FALSE, ntraits=1, nstates=2, model=\"ARD\") #unequal transition rates\r\nrmat.er #all rates the same\r\nrmat.ard #all rates different \r\n #Set up data\r\ntrt1<-cbind(row.names(mydata),mydata[,1])\r\ntrt2<-cbind(row.names(mydata),mydata[,3])\r\ntrt3<-cbind(row.names(mydata),mydata[,5])\r\n\r\n#Fit models: TRAIT 1\r\nplot.phylo(tree,show.tip.label = F)\r\ntiplabels(pie = to.matrix(mydata[,1],sort(unique(mydata[,1]))),piecol=c(\"red\", \"black\"),cex=.3, offset=0)\r\nres1.er<-corHMM(tree,trt1,rate.cat=1,rate.mat=rmat.er,node.states=\"marginal\") \r\nres1.er\r\nres1.ard<-corHMM(tree,trt1,rate.cat=1,rate.mat=rmat.ard,node.states=\"marginal\") \r\n #compare models: logL and AIC\r\nres1.er$loglik \r\nres1.ard$loglik #identical in this case: no need for formal LRT \r\nres1.er$AICc\r\nres1.ard$AICc #Choose simpler model (lower AICc b/c fewer parameters and same logL)\r\nres1.er$solution #rate transition parameters\r\n\r\n#Fit models: TRAIT 2\r\nplot.phylo(tree,show.tip.label = F)\r\ntiplabels(pie = to.matrix(mydata[,3],sort(unique(mydata[,3]))),piecol=c(\"red\", \"black\"),cex=.3, offset=0)\r\nres2.er<-corHMM(tree,trt2,rate.cat=1,rate.mat=rmat.er,node.states=\"marginal\") \r\nres2.ard<-corHMM(tree,trt2,rate.cat=1,rate.mat=rmat.ard,node.states=\"marginal\") \r\n#compare models: logL and AIC\r\nres2.er$loglik \r\nres2.ard$loglik #identical in this case: no need for formal LRT \r\nres2.er$AICc\r\nres2.ard$AICc #Choose simpler model (lower AICc b/c fewer parameters and same logL)\r\nres2.er$solution #rate transition parameters\r\n\r\n#Fit models: TRAIT 3\r\nplot.phylo(tree,show.tip.label = F)\r\ntiplabels(pie = to.matrix(mydata[,5],sort(unique(mydata[,5]))),piecol=c(\"red\", \"black\"),cex=.3, offset=0)\r\nres3.er<-corHMM(tree,trt3,rate.cat=1,rate.mat=rmat.er,node.states=\"marginal\") \r\nres3.ard<-corHMM(tree,trt3,rate.cat=1,rate.mat=rmat.ard,node.states=\"marginal\") \r\n#compare models: logL and AIC\r\nres3.er$loglik \r\nres3.ard$loglik\r\nLRT<- -2*(res3.er$loglik - res3.ard$loglik) #LRT test\r\nLRT\r\n1-pchisq( LRT,df = 1) #probability of LRT\r\nres3.er$AICc\r\nres3.ard$AICc #ARD preferred\r\nres3.ard$solution #rate transition parameters\r\n #transition from \r\n\r\n\r\n#####2: Two trait analysis: Trait association\r\n#Compare dependent (correlated) change model (ARD) with independent change model (ER) #See Pagel 1994\r\ntrtset12<-cbind(row.names(mydata),mydata[,1:2])\r\ntrtset34<-cbind(row.names(mydata),mydata[,3:4])\r\n\r\n#Traits 1&2\r\nplot.phylo(tree,show.tip.label = F)\r\ntiplabels(pie = to.matrix(mydata[,1],sort(unique(mydata[,1]))),piecol=c(\"red\", \"black\"),cex=.3, offset=0)\r\ntiplabels(pie = to.matrix(mydata[,2],sort(unique(mydata[,2]))),piecol=c(\"green\", \"orange\"),cex=.3, offset=0.2)\r\ndisc.res12.er<-corDISC(tree,trtset12,ntraits=2,model=\"ER\",node.states=\"marginal\") \r\ndisc.res12.ard<-corDISC(tree,trtset12,ntraits=2,model=\"ARD\",node.states=\"marginal\") \r\ndisc.res12.er$loglik\r\ndisc.res12.ard$loglik\r\nLRT<- -2*(disc.res12.er$loglik - disc.res12.ard$loglik) #LRT test\r\n1-pchisq( LRT,df = 4) #probability of LRT\r\ndisc.res12.er$AIC\r\ndisc.res12.ard$AIC #STRONG SUPPORT for correlated (dependent) evolution\r\n #Implemented in phytools\r\ntr1<-mydata[,1]; names(tr1)<-row.names(mydata)\r\ntr2<-mydata[,2]; names(tr2)<-row.names(mydata)\r\ndisc.res12.b<-fitPagel(tree,x=tr1,y=tr2)\r\ndisc.res12.b\r\n #NOTE: LOOK AT PHYLOGENY AND DATA HERE! WE have strong support for correlated evolution, but both traits are co-distributed and clustered\r\n #essentially, 1 evolutionary shift could explain the pattern: the 'BiSSE' problem (need ancestral state estimation: next week)\r\n ##### ALWAYS PLOT YOUR DATA! DON'T JUST RUN STATISTICS (statistical anlaysis alone here leads to mis-interpretation)\r\n\r\n#Traits 3&4\r\nplot.phylo(tree,show.tip.label = F)\r\ntiplabels(pie = to.matrix(mydata[,3],sort(unique(mydata[,3]))),piecol=c(\"red\", \"black\"),cex=.3, offset=0)\r\ntiplabels(pie = to.matrix(mydata[,4],sort(unique(mydata[,4]))),piecol=c(\"green\", \"orange\"),cex=.3, offset=0.2)\r\ndisc.res34.er<-corDISC(tree,trtset34,ntraits=2,model=\"ER\",node.states=\"marginal\") \r\ndisc.res34.ard<-corDISC(tree,trtset34,ntraits=2,model=\"ARD\",node.states=\"marginal\") \r\ndisc.res34.er$loglik\r\ndisc.res34.ard$loglik \r\ndisc.res34.er$AIC\r\ndisc.res34.ard$AIC # VERY STRONG SUPPORT FOR CORRELATED EVOLUTION\r\n\r\n\r\n#####3: Two trait analysis: Order to transitions matters (changes in trait 2 DEPEND on values of trait 1: Maddison 1990)\r\n #Must define the two models for comparison\r\nplot.phylo(tree,show.tip.label = F)\r\ntiplabels(pie = to.matrix(mydata[,3],sort(unique(mydata[,3]))),piecol=c(\"red\", \"black\"),cex=.3, offset=0)\r\ntiplabels(pie = to.matrix(mydata[,4],sort(unique(mydata[,4]))),piecol=c(\"green\", \"orange\"),cex=.3, offset=0.2)\r\neq.rat<-rate.mat.maker(rate.cat=1, hrm=FALSE, ntraits=2, nstates=2, model=\"ER\")\r\neq.rat\r\ndir.rat<-eq.rat; dir.rat[1,2]<-2; dir.rat[3,4]<-3 #e.g., q12<>q34: see Table in Pagel 1994\r\ndir.rat\r\n #NOTE: which values are allowed to differ dependently depends upon the directional hypothesis under investigation!\r\n\r\ndisc.res34.er<-corDISC(tree,trtset34,ntraits=2,model=\"ER\",node.states=\"marginal\") \r\ndisc.res34.dir<-corDISC(tree,trtset34,ntraits=2,model=\"ER\",rate.mat=dir.rat,node.states=\"marginal\") #run with directional hypothesis\r\ndisc.res34.er$loglik\r\ndisc.res34.dir$loglik \r\ndisc.res34.er$AIC\r\ndisc.res34.dir$AIC # # VERY STRONG SUPPORT FOR DIRECTIONAL DEPENDENT EVOLUTION\r\n\r\n", "meta": {"hexsha": "e4070055c0478f63c2d3e02996e4fb99afed9437", "size": 6372, "ext": "r", "lang": "R", "max_stars_repo_path": "practicals/TutorialData/PhyloAssocDiscrete.r", "max_stars_repo_name": "EEOB-Macroevolution/EEOB-565X-Spring2018", "max_stars_repo_head_hexsha": "da192d6744dfaaa68a7f02d6a00ed2642192d43a", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "practicals/TutorialData/PhyloAssocDiscrete.r", "max_issues_repo_name": "EEOB-Macroevolution/EEOB-565X-Spring2018", "max_issues_repo_head_hexsha": "da192d6744dfaaa68a7f02d6a00ed2642192d43a", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "practicals/TutorialData/PhyloAssocDiscrete.r", "max_forks_repo_name": "EEOB-Macroevolution/EEOB-565X-Spring2018", "max_forks_repo_head_hexsha": "da192d6744dfaaa68a7f02d6a00ed2642192d43a", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 47.9097744361, "max_line_length": 141, "alphanum_fraction": 0.7176710609, "num_tokens": 2134, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7905303186696748, "lm_q2_score": 0.6513548511303338, "lm_q1q2_score": 0.5149157580311013}} {"text": "##Set Working Directory\r\nsetwd(\"C:/Users/GitHubRepository/scwrp-rsu-slr-model\")\r\n\r\n##Set Desired Output Directory\r\nOutputDirectory=\"C:/Users/GitHubRepository/scwrp-rsu-slr-model/model output/\"\r\n #This directory is set as a variable that will be used below to output the model data\r\n #Suggested output directory is a subdirectory of the working directory\r\n #Output directory path must end with a forward slash \"/\"\r\n\r\n##############################################################################\r\n#Part 1\r\n##Import csv with input data for all sites in the region\r\ninput <- read.csv(file=\"SLR_Model_Inputs.csv\", head=TRUE, sep=\",\")\r\n\r\n##Create input variables for model\r\n#Time points for baseline and SLR projections\r\nt0 <- 2016\r\nt1 <- 2050\r\nt2 <- 2100\r\n\r\n##Relative SLR (mm/yr)\r\nSL2050 <-input$SLR_2050_mmyr*(t1-t0)\r\nSL2100 <-input$SLR_2100_mmyr*(t2-t0)\r\n\r\n##Accretion: inputs from Lit Review (mm/yr)\r\nAcr2050 <- input$Accretion_mmyr*(t1-t0)\r\nAcr2100 <- input$Accretion_mmyr*(t2-t0)\r\n\r\n##Change in Elevation (m)\r\nElev2050 <- Acr2050*0.001 #convert accretion to m\r\nElev2100 <- Acr2100*0.001\r\n\r\n\r\n##Change in Water level: inputs from rSLR and Mouth Dynamics Modeling\r\n#Mouth Dynamics contribution (m)\r\nMD2050 <- input$MD_WL_2050\r\nMD2100 <- input$MD_WL_2100\r\n\r\n#Change in Water level (m)\r\nWL2050 <-(SL2050*0.001)+(MD2050) #SLR contribution (m)\r\nWL2100 <-(SL2100*0.001)+(MD2100)\r\n\r\n#Intermediate outputs for elevation and water level\r\noutput <-cbind(input, Elev2050, Elev2100, WL2050, WL2100)\r\n #Appends elevation and water level calculations per site to input data\r\n\r\n##Hypsometric curves\r\nHyps <- read.csv(file=\"Hypsometry.csv\", head=TRUE, sep=\",\")\r\n #Import long-format csv containing hypsometrtic curves (standardized to Z*) for all sites in the region\r\n\r\n###Z breaks, by archetype\r\nzbreaks <- read.csv(file=\"Zbreaks_Archetypes.csv\", head=TRUE, sep=\",\")\r\n\r\n##############################################################################\r\n#Part 2\r\n#Create list of sites from input/intermediate output\r\nSites=output$Site_Name\r\nSites = levels(Sites)\r\n\r\nfor (i in 1:length(Sites)){\r\n\r\n #SITE Inputs\r\n Site = subset(Hyps, Name==Sites[i]) #extracts site hypsometry data\r\n PC = subset(output, Site_Name==Sites[i], select=Perch.Correction) #site perch correction\r\n z.bin.max.pc = Site$z.bin.max - (PC[1,1]*1.25) #adjusts Z* (elevation) to correct for perched systems\r\n Site = cbind.data.frame(Site, z.bin.max.pc) #appends adjusted Z* (elevation)\r\n Site_archetype = subset(output, Site_Name==Sites[i], select=ArchetypeCode) #Site-specific archetype classification\r\n \r\n ##Archetype-specific Z* breaks\r\n z.subtidal=subset(zbreaks, Arch_Code==Site_archetype[1,1], select=\"Subtidal\")\r\n z.mudflat=subset(zbreaks, Arch_Code==Site_archetype[1,1], select=\"Intertidal.Mudflat\")\r\n z.low=subset(zbreaks, Arch_Code==Site_archetype[1,1], select=\"Low.Marsh\")\r\n z.mid=subset(zbreaks, Arch_Code==Site_archetype[1,1], select=\"Mid.Marsh\")\r\n z.high=subset(zbreaks, Arch_Code==Site_archetype[1,1], select=\"High.Marsh\")\r\n z.trans=subset(zbreaks, Arch_Code==Site_archetype[1,1], select=\"Transition\")\r\n z.stop=subset(zbreaks, Arch_Code==Site_archetype[1,1], select=\"Stop\")\r\n \r\n #Calculate current area \r\n subtidal=with(Site, sum(Site[z.bin.max.pc=z.subtidal[1,1] & z.bin.max.pc=z.mudflat[1,1] & z.bin.max.pc=z.low[1,1] & z.bin.max.pc=z.mid[1,1] & z.bin.max.pc=z.high[1,1] & z.bin.max.pc=z.trans[1,1] & z.bin.max.pc=z.shallowsubtidal.2050[1,1] & z.bins.2050=z.mudflat.2050[1,1] & z.bins.2050=z.low.2050[1,1] & z.bins.2050=z.mid.2050[1,1] & z.bins.2050=z.high.2050[1,1] & z.bins.2050=z.trans.2050[1,1] & z.bins.2050=z.shallowsubtidal.2100[1,1] & z.bins.2100=z.mudflat.2100[1,1] & z.bins.2100=z.low.2100[1,1] & z.bins.2100=z.mid.2100[1,1] & z.bins.2100=z.high.2100[1,1] & z.bins.2100=z.trans.2100[1,1] & z.bins.21001) for (ki in 2:p) x <- cbind(x,dy[(p+1-ki):(t-1-ki)])\r\n\r\n if (sum(coef)==0){\r\n xi_ns <- as.matrix(seq(1,p,1)+1+constant+trend)\r\n }else{\r\n idex <- colSums(as.matrix(coef%*%matrix(1,1,p))==(matrix(1,nrow(coef),1)%*%seq(1,p,1)))\r\n xi_ns <- seq(1,p,1)\r\n xi_ns <- as.matrix(xi_ns[idex==0]+1+constant+trend)\r\n }\r\n\r\n if (trend==0){\r\n if(constant==0){\r\n\t if (sum(coef)==0){\r\n xi_s <- rbind(1)\r\n }else{xi_s <- rbind(1,(coef+1))\r\n }\r\n }\r\n }\r\n if (trend==0){\r\n if(constant==1){\r\n\tif (sum(coef)==0){\r\n xi_s <- rbind(1,2)\r\n }else{\r\n xi_s <- rbind(1,2,(coef+2))\r\n }\r\n }\r\n }\r\n if (trend==1){\r\n if(constant==1){\r\n\tif (sum(coef)==0){\r\n xi_s <- rbind(1,2,3)\r\n }else{\r\n xi_s <- rbind(1,2,3,(coef+3))\r\n }\r\n }\r\n }\r\n\r\n xs <- as.matrix(x[,xi_s])\r\n xns <- as.matrix(x[,xi_ns])\r\n\r\n if (thresh==1){\r\n qs <- dat[(1+p):(t-1)]-dat[(1+p-mmin):(t-1-mmin)]\r\n qs <- as.matrix(qs)\r\n if(p>1){\r\n\t for (mi in (mmin+1):mmax){\r\n qs <- cbind(qs,(dat[(1+p):(t-1)]-dat[(1+p-mi):(t-1-mi)]))\r\n }\r\n\t }\r\n }\r\n if (thresh==2){\r\n qs <- dat[(1+p-mmin+1):(t-1-mmin+1)]-dat[(1+p-mmin):(t-1-mmin)]\r\n qs <- as.matrix(qs)\r\n if(p>1){\r\n\t for (mi in (mmin+1):mmax){\r\n qs <- cbind(qs,(dat[(1+p-mi+1):(t-1-mi+1)]-dat[(1+p-mi):(t-1-mi)]))\r\n }\r\n\t }\r\n }\r\n if (thresh==3){\r\n qs <- dat[(1+p-mmin+1):(t-mmin)]\r\n qs <- as.matrix(qs)\r\n if(p>1){\r\n\t for (mi in (mmin+1):mmax){\r\n qs <- cbind(qs,(dat[(1+p-mi+1):(t-mi)]))\r\n }\r\n\t }\r\n }\r\n kx <- ncol(x)\r\n ks <- nrow(coef)+1+constant+trend\r\n xx <- solve(t(x)%*%x, tol=1e-25)\r\n bols <- xx%*%(t(x)%*%y) \r\n eols <- y - x%*%bols\r\n sols <- t(eols)%*%eols\r\n wwss <- matrix(0,mn,1)\r\n smins <- matrix(0,mn,1)\r\n lams <- matrix(0,mn,1)\r\n ws <- matrix(0,mn,1)\r\n r1 <- matrix(0,mn,1)\r\n r2 <- matrix(0,mn,1)\r\n t1 <- matrix(0,mn,1)\r\n t2 <- matrix(0,mn,1)\r\n rur <- rbind(1,(ks+1))+trend+constant\r\n indx <- 1:(n)\r\n pi1 <-.2-.05*trim\r\n pi2 <-.8+.05*trim\r\n for (mi in 1:mn){\r\n q <- qs[,mi]\r\n qq <- unique(q)\r\n qq <- as.matrix(sort(qq))\r\n qq <- as.matrix(qq[(floor(n*pi1)+1):(ceiling(n*pi2)-1)])\r\n qn <- nrow(qq)\r\n s <- matrix(0,qn,1)\r\n wws<- matrix(0,qn,1)\r\n for (qi in 1:qn){\r\n d1 <- as.matrix((qlam)\r\nlist(mhat=mhat,lam=lam,e=e,bs_s=bs_s,ar01=ar01,ar02=ar02,b1=b1,b2=b2,r1=r1,r2=r2,t1=t1,t2=t2,ts=ts,ws=ws,w=w,\r\nwws=wws, wwss=wwss,leng=leng,vd=vd)\r\n}\r\n\r\n###########################################################\r\n# Caner & Hansen (2001) TAR Model Estimation Settings #####\r\nthresh <- 2 \r\ntrim <- 1 \r\np <- 1\r\nm <- 0\r\nmmin <- 1\r\nmmax <- p\r\ncoef <- matrix(c(1:p),p,1)\r\n###########################################################\r\n# Simulation Settings #####################################\r\nrep <- 10000\r\nn <- 100\r\nalpha <- 0.8\r\ntr <-as.matrix(c(1:n))\r\ncc <-as.matrix(rep(1,n))\r\n###########################################################\r\n# 5% Criticals T=100 ######################################\r\n R1A <- 21.15948\r\n R2A <- 21.27445\r\n R1B <- 27.69705\r\n R2B <- 27.77254\r\n R1C <- 30.93129\r\n R2C <- 30.95712\r\n# 5% Criticals T=200 ######################################\r\n# R1A <- 20.82907\r\n# R2A <- 21.07073\r\n# R1B <- 26.67934\r\n# R2B <- 26.80033\r\n# R1C <- 29.65022\r\n# R2C <- 29.69269\r\n###########################################################\r\n# Simulation Process\r\nptm <- proc.time()\r\nset.seed(012345)\r\nRTDist <- matrix(0,rep,6)\r\n\r\nfor(i in 1:rep){\r\nyt <- as.matrix(CHSizeDGP(n,alpha))\r\nLNV <- startLNV(yt)\r\nb0a <- LNV$betasA\r\nb0b <- LNV$betasB\r\nb0c <- LNV$betasC\r\nX <- as.matrix(cbind(cc,tr))\r\nELNV <- estLNV(yt, X, b0a, b0b,b0c)\r\nD <- as.matrix(ELNV$ehat)\r\n\r\ntarA <- tur_est2(as.matrix(D[,1]),p,mmin,mmax,m,0,0);\r\ntarB <- tur_est2(as.matrix(D[,2]),p,mmin,mmax,m,0,0);\r\ntarC <- tur_est2(as.matrix(D[,3]),p,mmin,mmax,m,0,0);\r\n\r\nRTDist[i,1] <- sum(tarA$r1[tarA$mhat]>R1A)\r\nRTDist[i,2] <- sum(tarB$r1[tarB$mhat]>R1B)\r\nRTDist[i,3] <- sum(tarC$r1[tarC$mhat]>R1C)\r\nRTDist[i,4] <- sum(tarA$r2[tarA$mhat]>R2A)\r\nRTDist[i,5] <- sum(tarB$r2[tarB$mhat]>R2B)\r\nRTDist[i,6] <- sum(tarC$r2[tarC$mhat]>R2C)\r\n}\r\nrA <- rbind((sum(RTDist[,1])*100)/rep,(sum(RTDist[,4])*100)/rep)\r\nrB <- rbind((sum(RTDist[,2])*100)/rep,(sum(RTDist[,5])*100)/rep)\r\nrC <- rbind((sum(RTDist[,3])*100)/rep,(sum(RTDist[,6])*100)/rep)\r\nCC <- cbind(rA,rB,rC)\r\nproc.time() - ptm\r\nprint(CC)\r\n\r\nprintres <- function(CC){\r\ncat (\"\\n\")\r\ncat (\"Empirical Size Values for Selected Sample Size & Alpha\" , \"\\n\")\r\ncat (\"\\n\")\r\ncat (\"Number of Observations (T):\", n , \"\\n\")\r\ncat (\" Alpha:\", alpha , \"\\n\")\r\ncat (\" Replication:\", rep , \"\\n\")\r\ncat (\" Nominal Size:\", \"%5\" , \"\\n\")\r\ncat (\"\\n\")\r\ncat (\"Model A\" , \"\\n\")\r\ncat (\" \",\"R1t\", \" R2t\" , \"\\n\")\r\ncat (\" \", CC[1,1],\" \",CC[2,1] , \"\\n\")\r\ncat (\"\\n\")\r\ncat (\"Model B\" , \"\\n\")\r\ncat (\" \",\"R1t\", \" R2t\" , \"\\n\")\r\ncat (\" \", CC[1,2],\" \",CC[2,2] , \"\\n\")\r\ncat (\"\\n\")\r\ncat (\"Model C\" , \"\\n\")\r\ncat (\" \",\"R1t\", \" R2t\" , \"\\n\")\r\ncat (\" \", CC[1,3],\" \",CC[2,3] , \"\\n\")\r\ncat (\"\\n\")\r\n}\r\n\r\nprintres(CC)\r\n# END #\r\n\r\n\r\n\r\n\r\n\r\n", "meta": {"hexsha": "5afa72c4cd5bbfa88e18fa6c206c4aeb4ba1d424", "size": 14703, "ext": "r", "lang": "R", "max_stars_repo_path": "Size-1.r", "max_stars_repo_name": "mehmet-ozcan/2021-bookchapter-1", "max_stars_repo_head_hexsha": "576ad6c2c047a152081769dfde62a8af3d4262c8", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Size-1.r", "max_issues_repo_name": "mehmet-ozcan/2021-bookchapter-1", "max_issues_repo_head_hexsha": "576ad6c2c047a152081769dfde62a8af3d4262c8", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Size-1.r", "max_forks_repo_name": "mehmet-ozcan/2021-bookchapter-1", "max_forks_repo_head_hexsha": "576ad6c2c047a152081769dfde62a8af3d4262c8", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 34.7588652482, "max_line_length": 176, "alphanum_fraction": 0.4642589948, "num_tokens": 5335, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.800692021119887, "lm_q2_score": 0.640635868562172, "lm_q1q2_score": 0.5129520284009398}} {"text": "#================================================================\r\nlibrary(truncnorm)\r\n#===============================================================\r\n#Auxiliary functions used during the simulations\r\n#===============================================================\r\n#Function that calculates euclidean distances. It receives two \r\n#matrices of x and y coordinates as parameters.\r\neuclid = function(m1, m2=m1)\r\n{\r\n sqrt(outer(m1[,1], m2[,1],\"-\")^2 + outer(m1[,2], m2[,2],\"-\")^2)\r\n}\r\n#===============================================================\r\n#Function that takes samples from a vector of probabilities\r\n#Given a vector prob of length y, it samples values from 1 to y, the \r\n#probability of value i (i <=y ) being sampled is prob[i].\r\nsampleProb = function(prob, n=1, replace=FALSE)\r\n{\r\n prob[is.na(prob)] = 0\r\n positiveProbs = sum(prob!=0)\r\n if (positiveProbs == 1)\r\n return(which(prob!=0))\r\n else if(positiveProbs > 1)\r\n return(sample(1:length(prob), size = n, replace=replace, prob=prob))\r\n else return(NA)\r\n}\r\n#===============================================================\r\n#Function that runs a simulation of a mating system with information\r\n#filtering during mate choice.\r\n#It simulates one mating season of a population with 1:1 sex ratio \r\n#The function returns a list containing three data.frames, if a \r\n#filename is given, all data is also registered in three csv files.\r\nsimScramble = function(N=100, w = 1, zmean=4, zsd=1, radius=0.10, \r\n B=2, filename=NA, append=TRUE, seed=NA,\r\n choice=1, zmin=1, zmax=Inf, autocorr = FALSE)\r\n{ \r\n \r\n #if a random seed is given, it is used\r\n if(!is.na(seed))\r\n set.seed(seed) \r\n \r\n #a trick to facilitate indexing\r\n xy = c(\"x\",\"y\")\r\n \r\n #creating the data.frames for males and females\r\n females = data.frame(wid=seed, id=1:N, x=runif(N, min=0, max=w),\r\n y=runif(N, min=0, max=w))\r\n males = data.frame(wid=seed, id=1:N, x=runif(N, min=0, max=w), y=runif(N, min=0, max=w), ms=0)\r\n \r\n zmatrix = NA\r\n difs = NA\r\n \r\n #creating data.frame for males\r\n zmean = as.numeric(zmean);zsd = as.numeric(zsd);\r\n zmin = as.numeric(zmin);zmax = as.numeric(zmax);\r\n \r\n #if there is spatial autocorrelation of traits\r\n if (autocorr)\r\n {\r\n MdistOrigin = euclid(cbind(males$x,males$y), matrix(c(0,0), nrow = 1, ncol=2))\r\n MmeanZ = (MdistOrigin/(sqrt(2)*w)*zmean)+1\r\n males$trait = rtruncnorm(N, a=MmeanZ/4, b=zmax, mean = MmeanZ, sd=zsd)\r\n }\r\n else\r\n {\r\n males$trait=rtruncnorm(n = N, a = zmin, b = zmax, mean=zmean, sd=zsd) \r\n }\r\n \r\n \r\n if(choice==1)#directional choice\r\n {\r\n #male trait matrix\r\n zbase = matrix(males$trait, nrow=N, ncol=N, byrow=TRUE)\r\n zmatrix = zbase^B\r\n \r\n }\r\n else #assortative choice\r\n {\r\n #then females also have a trait value(from the same distribution) \r\n \r\n if(autocorr)\r\n {\r\n FdistOrigin = euclid(cbind(females$x,females$y), matrix(c(0,0), nrow = 1, ncol=2))\r\n FmeanZ = (FdistOrigin/(sqrt(2)*w)*zmean)+1\r\n females$trait=rtruncnorm(n = N, a = FmeanZ/4, b = zmax, mean=FmeanZ, sd=zsd)\r\n }\r\n else\r\n {\r\n females$trait=rtruncnorm(n = N, a = zmin, b = zmax, mean=zmean, sd=zsd)\r\n }\r\n \r\n \r\n #calculates all pairwise differences\r\n difs = outer(females$trait, males$trait, \"-\")\r\n \r\n #divides by female trait\r\n fzmatrix = matrix(females$trait, nrow=N, ncol=N, byrow=TRUE)\r\n difs = difs/fzmatrix\r\n difs = abs(difs)\r\n \r\n zbase = difs\r\n zmatrix = exp(-B*(difs))\r\n } \r\n \r\n #In all the matrices, females are lines, males are columns\r\n #distance matrix\r\n smatrix = euclid(females[,xy], males[,xy])\r\n \r\n #adjacency matrix of neighborhood (males available)\r\n #females are connected to males within r distance\r\n nmatrix = ifelse(smatrix<=radius, 1, 0)\r\n \r\n #also, by default females are connected to the closest male\r\n nmatrix[cbind(1:N, apply(smatrix, 1, which.min))] = 1 \r\n \r\n #choice matrix\r\n cmatrix = nmatrix * zmatrix\r\n \r\n #choosing the males\r\n chosenMales = apply(cmatrix, 1, sampleProb, replace=FALSE)\r\n \r\n #data.frame that registers the copulations\r\n groups = data.frame(wid=seed, female = females$id, male = chosenMales)\r\n \r\n #adding mating success to the males\r\n males$ms=0 \r\n tms = tabulate(chosenMales, nbins = N)#table of mating success\r\n males$ms = tms\r\n \r\n #If there is a filename, files must be written\r\n #Columns are separated by \";\" and \",\" is used as decimal separator\r\n if(!(is.na(filename)))\r\n {\r\n write.table(x = females, file = paste(filename, \"Females.csv\", sep=\"\"),\r\n append = append, sep = \";\", dec=\",\", row.names = FALSE, col.names = !append)\r\n write.table(x = males, file = paste(filename, \"Males.csv\", sep=\"\"),\r\n append = append, sep = \";\", dec=\",\", row.names = FALSE, col.names = !append)\r\n write.table(x = groups, file = paste(filename, \"Groups.csv\", sep=\"\"),\r\n append = append, sep = \";\", dec=\",\", row.names = FALSE, col.names = !append)\r\n }\r\n \r\n invisible(list(females=females, males=males,groups=groups))\r\n \r\n}\r\n#====================================================================", "meta": {"hexsha": "ad49f6f1fd277212644a3104bac7bfa5c4c4c90b", "size": 5173, "ext": "r", "lang": "R", "max_stars_repo_path": "IF.simulation.base.r", "max_stars_repo_name": "danilogmuniz/informationFiltering", "max_stars_repo_head_hexsha": "5aa60cd42a246665c162521104d9501dfeec8e29", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "IF.simulation.base.r", "max_issues_repo_name": "danilogmuniz/informationFiltering", "max_issues_repo_head_hexsha": "5aa60cd42a246665c162521104d9501dfeec8e29", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "IF.simulation.base.r", "max_forks_repo_name": "danilogmuniz/informationFiltering", "max_forks_repo_head_hexsha": "5aa60cd42a246665c162521104d9501dfeec8e29", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.9236111111, "max_line_length": 97, "alphanum_fraction": 0.5735549971, "num_tokens": 1482, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8006920116079208, "lm_q2_score": 0.6406358548398982, "lm_q1q2_score": 0.512952011319918}} {"text": "# Conditional Latent Growth Structural Equation Model Power Analysis\n\n# Raymond Viviano\n# rayviviano@gmail.com\n# February 12th, 2020\n\n# TODO: Vary models by small, medium, and large residual variance\n\nsuppressPackageStartupMessages(library(lavaan))\nsuppressPackageStartupMessages(library(mnormt))\nsuppressPackageStartupMessages(library(ggplot2))\nsuppressPackageStartupMessages(library(RColorBrewer))\n\noptions(warn=-1)\n\n################### CHANGE MODEL AND SCRIPT PARAMETERS HERE ####################\n\nintercept_mean_control <- 0 # Doesn't currently do anything.\nintercept_mean_mci <- 0\n\nslope_mean_control <- 0\nmax_slope_mean_mci <- 1\n\nintercept_variance <- 1\nmax_slope_variance <- 1\n\nintercept_slope_covariance <- .2\n\nresidual_variances <- .5\n\nintercept_age_coef_unstd <- .2\nintercept_edu_coef_unstd <- -.2\nintercept_sex_coef_unstd <- .1\nintercept_accult_coef_unstd <- .1\n\nslope_age_coef_unstd <- .2\nslope_edu_coef_unstd <- -.2\nslope_sex_coef_unstd <- .1\nslope_accult_coef_unstd <- .1\n\nnsimulations = 5000\n\n##################### END USE DEFINED PARAMETERS SECTION ######################\n\n\n# Set Base Model, i.e., componenets that do not vary between models\ngrowth_model_base <- \"i =~ 1*y1 + 1*y2 + 1*y3\\ns =~ 0*y1 + 1*y2 + 2*y3\\n\"\ngrowth_model_base <- paste0(growth_model_base, \"i ~~ \", intercept_variance, \"*i\\n\")\n\n\n# Age, edu, sex, and acculturation path coefficients\nage_edu_terms <- paste0(\"i ~ \", intercept_age_coef_unstd, \"*age\\n\")\nage_edu_terms <- paste0(age_edu_terms, \"s ~ \", slope_age_coef_unstd, \"*age\\n\")\nage_edu_terms <- paste0(age_edu_terms, \"i ~ \", intercept_edu_coef_unstd, \"*edu\\n\")\nage_edu_terms <- paste0(age_edu_terms, \"s ~ \", slope_edu_coef_unstd, \"*edu\\n\")\nage_edu_terms <- paste0(age_edu_terms, \"i ~ \", intercept_sex_coef_unstd, \"*sex\\n\")\nage_edu_terms <- paste0(age_edu_terms, \"s ~ \", slope_sex_coef_unstd, \"*sex\\n\")\nage_edu_terms <- paste0(age_edu_terms, \"i ~ \", intercept_accult_coef_unstd, \"*acculturation\\n\")\nage_edu_terms <- paste0(age_edu_terms, \"s ~ \", slope_accult_coef_unstd, \"*acculturation\\n\")\n\n# Set slope variance/covariance and residual variance for Ys\nfor(i in c(max_slope_variance/2, max_slope_variance)){\n # Vary slope variance between .1 and .5\n eval(parse(text=paste0('lgm_svar', i, ' <-paste0(growth_model_base, ',\n '\"s ~~ ', i, '*s +', intercept_slope_covariance, \n '*i\\n\")')))\n \n # Set residual variances to .5\n eval(parse(text=paste0('lgm_svar', i, ' <-paste0(lgm_svar', i, \n ', \"\\ny1 ~~ ', residual_variances,'*y1\\ny2 ~~ ', \n residual_variances,'*y2\\n\",', '\"y3 ~~ ', \n residual_variances,'*y3\\n\")'))) \n\n # Add age and edu covariates\n eval(parse(text=paste0('lgm_svar', i, ' <- paste0(lgm_svar', i, \n ', \"\\n\", age_edu_terms)')))\n}\n\n# Set mean slope for MCI group\nfor(i in c(max_slope_variance/2, max_slope_variance)){\n for(j in seq(0, max_slope_mean_mci, by=max_slope_mean_mci/10)){\n # Add path coefficients between mci and slope (vary slope means\n # for the mci group across simulations)\n eval(parse(text=paste0('lgm_svar', i, '_mci_slp_mean', j,\n ' <- paste0(lgm_svar', i, \n ', \"\\ni~0*1\\ns~', j, '*1\\n\")')))\n }\n}\n\n# Test Model - Correctly Specified\nm1 <- \"\ni =~ 1*y1 + 1*y2 + 1*y3\ns =~ 0*y1 + 1*y2 + 2*y3\ni ~~ i + s\ns ~~ s\ni ~ mci + age + edu + sex + acculturation\ns ~ mci + age + edu + sex + acculturation\n\"\n\n# Test Model - Misspecification 1 - Only MCI\nm2 <- \"\ni =~ 1*y1 + 1*y2 + 1*y3\ns =~ 0*y1 + 1*y2 + 2*y3\ni ~~ i + s\ns ~~ s\ni ~ mci \ns ~ mci\n\"\n\n# Test Model - Misspecification 2 - Add test site and language\nm3 <- \"\ni =~ 1*y1 + 1*y2 + 1*y3\ns =~ 0*y1 + 1*y2 + 2*y3\ni ~~ i + s\ns ~~ s\ni ~ mci + age + edu + sex + site + acculturation + language\ns ~ mci + age + edu + sex + site + acculturation + language\n\"\n\n# Create dataframes to store simulation results\nsim_df_m1 <- data.frame(matrix(ncol = 5, nrow = 0))\nsim_df_m2 <- data.frame(matrix(ncol = 5, nrow = 0))\nsim_df_m3 <- data.frame(matrix(ncol = 5, nrow = 0))\ncolnames(sim_df_m1) <- c('mciSlopeMean', 's_mci', 'svar', 'r05', 'r05p05')\ncolnames(sim_df_m2) <- c('mciSlopeMean', 's_mci', 'svar', 'r05', 'r05p05')\ncolnames(sim_df_m3) <- c('mciSlopeMean', 's_mci', 'svar', 'r05', 'r05p05')\n\n# Run the simulations\nfor(i in c(max_slope_variance/2, max_slope_variance)){\n for(j in seq(0, max_slope_mean_mci, by=max_slope_mean_mci/10)){\n cat(i,j,'\\n')\n eval(parse(text=paste0('lgm_con <- lgm_svar', i, '_mci_slp_mean', j*0.0))) \n eval(parse(text=paste0('lgm_mci <- lgm_svar', i, '_mci_slp_mean', j))) \n \n rmsea05_count <- 0\n r05_path05 <- 0\n mci_slope_path <- 0\n\n for(k in 1:nsimulations){\n # Simulate 100 MCI and 100 Control Observations\n df_con <- lavaan::simulateData(model=lgm_con, model.type=\"growth\",\n sample.nobs=100)\n\n df_mci <- lavaan::simulateData(model=lgm_mci, model.type=\"growth\",\n sample.nobs=100)\n\n # Add MCI columns to both data frames\n df_con[,\"mci\"] <- 0\n df_mci[,\"mci\"] <- 1\n\n # Combine Control and MCI datasets\n df <- rbind(df_con, df_mci)\n\n # Test the correctly specified and misspecified models\n m1_fit <- growth(m1, data=df, missing='fiml') \n # m2_fit <- growth(m2, data=df, missing='fiml')\n # m3_fit <- growth(m3, data=df, missing='fiml')\n\n # Get RMSEA, s~mci estimate, and s~mci pvalue - Model 1\n smci_est <- as.double(parameterEstimates(m1_fit, standardized = FALSE)[15,'est'])\n pval <- as.double(parameterEstimates(m1_fit, standardized = FALSE)[15,'pvalue'])\n rmsea <- as.double(fitMeasures(m1_fit, c('rmsea'))[1])\n\n mci_slope_path <- mci_slope_path + smci_est\n\n if(rmsea <= .05){\n rmsea05_count <- rmsea05_count + 1\n if(pval <= .05){\n r05_path05 <- r05_path05 + 1\n }\n } \n }\n\n rmsea05_prop <- rmsea05_count/nsimulations\n r05_path05_prop <- r05_path05/nsimulations\n mci_slope_path_avg <- mci_slope_path/nsimulations\n\n temp_df <- data.frame(j, mci_slope_path_avg, i, rmsea05_prop, r05_path05_prop)\n colnames(temp_df) <- c('mciSlopeMean', 's_mci', 'svar', 'r05', 'r05p05')\n\n sim_df_m1 <- rbind(sim_df_m1, temp_df)\n \n }\n}\n\n# Sort data\nsim_df_m1 <- sim_df_m1[order(sim_df_m1$mciSlopeMean, sim_df_m1$s_mci),]\n\nprint(sim_df_m1)\n\nsave(sim_df_m1, file=\"./sim_df_m1.Rda\")", "meta": {"hexsha": "935c5811089c37f9854d5604d2f228c16fe1f4db", "size": 6756, "ext": "r", "lang": "R", "max_stars_repo_path": "scripts/lgm-power-analysis.r", "max_stars_repo_name": "rviviano/data-tools", "max_stars_repo_head_hexsha": "97ab15584d01ac81ffe4349e337ed2d76a042ece", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "scripts/lgm-power-analysis.r", "max_issues_repo_name": "rviviano/data-tools", "max_issues_repo_head_hexsha": "97ab15584d01ac81ffe4349e337ed2d76a042ece", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "scripts/lgm-power-analysis.r", "max_forks_repo_name": "rviviano/data-tools", "max_forks_repo_head_hexsha": "97ab15584d01ac81ffe4349e337ed2d76a042ece", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.3717277487, "max_line_length": 95, "alphanum_fraction": 0.6105683837, "num_tokens": 2072, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8289388167733099, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5129311205901026}} {"text": "a = 0\nb = 1\nfor(i in seq(1, 20)) {\n cat(i, b, '\\n')\n c = a + b\n a = b\n b = c\n}", "meta": {"hexsha": "6480d4c3f18e00f747de3e0f6ac6a84f9ca0e958", "size": 82, "ext": "r", "lang": "R", "max_stars_repo_path": "resolucao/r/fibonacci.r", "max_stars_repo_name": "rafaelbes/numericalAnalysis", "max_stars_repo_head_hexsha": "31f2b9cd5fb62cee9a649ac0257024de757eede4", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "resolucao/r/fibonacci.r", "max_issues_repo_name": "rafaelbes/numericalAnalysis", "max_issues_repo_head_hexsha": "31f2b9cd5fb62cee9a649ac0257024de757eede4", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "resolucao/r/fibonacci.r", "max_forks_repo_name": "rafaelbes/numericalAnalysis", "max_forks_repo_head_hexsha": "31f2b9cd5fb62cee9a649ac0257024de757eede4", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 4, "max_forks_repo_forks_event_min_datetime": "2020-12-15T00:31:38.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-09T14:52:19.000Z", "avg_line_length": 10.25, "max_line_length": 22, "alphanum_fraction": 0.3536585366, "num_tokens": 48, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7826624789529376, "lm_q2_score": 0.6548947155710233, "lm_q1q2_score": 0.5125615215419961}} {"text": " DGGESX Example Program Results\n\n Matrix A\n 1 2 3 4\n 1 3.9000 12.5000 -34.5000 -0.5000\n 2 4.3000 21.5000 -47.5000 7.5000\n 3 4.3000 21.5000 -43.5000 3.5000\n 4 4.4000 26.0000 -46.0000 6.0000\n\n Matrix B\n 1 2 3 4\n 1 1.0000 2.0000 -3.0000 1.0000\n 2 1.0000 3.0000 -5.0000 4.0000\n 3 1.0000 3.0000 -4.0000 3.0000\n 4 1.0000 3.0000 -4.0000 4.0000\n\n Number of eigenvalues for which SELCTG is true = 2\n (dimension of deflating subspaces)\n\n Selected generalized eigenvalues\n 1 ( 2.000, 0.000)\n 2 ( 4.000, 0.000)\n\n Reciprocals of left and right projection norms onto\n the deflating subspaces for the selected eigenvalues\n RCONDE(1) = 1.9E-01, RCONDE(2) = 1.8E-02\n\n Reciprocal condition numbers for the left and right\n deflating subspaces\n RCONDV(1) = 5.4E-02, RCONDV(2) = 9.0E-02\n\n Approximate asymptotic error bound for selected eigenvalues = 1.1E-13\n Approximate asymptotic error bound for the deflating subspaces = 2.4E-13\n", "meta": {"hexsha": "f510706d3259d7e19e0b563aba3e901efa10ef22", "size": 1139, "ext": "r", "lang": "R", "max_stars_repo_path": "examples/baseresults/dggesx_example.r", "max_stars_repo_name": "numericalalgorithmsgroup/LAPACK_examples", "max_stars_repo_head_hexsha": "0dde05ae4817ce9698462bbca990c4225337f481", "max_stars_repo_licenses": ["BSD-3-Clause-Open-MPI"], "max_stars_count": 28, "max_stars_repo_stars_event_min_datetime": "2018-01-28T15:48:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-18T09:26:43.000Z", "max_issues_repo_path": "examples/baseresults/dggesx_example.r", "max_issues_repo_name": "numericalalgorithmsgroup/LAPACK_examples", "max_issues_repo_head_hexsha": "0dde05ae4817ce9698462bbca990c4225337f481", "max_issues_repo_licenses": ["BSD-3-Clause-Open-MPI"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "examples/baseresults/dggesx_example.r", "max_forks_repo_name": "numericalalgorithmsgroup/LAPACK_examples", "max_forks_repo_head_hexsha": "0dde05ae4817ce9698462bbca990c4225337f481", "max_forks_repo_licenses": ["BSD-3-Clause-Open-MPI"], "max_forks_count": 18, "max_forks_repo_forks_event_min_datetime": "2019-04-19T12:22:40.000Z", "max_forks_repo_forks_event_max_datetime": "2022-02-18T03:32:12.000Z", "avg_line_length": 33.5, "max_line_length": 74, "alphanum_fraction": 0.5794556629, "num_tokens": 469, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7662936324115011, "lm_q2_score": 0.668880247169804, "lm_q1q2_score": 0.5125586742520517}} {"text": "\n\n\n\n#' Fitting bivariate/multivariate MGL and MGL-EV copula regression models.\n#'\n#' @description \\code{MGL.reg} is used to fit bivariate MGL and MGL-EV copula regression models for two continuous variables.\n#' @param U two-dimensional matrix with values in \\eqn{[0,1]}.\n#' @param X design matrix\n#' @param copula 'MGL', 'MGL180', \"MGL-EV\", \"MGL-EV180\", \"Gumbel\", \"MGB2\".\n#' @param initpar Initial values for the parameters to be optimized over.\n#' @param hessian Logical. Should a numerically differentiated Hessian matrix be returned?\n#' @param ... additional arguments, see \\code{\\link[stats]{optim}} for more details.\n#' @importFrom stats qbeta\n#' @importFrom stats optim\n#' @return A list containing the following components:\n#' * loglike: the value of the estimated maximum of the loglikelihood function.\n#' * copula: the name of the fitted copula. \"MGL180\" and \"MGL-EV180\" denote the survival MGL and MGL-EV copula respectively.\n#' * estimates: the point at which the maximum value of the loglikelihood is obtained.\n#' * se: the standard errors of the estimators.\n#' * AIC, BIC: the goodness fit of the regression models.\n#' * hessian: the hessian at the estimated maximum of the loglikelihood (if requested).\n#' @details\n#' The estimation method is performed via \\code{\\link[stats]{optim}} function. Y1 and Y2 are both continuous variables.\n#'\n#' copula: \"MGL180\" and \"MGLEV180\" denote the survival MGL and survival MGL-EV copula respectively.\n#' * For \"Gumbel\" regression model, the copula parameter \\deqn{\\delta_i = \\exp(X_i^T\\beta) + 1,} where \\eqn{{X}_{i}=(1,x_{i1}...,x_{ik})} denotes the vector of covariates and \\eqn{\\beta} is the vector of coefficients to be estimated in the copula regression.\n#' * For \"MGL\", \"MGL180\", \"MGL-EV\", \"MGL-EV180\" regression model, the copula parameter \\deqn{\\delta_i = \\exp(X_i^T\\beta).}\n#' * For \"MGB2\", the copula parameter \\deqn{\\q_i = \\exp(X_i^T\\beta)} and \\eqn{(p_1,p_2)} remain to be constant.\n#' * Note that the regression modelling can be extended to the high-dimensional case when copula is \"MGL180\" and \"MGL\".\n#'\n#' @examples\n#' # 10-dimensional regression models\n#' set.seed(111)\n#' d <- 10\n#' n <- 1000 # sample size\n#' beta.true <- c(-0.6, 0.5, 0.2) # true regression coefficients\n#' x1 <- rnorm(n, 0, 1)\n#' x2 <- rnorm(n, 0, 1)\n#' X <- model.matrix(~ x1 + x2) # design matrix\n#' delta.sim <- as.vector(exp(X%*%beta.true)) # true copula parameters\n#' Usim <- matrix(0, nrow = n, ncol = d)\n#' for (i in 1:n){\n#' Usim[i, ] <- rcMGL.multi(n = 1, d = d, pars = delta.sim[i])\n#' }\n#' m.MGLMGA <- MGL.reg(U = Usim, copula = \"MGL\",\n#' X = X, method = \"Nelder-Mead\",\n#' initpar = c(-0.32, 0.001, 0.001)\n#' )\n#' m.MGLMGA\n#'\n#' @export\n#'\n#'\nMGL.reg <- function(U, X, copula = c(\n \"MGL\", \"MGL180\", \"MGL-EV\",\n \"MGL-EV180\",\n \"Gumbel\", \"MGB2\"),\n hessian = TRUE, initpar, ...) {\n dcMGL.reg <- function(U, param) {\n dim <- length(U)\n a <- 1 / param[1]\n q <- qbeta(1 - U, shape1 = 0.5, shape2 = a) / (1 - qbeta(1 - U, shape1 = 0.5, shape2 = a))\n logdc <- (dim - 1) * lgamma(a) + lgamma(a + dim / 2) - dim * lgamma(a + 0.5) + (a + 0.5) * sum(log(q + 1)) - (a + dim / 2) * log(sum(q) + 1)\n out <- exp(logdc)\n return(out)\n }\n\n dcMGL180.reg <- function(U, param) {\n dcMGL.reg(1 - U, param = param)\n }\n\n dcMGLEV180.reg <- function(U, param) {\n u1 <- U[1]\n u2 <- U[2]\n as.numeric(dcMGLEV180.bivar(u1, u2, param = param[1]))\n }\n\n dcMGLEV.reg <- function(U, param) {\n u1 <- U[1]\n u2 <- U[2]\n as.numeric(dcMGLEV.bivar(u1, u2, param = param[1]))\n }\n\n dgumcop.reg <- function(U, param) {\n as.numeric(fCopulae::devCopula(u = U[1], v = U[2], type = \"gumbel\", param = param[1])) # Bivariate Extreme\n }\n\n\n if (copula == \"MGL\") {\n dcop <- dcMGL.reg\n } else if (copula == \"MGL180\") {\n dcop <- dcMGL180.reg\n } else if (copula == \"MGL-EV\") {\n dcop <- dcMGLEV.reg\n } else if (copula == \"MGL-EV180\") {\n dcop <- dcMGLEV180.reg\n } else if (copula == \"Gumbel\") {\n dcop <- dgumcop.reg\n }\n\n copLogL <- function(pars, X) {\n ll <- 0\n\n if (copula == \"Gumbel\") {\n delta <- exp(X %*% pars) + 1\n for (i in seq_len(nrow(X))) {ll[i] <- dcop(U[i, ], param = as.vector(delta[i]))}\n } else if (copula == \"MGB2\"){\n p1 <- exp(pars[ncol(X) + 1])\n p2 <- exp(pars[ncol(X) + 2])\n q <- exp(X%*%pars[1:(ncol(X))])\n for (i in seq_len(nrow(X))) {ll[i] <- dcMGB2.bivar(u1 = U[i,1], u2 = U[i,2], pars1 = p1, pars2 = p2, pars3 = q[i])}\n } else {\n delta <- exp(X %*% pars)\n for (i in seq_len(nrow(X))) {ll[i] <- dcop(U[i, ], param = as.vector(delta[i]))}\n }\n res <- -sum((log(ll)))\n return(res)\n }\n\n resopt <- optim(\n par = initpar,\n fn = copLogL,\n X = X,\n hessian = hessian, ...\n )\n\n\n if (hessian == TRUE){\n out <- list(\n loglike = -resopt$value,\n copula = list(name = copula),\n estimates = resopt$par,\n se = sqrt(diag(solve(resopt$hessian))),\n hessian = -resopt$hessian,\n AIC = 2 * length(resopt$par) + 2 * resopt$value,\n BIC = log(nrow(U)) * length(resopt$par) + 2 * resopt$value)\n } else {\n out <- list(\n loglike = -resopt$value,\n copula = list(name = copula),\n estimates = resopt$par,\n AIC = 2 * length(resopt$par) + 2 * resopt$value,\n BIC = log(nrow(U)) * length(resopt$par) + 2 * resopt$value)\n }\n out\n\n}\n", "meta": {"hexsha": "b539606549808b176f60d666b8d34a71f3b2d8e2", "size": 5480, "ext": "r", "lang": "R", "max_stars_repo_path": "R/MGLReg.r", "max_stars_repo_name": "lizhengxiao/rMGLReg", "max_stars_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/MGLReg.r", "max_issues_repo_name": "lizhengxiao/rMGLReg", "max_issues_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2021-08-10T13:04:07.000Z", "max_issues_repo_issues_event_max_datetime": "2021-11-23T11:13:27.000Z", "max_forks_repo_path": "R/MGLReg.r", "max_forks_repo_name": "lizhengxiao/rMGLReg", "max_forks_repo_head_hexsha": "8823d8a0409616ddb3bb75cdba436865dd979604", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 36.7785234899, "max_line_length": 258, "alphanum_fraction": 0.5846715328, "num_tokens": 1862, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8479677737461007, "lm_q2_score": 0.6039318337259584, "lm_q1q2_score": 0.5121147325390012}} {"text": "#' @title calcActualDisp\n#'\n#' @description Estimate actual loaded displacement (\\code{actualDisplacement})\n#' (m^3) using a vessel's actual draft.\n#'\n#' @param Cb Maximum block coefficient (vector of numericals, dimensionless) (see \\code{\\link{calcCb}})\n#' @param Cbw Waterline block coefficient (vector of numericals, dimensionless) (see \\code{\\link{calcCbw}})\n#' @param actualDraft Actual draft (vector of numericals, m)\n#' @param maxDraft Maximum summer load line draft (vector of numericals, m)\n#' @param maxDisplacement Maximum ship displacement (vector of numericals, m^3)\n#'\n#' @details\n#' Loaded displacement is estimated using actual draft. Actual draft is\n#' typically obtained from sources such as AIS messages or ship records.\n#' Uses Riddlesworth method (MAN, 2011).\n#'\n#' @return \\code{actualDisplacement} (vector of numericals, m^3)\n#'\n#' @references\n#' \\href{https://www.man-es.com/marine/products/propeller-aft-ship}{MAN Energy\n#' Solutions. 2011. \"Basic Principles of Propulsion.\" pg. 9.}\n#'\n#' @seealso \\code{\\link{calcCb}}, \\code{\\link{calcCbw}}\n#'\n#' @examples\n#' calcActualDisp(c(0.82,0.66), c(0.81,0.65), c(12.5,11.1), c(13.6,11.5), c(80097,52382))\n#' calcActualDisp(0.82, 0.81, 12.5, 13.6, 80097)\n#'\n#' @export\n\ncalcActualDisp <- function(Cb, Cbw, actualDraft, maxDraft, maxDisplacement){\n\n actualDisplacement<- ((pmin(0.99,\n Cb\n ))/Cbw)*(actualDraft/maxDraft)*maxDisplacement\n return(actualDisplacement)\n\n}\n", "meta": {"hexsha": "24826c836ace032767117b3b836ec7864caf587d", "size": 1458, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcActualDisp.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcActualDisp.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcActualDisp.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 37.3846153846, "max_line_length": 107, "alphanum_fraction": 0.70781893, "num_tokens": 438, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.8128673178375734, "lm_q2_score": 0.6297746213017459, "lm_q1q2_score": 0.5119232072597237}} {"text": "## this script produces a bifurcation diagramm for the 3-species food chain\n\nrm(list=ls())\n\nif (!\"EMD\" %in% installed.packages()) install.packages(\"EMD\")\nlibrary(EMD)\n\n\n\n\n\n\n# bifurcation data\nbdd <- data.frame(\n q = numeric(),\n b1.min = numeric(),\n b1.max = numeric(),\n b2.min = numeric(),\n b2.max = numeric(),\n b3.min = numeric(),\n b3.max = numeric()\n )\n\n\nnutrient<-seq(0,10,0.1)\nfor (j in 1:5)\n{\nfor (i in 1:length(nutrient))\n{\n out2 <-read.table(paste(\"../../Simulations/Nutrients/timeseries_\",i,\"_\",j,\".out\",sep=\"\"),header=T)\n\n \n ### searching mins and maxs\n ## b1min\n ifelse(\n extrema(out2$B0P0)$nextreme<2,\n b1.min <- rep(min(out2$B0P0),25), # if no extrema have been found, take the minimum value 25 times (note: minima are in this case identical to the fixed point value!)\n b1.min <- out2$B0P0[extrema(out2$B0P0)$minindex[1:25]] # otherwise, take the first 25 minima that have been found\n )\n ## b1max\n ifelse(\n extrema(out2$B0P0)$nextreme<2,\n b1.max <- rep(min(out2$B0P0),25),\n b1.max <- out2$B0P0[extrema(out2$B0P0)$maxindex[1:25]]\n ) \n ## b2min\n ifelse(\n extrema(out2$B1P0)$nextreme<2,\n b2.min <- rep(min(out2$B1P0),25),\n b2.min <- out2$B1P0[extrema(out2$B1P0)$minindex[1:25]]\n )\n ## b2max\n ifelse(\n extrema(out2$B1P0)$nextreme<2,\n b2.max <- rep(min(out2$B1P0),25),\n b2.max <- out2$B1P0[extrema(out2$B1P0)$maxindex[1:25]]\n )\n ## b3min\n ifelse(\n extrema(out2$B2P0)$nextreme<2,\n b3.min <- rep(min(out2$B2P0),25),\n b3.min <- out2$B2P0[extrema(out2$B2P0)$minindex[1:25]]\n )\n ## b3max\n ifelse(\n extrema(out2$B2P0)$nextreme<2,\n b3.max <- rep(min(out2$B2P0),25),\n b3.max <- out2$B2P0[extrema(out2$B2P0)$maxindex[1:25]]\n ) \n bddx <- data.frame(\n q = rep(nutrient[i],25),\n b1.min = b1.min,\n b1.max = b1.max,\n b2.min = b2.min,\n b2.max = b2.max,\n b3.min = b3.min,\n b3.max = b3.max\n )\n bdd <- rbind(bdd,bddx)\n print(i)\n \n}\n}\n\nwrite.table(bdd,\"Nutrient.txt\", sep=\",\", row.names = F)\n", "meta": {"hexsha": "d84a561ca5329bba5429288ac6130c995449ee0e", "size": 1992, "ext": "r", "lang": "R", "max_stars_repo_path": "Code-Metafoodweb/R/1-patch/nutrient.r", "max_stars_repo_name": "RemoRyser/Metafoodweb", "max_stars_repo_head_hexsha": "524d0a5362d914cafb7cca5efd08ad20f437a57e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2022-03-25T21:49:06.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-25T21:49:06.000Z", "max_issues_repo_path": "Code-Metafoodweb/R/1-patch/nutrient.r", "max_issues_repo_name": "RemoRyser/Metafoodweb", "max_issues_repo_head_hexsha": "524d0a5362d914cafb7cca5efd08ad20f437a57e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Code-Metafoodweb/R/1-patch/nutrient.r", "max_forks_repo_name": "RemoRyser/Metafoodweb", "max_forks_repo_head_hexsha": "524d0a5362d914cafb7cca5efd08ad20f437a57e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 23.1627906977, "max_line_length": 170, "alphanum_fraction": 0.6129518072, "num_tokens": 796, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8080672135527632, "lm_q2_score": 0.6334102775181399, "lm_q1q2_score": 0.5118380779897658}} {"text": "#' @rdname geom_boxplot\n#' @param coef Length of the whiskers as multiple of IQR. Defaults to 1.5.\n#' @inheritParams stat_identity\n#' @section Computed variables:\n#' `stat_boxplot()` provides the following variables, some of which depend on the orientation:\n#' \\describe{\n#' \\item{width}{width of boxplot}\n#' \\item{ymin *or* xmin}{lower whisker = smallest observation greater than or equal to lower hinge - 1.5 * IQR}\n#' \\item{lower *or* xlower}{lower hinge, 25% quantile}\n#' \\item{notchlower}{lower edge of notch = median - 1.58 * IQR / sqrt(n)}\n#' \\item{middle *or* xmiddle}{median, 50% quantile}\n#' \\item{notchupper}{upper edge of notch = median + 1.58 * IQR / sqrt(n)}\n#' \\item{upper *or* xupper}{upper hinge, 75% quantile}\n#' \\item{ymax *or* xmax}{upper whisker = largest observation less than or equal to upper hinge + 1.5 * IQR}\n#' }\n#' @export\nstat_boxplot <- function(mapping = NULL, data = NULL,\n geom = \"boxplot\", position = \"dodge2\",\n ...,\n coef = 1.5,\n na.rm = FALSE,\n orientation = NA,\n show.legend = NA,\n inherit.aes = TRUE) {\n layer(\n data = data,\n mapping = mapping,\n stat = StatBoxplot,\n geom = geom,\n position = position,\n show.legend = show.legend,\n inherit.aes = inherit.aes,\n params = list2(\n na.rm = na.rm,\n orientation = orientation,\n coef = coef,\n ...\n )\n )\n}\n\n\n#' @rdname ggplot2-ggproto\n#' @format NULL\n#' @usage NULL\n#' @export\nStatBoxplot <- ggproto(\"StatBoxplot\", Stat,\n required_aes = c(\"y|x\"),\n non_missing_aes = \"weight\",\n setup_data = function(self, data, params) {\n data <- flip_data(data, params$flipped_aes)\n data$x <- data$x %||% 0\n data <- remove_missing(\n data,\n na.rm = params$na.rm,\n vars = \"x\",\n name = \"stat_boxplot\"\n )\n flip_data(data, params$flipped_aes)\n },\n\n setup_params = function(self, data, params) {\n params$flipped_aes <- has_flipped_aes(data, params, main_is_orthogonal = TRUE,\n group_has_equal = TRUE,\n main_is_optional = TRUE)\n data <- flip_data(data, params$flipped_aes)\n\n has_x <- !(is.null(data$x) && is.null(params$x))\n has_y <- !(is.null(data$y) && is.null(params$y))\n if (!has_x && !has_y) {\n cli::cli_abort(\"{.fn {snake_class(self)}} requires an {.field x} or {.field y} aesthetic.\")\n }\n\n params$width <- params$width %||% (resolution(data$x %||% 0) * 0.75)\n\n if (is.double(data$x) && !has_groups(data) && any(data$x != data$x[1L])) {\n cli::cli_warn(c(\n \"Continuous {.field {flipped_names(params$flipped_aes)$x}} aesthetic\",\n \"i\" = \"did you forget {.code aes(group = ...)}?\"\n ))\n }\n\n params\n },\n\n extra_params = c(\"na.rm\", \"orientation\"),\n\n compute_group = function(data, scales, width = NULL, na.rm = FALSE, coef = 1.5, flipped_aes = FALSE) {\n data <- flip_data(data, flipped_aes)\n qs <- c(0, 0.25, 0.5, 0.75, 1)\n\n if (!is.null(data$weight)) {\n mod <- quantreg::rq(y ~ 1, weights = weight, data = data, tau = qs)\n stats <- as.numeric(stats::coef(mod))\n } else {\n stats <- as.numeric(stats::quantile(data$y, qs))\n }\n names(stats) <- c(\"ymin\", \"lower\", \"middle\", \"upper\", \"ymax\")\n iqr <- diff(stats[c(2, 4)])\n\n outliers <- data$y < (stats[2] - coef * iqr) | data$y > (stats[4] + coef * iqr)\n if (any(outliers)) {\n stats[c(1, 5)] <- range(c(stats[2:4], data$y[!outliers]), na.rm = TRUE)\n }\n\n if (length(unique(data$x)) > 1)\n width <- diff(range(data$x)) * 0.9\n\n df <- new_data_frame(as.list(stats))\n df$outliers <- list(data$y[outliers])\n\n if (is.null(data$weight)) {\n n <- sum(!is.na(data$y))\n } else {\n # Sum up weights for non-NA positions of y and weight\n n <- sum(data$weight[!is.na(data$y) & !is.na(data$weight)])\n }\n\n df$notchupper <- df$middle + 1.58 * iqr / sqrt(n)\n df$notchlower <- df$middle - 1.58 * iqr / sqrt(n)\n\n df$x <- if (is.factor(data$x)) data$x[1] else mean(range(data$x))\n df$width <- width\n df$relvarwidth <- sqrt(n)\n df$flipped_aes <- flipped_aes\n flip_data(df, flipped_aes)\n }\n)\n", "meta": {"hexsha": "8d951e16543d2cae1ca4732eb04e4692802738b3", "size": 4293, "ext": "r", "lang": "R", "max_stars_repo_path": "R/stat-boxplot.r", "max_stars_repo_name": "sthagen/tidyverse-ggplot2", "max_stars_repo_head_hexsha": "31dce56e38c091725e16a577867ffd547352de85", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/stat-boxplot.r", "max_issues_repo_name": "sthagen/tidyverse-ggplot2", "max_issues_repo_head_hexsha": "31dce56e38c091725e16a577867ffd547352de85", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/stat-boxplot.r", "max_forks_repo_name": "sthagen/tidyverse-ggplot2", "max_forks_repo_head_hexsha": "31dce56e38c091725e16a577867ffd547352de85", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 33.2790697674, "max_line_length": 113, "alphanum_fraction": 0.5660377358, "num_tokens": 1261, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7461389817407016, "lm_q2_score": 0.6859494550081926, "lm_q1q2_score": 0.5118136278854021}} {"text": "# Simulation Tutorial in R for Engineers and Geoscientists \r\n# Michael Pyrcz, University of Texas at Austin, Twitter @GeostatsGuy\r\n\r\n# This will be used in my Introduction to Geostatistics undergraduate class \r\n# It is assumed that students have no previous R experience. \r\n# This utilizes the gstat library by Edzer Pedesma, appreciation to Dr. Pedesma for assistance.\r\n\r\n# Load the required libraries, you may have to first go to \"Tools/Install Packages...\" to install these first\r\nlibrary(gstat) # geostatistical methods by Edzer Pebesma\r\nlibrary(sp) # spatial points addition to regular data frames\r\nlibrary(plyr) # splitting, applying and combining data by Hadley Wickham \r\nlibrary(fields) # required for the image plots\r\n\r\n# Specify the grid parameters (same as GSLIB / GEO-DAS parameterization in 2D)\r\nnx = 100\r\nny = 100\r\nxmn = 5.0\r\nymn = 5.0\r\nxsize = 40.0\r\nysize = 40.0\r\nxmin = xmn - xsize*0.5\r\nymin = ymn - ysize*0.5\r\nxmax = xmin + nx * xsize\r\nymax = ymin + ny * ysize\r\nx<-seq(xmin,xmax,by=xsize) # used for axes on image plots\r\ny<-seq(ymin,ymax,by=ysize) \r\n\r\n# Specify the parameters For plotting \r\ncolmap = topo.colors(100) # define the color map and descritation\r\n\r\n# Declare functions\r\n\r\n# This function completes standard normal transform on a data vector\r\n\r\nnscore <- function(x) { # written by Ashton Shortridge, May/June, 2008\r\n # Takes a vector of values x and calculates their normal scores. Returns \r\n # a list with the scores and an ordered table of original values and\r\n # scores, which is useful as a back-transform table. See backtr().\r\n nscore <- qqnorm(x, plot.it = FALSE)$x # normal score \r\n trn.table <- data.frame(x=sort(x),nscore=sort(nscore))\r\n return (list(nscore=nscore, trn.table=trn.table))\r\n}\r\n\r\n# This function builds a spatial points dataframe with the locations for estimation / simulation \r\n\r\naddcoord <- function(nx,xmin,xsize,ny,ymin,ysize) { # Michael Pyrcz, March, 2018 \r\n # makes a 2D dataframe with coordinates based on GSLIB specification\r\n coords = matrix(nrow = nx*ny,ncol=2)\r\n ixy = 1\r\n for(iy in 1:nx) {\r\n for(ix in 1:ny) {\r\n coords[ixy,1] = xmin + (ix-1)*xsize \r\n coords[ixy,2] = ymin + (iy-1)*ysize \r\n ixy = ixy + 1\r\n }\r\n }\r\n coords.df = data.frame(coords)\r\n colnames(coords.df) <- c(\"X\",\"Y\")\r\n coordinates(coords.df) =~X+Y\r\n return (coords.df)\r\n} \r\n\r\nsim2darray <- function(spdataframe,nx,ny,ireal) { # Michael Pyrcz, March, 2018 \r\n # makes a 2D array from realizations spatial point dataframe\r\n model = matrix(nrow = nx,ncol = ny)\r\n ixy = 1\r\n for(iy in 1:ny) {\r\n for(ix in 1:nx) {\r\n model[ix,iy] = spdataframe@data[ixy,ireal] \r\n ixy = ixy + 1\r\n }\r\n }\r\n return (model)\r\n} \r\n\r\nsim2vector <- function(spdataframe,nx,ny,ireal) { # Michael Pyrcz, March, 2018 \r\n # makes a 1D vector from spatial point dataframe\r\n model = rep(0,nx*ny)\r\n ixy = 1\r\n for(iy in 1:ny) {\r\n for(ix in 1:nx) {\r\n model[ixy] = spdataframe@data[ixy,ireal] \r\n ixy = ixy + 1\r\n }\r\n }\r\n return (model)\r\n} \r\n\r\n# Set the working directory, I always like to do this so I don't lose files and to simplify subsequent read and writes\r\nsetwd(\"C:/PGE337\")\r\n\r\n# Read the data table from a comma delimited file - data on GitHub/GeostatsGuy/GeoDataSets\r\nmydata = read.csv(\"2D_MV_200Wells.csv\") # read in comma delimited data file\r\n\r\n# Let's visualize the first several rows of our data so we can make sure we successfully loaded it\r\nhead(mydata) # show the first several rows of a data table in the console\r\n# The columns are variables with variable names at the top and the rows are samples\r\n\r\n# Convert the dataframe to a spatial points dataframe\r\nclass(mydata) # confirms that it is a dataframe\r\ncoordinates(mydata) = ~X+Y # indicate the X, Y spatial coordinates\r\nsummary(mydata) # confirms that it is now a spatial points dataframe\r\nhead(coordinates(mydata)) # check the first several coordinates\r\n\r\n# Normal scores transform of the porosity data to assist with variogram calculation\r\nnpor.trn = nscore(mydata$porosity) # normal scores transform\r\nmydata[[\"NPorosity\"]]<-npor.trn$nscore # append the normal scores transform into the spatial data table\r\nhead(mydata)\r\n\r\n# Make the model 2D grid\r\ncoords <- addcoord(nx,xmin,xsize,ny,ymin,ysize) # make a dataframe with all the estimation locations\r\nsummary(coords) # check the coordinates\r\n\r\n# What is the sill of the raw variable\r\nsill = var(mydata$porosity)\r\nmin = min(mydata$porosity)\r\nmax = max(mydata$porosity)\r\nzlim = c(min,max) # define the property min and max\r\n\r\n# Anisotropic variogram\r\nvm1 <- vgm(psill = 0.5*sill, \"Sph\", 400, anis = c(000, 1.0),nugget=0.5*sill)\r\nvm1\r\nvm2 <- vgm(psill = 1.0*sill, \"Exp\", 200, anis = c(035, 0.5),nugget=0.0*sill)\r\nvm2\r\nvm3 <- vgm(psill = 1.0*sill, \"Sph\", 600, anis = c(060, 0.2),nugget=0.0*sill)\r\nvm3\r\n\r\n# Gaussian simulation\r\n\r\ncondsim1 = krige(porosity~1, mydata, coords, model = vm1, nmax = 100, nsim = 4)\r\ncondsim2 = krige(porosity~1, mydata, coords, model = vm2, nmax = 100, nsim = 4)\r\ncondsim3 = krige(porosity~1, mydata, coords, model = vm3, nmax = 100, nsim = 4)\r\n\r\nkrige1 = krige(porosity~1, mydata, coords, model = vm1, nmax = 100)\r\nkrige2 = krige(porosity~1, mydata, coords, model = vm2, nmax = 100)\r\nkrige3 = krige(porosity~1, mydata, coords, model = vm3, nmax = 100)\r\n\r\n# nmax is the maximum number of data for each kriging solution\r\n# this may take a long time to run, decrease nmax to spped up (e.g. 100)\r\n\r\n# Plot all the realizations\r\n#par(mar=c(1,1,1,1))\r\npar(mfrow=c(2,2))\r\nreal1 <- sim2darray(condsim1,nx,ny,1) # extract realization #1 to a 2D array and plot\r\n#image.plot(real1,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #1\", outer=F);box(which=\"plot\")\r\nimage.plot(real1,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim);\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nreal2 <- sim2darray(condsim1,nx,ny,2) # extract realization #2 to a 2D array and plot\r\nimage.plot(real2,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #2\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nreal3 <- sim2darray(condsim1,nx,ny,3) # extract realization #3 to a 2D array and plot\r\nimage.plot(real3,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #3\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nest <- sim2darray(krige1,nx,ny,1) # extract realization #4 to a 2D array and plot\r\nimage.plot(est,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Kriging\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\npar(mfrow=c(2,2))\r\nreal1 <- sim2darray(condsim2,nx,ny,1) # extract realization #1 to a 2D array and plot\r\nimage.plot(real1,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #1\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nreal2 <- sim2darray(condsim2,nx,ny,2) # extract realization #2 to a 2D array and plot\r\nimage.plot(real2,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #2\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nreal3 <- sim2darray(condsim2,nx,ny,3) # extract realization #3 to a 2D array and plot\r\nimage.plot(real3,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #3\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nest <- sim2darray(krige2,nx,ny,1) # extract realization #4 to a 2D array and plot\r\nimage.plot(est,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Kriging\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\npar(mfrow=c(2,2))\r\nreal1 <- sim2darray(condsim3,nx,ny,1) # extract realization #1 to a 2D array and plot\r\nimage.plot(real1,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #1\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nreal2 <- sim2darray(condsim3,nx,ny,2) # extract realization #2 to a 2D array and plot\r\nimage.plot(real2,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #2\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nreal3 <- sim2darray(condsim3,nx,ny,3) # extract realization #3 to a 2D array and plot\r\nimage.plot(real3,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Realization #3\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\nest <- sim2darray(krige3,nx,ny,1) # extract realization #4 to a 2D array and plot\r\nimage.plot(est,x=x,y=y,xlab=\"X(m)\",ylab=\"Y(m)\",zlim = zlim,col=colmap,legend.shrink = 0.6); mtext(line=1, side=3, \"Kriging\", outer=F);box(which=\"plot\")\r\npoints(mydata$X,mydata$Y,pch=\"+\",cex=1.0,col=\"black\") # add well locations\r\n\r\npar(mfrow=c(3,2))\r\n\r\n# Minimum acceptance checks, check the distributions and variagrams of the realization\r\n\r\n# Variogram 1\r\n# Plot the CDFs for the 4 realizations and compare to the raw data CDF\r\nplot(ecdf(condsim1@data[,1]),main=\"Variogram 1\",xlab=\"Gaussian Values\",ylab=\"Cumulative Probability\",col=\"red\")\r\nplot(ecdf(condsim1@data[,2]),add=TRUE,col=\"red\")\r\nplot(ecdf(condsim1@data[,3]),add=TRUE,col=\"red\")\r\nplot(ecdf(condsim1@data[,4]),add=TRUE,col=\"red\")\r\nplot(ecdf(mydata$porosity),add=TRUE,col=\"black\")\r\n\r\n# Calculate the varograms for the major and minor directions and compare to the variogram model\r\n# We just arbitrarily picked 035, 125 azimuth, model is isotropic so direction shouldn't matter\r\nvg.sim1.035 = variogram(sim1~1,condsim1,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim1.125 = variogram(sim1~1,condsim1,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\nvg.sim2.035 = variogram(sim2~1,condsim1,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim2.125 = variogram(sim2~1,condsim1,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\nvg.sim3.035 = variogram(sim3~1,condsim1,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim3.125 = variogram(sim3~1,condsim1,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\nvg.sim4.035 = variogram(sim4~1,condsim1,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim4.125 = variogram(sim4~1,condsim1,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\n\r\n# Plot the vairograms from the realizations and the model variogram.\r\nplot(vg.sim1.035$dist,vg.sim1.035$gamma,pch=19,cex=0.1,main=\"Variogram 1\",xlab=\" Lag Distance (m) \",ylab=\" Semivariogram \", col=\"black\",xlim=c(0,1000),ylim=c(0,1.2*sill))\r\npoints(vg.sim1.125$dist,vg.sim1.125$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim2.035$dist,vg.sim2.035$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim2.125$dist,vg.sim2.125$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim3.035$dist,vg.sim3.035$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim3.125$dist,vg.sim3.125$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim4.035$dist,vg.sim4.035$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim4.125$dist,vg.sim4.125$gamma,pch=19,cex=0.1,col=\"red\")\r\nabline(h = 1.0*sill)\r\n# Include variogram model\r\nunit_vector = c(0,1,0) # unit vector for 000 azimuth\r\nvm1.000 <- variogramLine(vm1,maxdist=1000,min=0.0001,n=100,dir=unit_vector,covariance=FALSE) # calculate 035 variogram model\r\nlines(vm1.000$dist,vm1.000$gamma,pch=19,cex=0.1,col=\"black\") # include variogram model \r\nunit_vector = c(1,0,0) # unit vector for 090 azimuth\r\nvm1.090 <- variogramLine(vm1,maxdist=1000,min=0.0001,n=100,dir=unit_vector,covariance=FALSE) # calculate 125 variogram model\r\nlines(vm1.090$dist,vm1.090$gamma,col=\"red\") \r\n\r\n# Repeat the above checks for the 2 other sets of simulated realizations\r\n# Variogram 2\r\n# Plot the CDFs for the 4 realizations and compare to the raw data CDF\r\nplot(ecdf(condsim2@data[,1]),main=\"Variogram 2\",xlab=\"Gaussian Values\",ylab=\"Cumulative Probability\",col=\"red\")\r\nplot(ecdf(condsim2@data[,2]),add=TRUE,col=\"red\")\r\nplot(ecdf(condsim2@data[,3]),add=TRUE,col=\"red\")\r\nplot(ecdf(condsim2@data[,4]),add=TRUE,col=\"red\")\r\nplot(ecdf(mydata$porosity),add=TRUE,col=\"black\")\r\n\r\nvg.sim1.035 = variogram(sim1~1,condsim2,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim1.125 = variogram(sim1~1,condsim2,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\nvg.sim2.035 = variogram(sim2~1,condsim2,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim2.125 = variogram(sim2~1,condsim2,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\nvg.sim3.035 = variogram(sim3~1,condsim2,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim3.125 = variogram(sim3~1,condsim2,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\nvg.sim4.035 = variogram(sim4~1,condsim2,cutoff = 1000,width =20,alpha = 35.0,tol.hor=22.5) \r\nvg.sim4.125 = variogram(sim4~1,condsim2,cutoff = 1000,width =20,alpha = 125.0,tol.hor=22.5) \r\n\r\nplot(vg.sim1.035$dist,vg.sim1.035$gamma,pch=19,cex=0.1,main=\"Variogram 2\",xlab=\" Lag Distance (m) \",ylab=\" Semivariogram \", col=\"black\",xlim=c(0,1000),ylim=c(0,1.2*sill))\r\npoints(vg.sim1.125$dist,vg.sim1.125$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim2.035$dist,vg.sim2.035$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim2.125$dist,vg.sim2.125$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim3.035$dist,vg.sim3.035$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim3.125$dist,vg.sim3.125$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim4.035$dist,vg.sim4.035$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim4.125$dist,vg.sim4.125$gamma,pch=19,cex=0.1,col=\"red\")\r\nabline(h = 1.0*sill)\r\n# Include variogram model\r\nunit_vector = c(sin(35*pi/180),cos(35*pi/180),0) # unit vector for 035 azimuth\r\nvm2.035 <- variogramLine(vm2,maxdist=1000,min=0.0001,n=100,dir=unit_vector,covariance=FALSE) # calculate 035 variogram model\r\nlines(vm2.035$dist,vm2.035$gamma,pch=19,cex=0.1,col=\"black\") # include variogram model \r\n\r\nunit_vector = c(sin(55*pi/180),-1*cos(35*pi/180),0) # unit vector for 125 azimuth\r\nvm2.125 <- variogramLine(vm2,maxdist=1000,min=0.0001,n=100,dir=unit_vector,covariance=FALSE) # calculate 125 variogram model\r\nlines(vm2.125$dist,vm2.125$gamma,col=\"red\") # include variogram model\r\n\r\n# Variogram 3\r\n# Plot the CDFs for the 4 realizations and compare to the raw data CDF\r\nplot(ecdf(condsim3@data[,1]),main=\"Variogram 3\",xlab=\"Gaussian Values\",ylab=\"Cumulative Probability\",col=\"red\")\r\nplot(ecdf(condsim3@data[,2]),add=TRUE,col=\"red\")\r\nplot(ecdf(condsim3@data[,3]),add=TRUE,col=\"red\")\r\nplot(ecdf(condsim3@data[,4]),add=TRUE,col=\"red\")\r\nplot(ecdf(mydata$porosity),add=TRUE,col=\"black\")\r\n\r\nc = sqrt(2)\r\n\r\nvg.sim1.060 = variogram(sim1~1,condsim3,cutoff = 1000,width =20,alpha = 60.0,tol.hor=22.5) \r\nvg.sim1.150 = variogram(sim1~1,condsim3,cutoff = 1000,width =20,alpha = 150.0,tol.hor=22.5) \r\nvg.sim2.060 = variogram(sim2~1,condsim3,cutoff = 1000,width =20,alpha = 60.0,tol.hor=22.5) \r\nvg.sim2.150 = variogram(sim2~1,condsim3,cutoff = 1000,width =20,alpha = 150.0,tol.hor=22.5) \r\nvg.sim3.060 = variogram(sim3~1,condsim3,cutoff = 1000,width =20,alpha = 60.0,tol.hor=22.5) \r\nvg.sim3.150 = variogram(sim3~1,condsim3,cutoff = 1000,width =20,alpha = 150.0,tol.hor=22.5) \r\nvg.sim4.060 = variogram(sim4~1,condsim3,cutoff = 1000,width =20,alpha = 60.0,tol.hor=22.5) \r\nvg.sim4.150 = variogram(sim4~1,condsim3,cutoff = 1000,width =20,alpha = 150.0,tol.hor=22.5) \r\n\r\nplot(vg.sim1.060$dist,vg.sim1.060$gamma,pch=19,cex=0.1,main=\"Varogram 3\",xlab=\" Lag Distance (m) \",ylab=\" Semivariogram \", col=\"black\",xlim=c(0,1000),ylim=c(0,1.2*sill))\r\npoints(vg.sim1.150$dist,vg.sim1.150$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim2.060$dist,vg.sim2.060$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim2.150$dist,vg.sim2.150$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim3.060$dist,vg.sim3.060$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim3.150$dist,vg.sim3.150$gamma,pch=19,cex=0.1,col=\"red\")\r\npoints(vg.sim4.060$dist,vg.sim4.060$gamma,pch=19,cex=0.1,col=\"black\")\r\npoints(vg.sim4.150$dist,vg.sim4.150$gamma,pch=19,cex=0.1,col=\"red\")\r\nabline(h = 1.0*sill)\r\n# Include variogram model\r\nunit_vector = c(sin(60*pi/180),cos(60*pi/180),0) # unit vector for 060 azimuth\r\nvm3.060 <- variogramLine(vm3,maxdist=1000,min=0.0001,n=100,dir=unit_vector,covariance=FALSE) # calculate 035 variogram model\r\nlines(vm3.060$dist,vm3.060$gamma,pch=19,cex=0.1,col=\"black\") # include variogram model \r\n\r\nunit_vector = c(sin(30*pi/180),-1*cos(60*pi/180),0) # unit vector for 125 azimuth\r\nvm3.150 <- variogramLine(vm3,maxdist=1000,min=0.0001,n=100,dir=unit_vector,covariance=FALSE) # calculate 125 variogram model\r\nlines(vm3.150$dist,vm3.150$gamma,col=\"red\") # include variogram model\r\n\r\n# On your own try changing the variogram parameters and observe the results. Also consider kriging with a trend.\r\n\r\n# Hope this was helpful,\r\n\r\n# Michael\r\n\r\n# Appreciation to Edzer Pedesma for the gstat package.\r\n", "meta": {"hexsha": "0acf8cac60661250a0ef31f01368a7f762e76fc7", "size": 18006, "ext": "r", "lang": "R", "max_stars_repo_path": "simulation_demo.r", "max_stars_repo_name": "GeostatsGuy/2DayCourse_Exercises", "max_stars_repo_head_hexsha": "f2e1131c269d7c4e14906a0f362fead10b9a702f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 11, "max_stars_repo_stars_event_min_datetime": "2018-07-24T17:54:15.000Z", "max_stars_repo_stars_event_max_datetime": "2019-09-09T06:15:03.000Z", "max_issues_repo_path": "simulation_demo.r", "max_issues_repo_name": "GeostatsGuy/2DayCourse_Exercises", "max_issues_repo_head_hexsha": "f2e1131c269d7c4e14906a0f362fead10b9a702f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "simulation_demo.r", "max_forks_repo_name": "GeostatsGuy/2DayCourse_Exercises", "max_forks_repo_head_hexsha": "f2e1131c269d7c4e14906a0f362fead10b9a702f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 9, "max_forks_repo_forks_event_min_datetime": "2019-02-15T23:40:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-04-25T19:12:37.000Z", "avg_line_length": 56.9810126582, "max_line_length": 172, "alphanum_fraction": 0.6766077974, "num_tokens": 6311, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7490872131147275, "lm_q2_score": 0.6825737408694988, "lm_q1q2_score": 0.511307261293227}} {"text": "num = as.integer(readline(prompt=\"Enter int: \"))\nfactorial = 1\nif(num < 0) {\nprint(\"Error: must be +\")\n} else if(num == 0) {\nprint(\"1\")\n} else {\nfor(i in 1:num) {\nfactorial = factorial * i\n}\nprint(paste(factorial))\n}", "meta": {"hexsha": "7d3b7e000b73a116e65c40d82d70222f9fd390f4", "size": 216, "ext": "r", "lang": "R", "max_stars_repo_path": "public/code/R/factorial_of_a_number.r", "max_stars_repo_name": "encap/coderush", "max_stars_repo_head_hexsha": "e55f7d6d57465b93c7c376d90f22b1e79e04fc2b", "max_stars_repo_licenses": ["Unlicense"], "max_stars_count": 45, "max_stars_repo_stars_event_min_datetime": "2020-10-17T22:28:52.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-28T12:36:40.000Z", "max_issues_repo_path": "public/code/R/factorial_of_a_number.r", "max_issues_repo_name": "encap/coderush", "max_issues_repo_head_hexsha": "e55f7d6d57465b93c7c376d90f22b1e79e04fc2b", "max_issues_repo_licenses": ["Unlicense"], "max_issues_count": 33, "max_issues_repo_issues_event_min_datetime": "2020-11-27T15:12:36.000Z", "max_issues_repo_issues_event_max_datetime": "2022-02-27T10:24:40.000Z", "max_forks_repo_path": "public/code/R/factorial_of_a_number.r", "max_forks_repo_name": "encap/coderush", "max_forks_repo_head_hexsha": "e55f7d6d57465b93c7c376d90f22b1e79e04fc2b", "max_forks_repo_licenses": ["Unlicense"], "max_forks_count": 5, "max_forks_repo_forks_event_min_datetime": "2020-11-27T00:29:29.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-21T15:53:04.000Z", "avg_line_length": 18.0, "max_line_length": 48, "alphanum_fraction": 0.6203703704, "num_tokens": 72, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6825737344123242, "lm_q2_score": 0.7490872187162397, "lm_q1q2_score": 0.5113072602796852}} {"text": "# OBRESTNE KRIVULJE\n# Finančni praktikum 2020/21\n# Mitja Mandić\n\n#prva naloga\n\neuribor.15 <- read.table(\"prva_naloga/podatki/hist_EURIBOR_2015.csv\", sep = \",\", header = TRUE, row.names = 1, check.names = FALSE)\neuribor.15 <- as.data.frame(t(euribor.15[,c(1, 22, 42, 64, 84, 104, 126, 149, 170, 192, 214, 235)]))\neuribor.16 <- read.table(\"prva_naloga/podatki/hist_EURIBOR_2016.csv\", sep = \",\", header = TRUE, row.names = 1, check.names = FALSE)\neuribor.16 <- as.data.frame(t(euribor.16[,c(1, 21, 42, 63, 84, 106, 128, 149, 172, 194, 215, 237)]))\neuribor.17 <- read.table(\"prva_naloga/podatki/hist_EURIBOR_2017.csv\", sep = \",\", header = TRUE, row.names = 1, check.names = FALSE)\neuribor.17 <- as.data.frame(t(euribor.17[,c(1, 23, 43, 66, 84, 106, 128, 149, 172, 193, 215, 237)]))\n\neuribor <- rbind(euribor.15,euribor.16,euribor.17)\n\nprva.c.6m <- ts(euribor[,\"6m\"], start = 2015, frequency = 12)\nprva.c.12m <- ts(euribor[,\"12m\"], start = 2015, frequency = 12)\n\nts.plot(prva.c.6m,prva.c.12m, gpars = list(xlab = \"Datum\", ylab = \"%\", col = c(\"red\", \"blue\")))\nlegend(\"topright\", bty=\"n\", lty=c(1,1), col=c(\"red\",\"blue\"),\n legend=c(\"6m\", \"12m\"))\ntitle(\"EURIBOR\") \ngraf.ts <- recordPlot()\n#druga naloga\n\n#1. julij 2015, 3. oktober 2016, 1. september 2017\n\ndruga.b <- euribor[c(\"01/07/2015\", \"03/10/2016\", \"01/09/2017\"),]\n#druga.b <- as.data.frame(t(druga.b))\n\n plot(t(druga.b[1,]), ylim = c(-0.4, 0.4), col = c(\"red\"),x=c(1/4,1/2,1,2,3,6,9,12), pch = 16, ylab = \"%\", xlab = \"Časovna enota\")\npoints(t(druga.b[2,]), col = c(\"blue\"),pch = 16,x=c(1/4,1/2,1,2,3,6,9,12))\npoints(t(druga.b[3,]), col = c(\"black\"),pch = 16,x=c(1/4,1/2,1,2,3,6,9,12))\nlines(t(druga.b[1,]),col = c(\"red\"),pch = 16,x=c(1/4,1/2,1,2,3,6,9,12))\nlines(t(druga.b[2,]),col = c(\"blue\"),pch = 16,x=c(1/4,1/2,1,2,3,6,9,12))\nlines(t(druga.b[3,]),col = c(\"black\"),pch = 16,x=c(1/4,1/2,1,2,3,6,9,12))\nlegend(\"topleft\",bty=\"n\",lty = c(1,1),col=c(\"red\", \"blue\", \"black\"), legend=c(\"1. julij 2015\", \"3. oktober 2016\", \"1. september 2017\"))\ntitle(\"Casovna struktura Euribor\")\ngraf.druga.b <- recordPlot()\n#Vse krivulje so naraščajoče, torej imajo višjo donosnost pri kasnejših datumih zapadlosti, torej imajo normalno obliko.\n#So precej položne in z izjemo leta 2015 skozi negativne.\n\n\n\n#tretja naloga\nT <- 6\nU <- 12\n\neuribor.tretja <- euribor[,c(\"6m\",\"12m\")]\neuribor.tretja[,\"terminska\"] <- 2 * ((1 + euribor.tretja[,\"12m\"]/100)/(1 + 1/2 * euribor.tretja[,\"6m\"]/100) - 1)*100 #račun terminske obrestne mere\n\nnapoved <- euribor.tretja[,\"terminska\"]\nnapoved[c(31,32,33,34,35,36)] <- 0\nnapoved[c(seq(7,36))] <- napoved[c(seq(1,30))]\n\nnapoved[c(1,2,3,4,5,6)] <- NA\neuribor.tretja[\"napoved\"] <- napoved #Dodan stolpec z napovedanimi vrednostmi.\n\n\n plot(x=euribor.tretja[seq(6,12),\"napoved\"],y=euribor.tretja[seq(6,12),\"6m\"], xlim = c(-0.3,0.5), ylim = c(-0.3,0.5), type=\"p\",\n xlab = \"napoved\", ylab = \"opazovano\",col = \"blue\")\npoints(x=euribor.tretja[seq(13,24),\"napoved\"],y=euribor.tretja[seq(13,24),\"6m\"], col = \"red\")\npoints(x=euribor.tretja[seq(25,36),\"napoved\"],y=euribor.tretja[seq(25,36),\"6m\"], col = \"green\")\nlines(x = c(-0.5,0,5), y = c(-0.5,0,5), col = \"grey\")\nlegend(\"topleft\",bty=\"n\",lty = c(1,1),col=c(\"blue\", \"red\", \"green\"), legend=c(\"2015\", \"2016\", \"2017\"))\nabline(lm(euribor.tretja[,\"6m\"]~euribor.tretja[,\"napoved\"]))\ntitle(\"6m euribor 2015-2017\") \ngraf.3c <- recordPlot()\n\nplot(x=euribor.tretja[seq(6,12),\"napoved\"],y=euribor.tretja[seq(6,12),\"6m\"], xlim = c(-0.1,0.5), ylim = c(-0.1,0.5), type=\"p\",\n xlab = \"napoved\", ylab = \"opazovano\",col = \"blue\")\nabline(lm(euribor.tretja[seq(6,12),\"6m\"]~euribor.tretja[seq(6,12),\"napoved\"]))\nlines(x = c(-0.5,0,5), y = c(-0.5,0,5), col = \"grey\")\ntitle(\"6m Euribor 2015\")\ngraf.3d.prva <- recordPlot()\n\nplot(x=euribor.tretja[seq(13,24),\"napoved\"],y=euribor.tretja[seq(13,24),\"6m\"], xlim = c(-0.3,0.3), ylim = c(-0.3,0.3), type=\"p\",\n xlab = \"napoved\", ylab = \"opazovano\",col = \"red\")\nabline(lm(euribor.tretja[seq(13,24),\"6m\"]~euribor.tretja[seq(13,24),\"napoved\"]))\nlines(x = c(-0.5,0,5), y = c(-0.5,0,5), col = \"grey\")\ntitle(\"6m Euribor 2016\")\ngraf.3d.druga <- recordPlot()\n\nplot(x=euribor.tretja[seq(25,36),\"napoved\"],y=euribor.tretja[seq(25,36),\"6m\"], xlim = c(-0.3,0.1), ylim = c(-0.3,0.1), type=\"p\",\n xlab = \"napoved\", ylab = \"opazovano\",col = \"green\")\nabline(lm(euribor.tretja[seq(25,36),\"6m\"]~euribor.tretja[seq(25,36),\"napoved\"]))\nlines(x = c(-0.5,0,5), y = c(-0.5,0,5), col = \"grey\")\ntitle(\"6m Euribor 2017\")\ngraf.3d.tretja <- recordPlot()\n\n#odgovor 3.e\n#Regresijska premica bi morala biti simetrala lihih kvadrantov, na kateri bi ležale vse narisane točke. Ker je napoved večja od\n#izmerjene vrednosti, so točke na grafih v desnem spodnjem kotu.\n\n#link na github repo: https://github.com/mitja-mandic/financni-praktikum", "meta": {"hexsha": "846c7a093bfe09248a69439a65c7fd3e3a4e8ad4", "size": 4846, "ext": "r", "lang": "R", "max_stars_repo_path": "prva_naloga/Mandic1.r", "max_stars_repo_name": "mitja-mandic/financni-praktikum", "max_stars_repo_head_hexsha": "fe7ececd401b7a002b9e11bd69938ba98a96a081", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "prva_naloga/Mandic1.r", "max_issues_repo_name": "mitja-mandic/financni-praktikum", "max_issues_repo_head_hexsha": "fe7ececd401b7a002b9e11bd69938ba98a96a081", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "prva_naloga/Mandic1.r", "max_forks_repo_name": "mitja-mandic/financni-praktikum", "max_forks_repo_head_hexsha": "fe7ececd401b7a002b9e11bd69938ba98a96a081", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 51.0105263158, "max_line_length": 147, "alphanum_fraction": 0.6223689641, "num_tokens": 2171, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.749087201911703, "lm_q2_score": 0.6825737214979745, "lm_q1q2_score": 0.5113072391353757}} {"text": "# The posterior conditional function utilized by the\n# Gibbs sampler in function.r. They have all the\n# same form maybe they should be implemented in a\n# smarter way.\n\n# The basic structure is the following\n\n# 1. Get the old parameter\n# 2. Propose a jump\n# 3. Accept it using the MH ratio\n# 4. Return the old or new value based on the MH ratio\n\nupdateTheta <- function(cp) {\n theta_star <- proposeTheta(cp$Theta)\n if (theta_star < 0) {\n return(cp)\n }\n\n old_lik <- cp$old_lik + priorTheta(cp$Theta)\n theta_star_mat <- getPmat(theta_star, cp$Rho, cp$acgt)\n new_lik_func <- logLikAll(cp$dat, theta_star_mat, cp$DeltaD, cp$DeltaS, cp$laVec, cp$nuVec, cp$m)\n new_lik <- new_lik_func + priorTheta(theta_star)\n\n if (metroDesc(new_lik, old_lik)) {\n cp$Theta <- theta_star\n cp$ThetaMat <- theta_star_mat\n cp$old_lik <- new_lik_func\n }\n\n return(cp)\n}\n\n\nupdateRho <- function(cp) {\n rho_star <- proposeRho(cp$Rho)\n if (rho_star <= 0) {\n return(cp)\n }\n\n old_lik <- cp$old_lik + priorRho(cp$Rho)\n rho_star_mat <- getPmat(cp$Theta, rho_star, cp$acgt)\n new_lik_func <- logLikAll(cp$dat, rho_star_mat, cp$DeltaD, cp$DeltaS, cp$laVec, cp$nuVec, cp$m)\n new_lik <- new_lik_func + priorRho(rho_star)\n\n if (metroDesc(new_lik, old_lik)) {\n cp$Rho <- rho_star\n cp$ThetaMat <- rho_star_mat\n cp$old_lik <- new_lik_func\n }\n\n return(cp)\n}\n\nupdateDeltaD <- function(cp) {\n deltad_star <- proposeDeltaD(cp$DeltaD)\n if (deltad_star < 0 || deltad_star > 1) {\n return(cp)\n }\n\n old_lik <- cp$old_lik + priorDeltaD(cp$DeltaD)\n new_lik_func <- logLikAll(cp$dat, cp$ThetaMat, deltad_star, cp$DeltaS, cp$laVec, cp$nuVec, cp$m)\n new_lik <- new_lik_func + priorDeltaD(deltad_star)\n\n if (metroDesc(new_lik, old_lik)) {\n cp$DeltaD <- deltad_star\n cp$old_lik <- new_lik_func\n }\n\n return(cp)\n}\n\nupdateDeltaS <- function(cp) {\n deltas_star <- proposeDeltaS(cp$DeltaS)\n if (deltas_star < 0 || deltas_star > 1) {\n return(cp)\n }\n\n old_lik <- cp$old_lik + priorDeltaS(cp$DeltaS)\n new_lik_func <- logLikAll(cp$dat, cp$ThetaMat, cp$DeltaD, deltas_star, cp$laVec, cp$nuVec, cp$m)\n new_lik <- new_lik_func + priorDeltaS(deltas_star)\n\n if (metroDesc(new_lik, old_lik)) {\n cp$DeltaS <- deltas_star\n cp$old_lik <- new_lik_func\n }\n\n return(cp)\n}\n\n\nupdateLambda <- function(cp) {\n lambda_star <- proposeLambda(cp$Lambda)\n if (lambda_star < 0 || lambda_star > 1) {\n return(cp)\n }\n\n old_lik <- cp$old_lik + priorLambda(cp$Lambda)\n laVecStarLeft <- seqProbVecLambda(lambda_star, cp$LambdaDisp, cp$m, cp$termini)\n if (cp$same_overhangs) {\n # It left and right are the same\n laVecStar <- laVecStarLeft\n } else {\n # The left and right overhangs are not the same\n laVecStar <- c(laVecStarLeft[1:(cp$m / 2)], cp$laVecRight[(cp$m / 2 + 1):cp$m])\n }\n\n new_lik_func <- logLikAll(cp$dat, cp$ThetaMat, cp$DeltaD, cp$DeltaS, laVecStar, cp$nuVec, cp$m)\n new_lik <- new_lik_func + priorLambda(lambda_star)\n\n if (metroDesc(new_lik, old_lik)) {\n cp$Lambda <- lambda_star\n cp$laVec <- laVecStar\n cp$old_lik <- new_lik_func\n }\n\n return(cp)\n}\n\n\nupdateLambdaRight <- function(cp) {\n stopifnot(!cp$same_overhangs, cp$termini == \"both\")\n\n lambda_right_star <- proposeLambdaRight(cp$LambdaRight)\n if (lambda_right_star < 0 || lambda_right_star > 1) {\n return(cp)\n }\n\n old_lik <- cp$old_lik + priorLambdaRight(cp$LambdaRight)\n laVecStarRight <- seqProbVecLambda(lambda_right_star, cp$LambdaDisp, cp$m, cp$termini)\n # The left and right overhangs are not the same\n laVecStar <- c(cp$laVec[1:(cp$m / 2)], laVecStarRight[(cp$m / 2 + 1):cp$m])\n new_lik_func <- logLikAll(cp$dat, cp$ThetaMat, cp$DeltaD, cp$DeltaS, laVecStar, cp$nuVec, cp$m)\n new_lik <- new_lik_func + priorLambdaRight(lambda_right_star)\n\n if (metroDesc(new_lik, old_lik)) {\n cp$LambdaRight <- lambda_right_star\n cp$laVecRight <- laVecStar\n cp$old_lik <- new_lik_func\n }\n\n return(cp)\n}\n\n\nupdateLambdaDisp <- function(cp) {\n lambda_disp_star <- proposeLambdaDisp(cp$LambdaDisp)\n if (lambda_disp_star < 0) {\n return(cp)\n }\n\n old_lik <- cp$old_lik + priorLambdaDisp(cp$LambdaDisp)\n if (cp$same_overhangs) {\n laVecStar <- seqProbVecLambda(cp$Lambda, lambda_disp_star, cp$m, cp$termini)\n } else {\n leftLaVecStar <- seqProbVecLambda(cp$Lambda, lambda_disp_star, cp$m, cp$termini)\n rightLaVecStar <- seqProbVecLambda(cp$LambdaRight, lambda_disp_star, cp$m, cp$termini)\n laVecStar <- c(leftLaVecStar[1:(cp$m / 2)], rightLaVecStar[(cp$m / 2 + 1):cp$m])\n }\n new_lik_func <- logLikAll(cp$dat, cp$ThetaMat, cp$DeltaD, cp$DeltaS, laVecStar, cp$nuVec, cp$m)\n new_lik <- new_lik_func + priorLambdaDisp(lambda_disp_star)\n\n if (metroDesc(new_lik, old_lik)) {\n cp$LambdaDisp <- lambda_disp_star\n cp$laVec <- laVecStar\n cp$old_lik <- new_lik_func\n }\n\n return(cp)\n}\n", "meta": {"hexsha": "4fd6302c3ff7375fcc581e5013a5afc175643483", "size": 4803, "ext": "r", "lang": "R", "max_stars_repo_path": "mapdamage/r/stats/postConditonal.r", "max_stars_repo_name": "ginolhac/mapDamage", "max_stars_repo_head_hexsha": "036806b434945594c2e642d03461c64e981507de", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 32, "max_stars_repo_stars_event_min_datetime": "2015-03-11T21:29:32.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-25T16:01:52.000Z", "max_issues_repo_path": "mapdamage/r/stats/postConditonal.r", "max_issues_repo_name": "ginolhac/mapDamage", "max_issues_repo_head_hexsha": "036806b434945594c2e642d03461c64e981507de", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 39, "max_issues_repo_issues_event_min_datetime": "2015-02-06T23:42:18.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-09T12:33:35.000Z", "max_forks_repo_path": "mapdamage/r/stats/postConditonal.r", "max_forks_repo_name": "ginolhac/mapDamage", "max_forks_repo_head_hexsha": "036806b434945594c2e642d03461c64e981507de", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 6, "max_forks_repo_forks_event_min_datetime": "2016-07-04T11:04:56.000Z", "max_forks_repo_forks_event_max_datetime": "2021-12-20T21:16:47.000Z", "avg_line_length": 28.2529411765, "max_line_length": 99, "alphanum_fraction": 0.6885280033, "num_tokens": 1621, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8354835207180245, "lm_q2_score": 0.6113819732941511, "lm_q1q2_score": 0.5107995635513306}} {"text": "library(igraph)\n# infect node by id\ninfectIDCont <- function(graph, idVector) {\n for (i in idVector) {\n # infect nodes in vector\n V(graph)[i]$currentState <- 1\n V(graph)[i]$wasInfected <- TRUE\n }\n return(graph)\n}\n\n# randomly infect nodes\nrandomlyInfectCont <- function(graph, number) {\n nodes <- as.vector(sample(V(graph), number, replace = FALSE))\n graph <- infectIDCont(graph, nodes)\n return(graph)\n}\n\n# run the infection\nrunContSIR <- function(graph, tau) {\n graph <- set_vertex_attr(graph, \"tau\", value = tau)\n graph <- set_vertex_attr(graph, \"currentState\", value = 0)\n graph <- set_vertex_attr(graph, \"wasInfected\", value = FALSE)\n graph <- set_vertex_attr(graph, \"Sduration\", value = 0)\n graph <- set_vertex_attr(graph, \"SIduration\", value = 0)\n graph <- randomlyInfectCont(graph, 1)\n \n # start with t=0\n t <- 0\n \n # while there are infected nodes\n while (length(V(graph)[V(graph)$currentState == 1]) > 0) {\n # neighbourList consists of nodes which may be infected in this turn\n # nodes occur multiple times if they share multiple edges with infected nodes\n neighbourList <-\n unlist(adjacent_vertices(graph, V(graph)[V(graph)$currentState == 1]), FALSE, FALSE)\n neighbourList <-\n neighbourList[!V(graph)[neighbourList]$wasInfected]\n \n # TAU is total infection rate\n TAU <- length(neighbourList) * tau\n GAMMA <- length(V(graph)[V(graph)$currentState == 1])\n rate <- TAU + GAMMA\n deltaT <- rexp(1, rate)\n t <- t + deltaT\n \n # infection is true if next event is infection, false if it is recovery\n infection <- sample(c(T, F), 1, prob = c(TAU / rate, GAMMA / rate))\n \n if (infection) {\n newInfected <- sample(neighbourList, 1)\n V(graph)[newInfected]$currentState <- 1\n V(graph)[newInfected]$Sduration <- round(t, 6)\n V(graph)[newInfected]$wasInfected <- TRUE\n }\n \n else{\n newRecovered <- sample(V(graph)[V(graph)$currentState == 1], 1)\n V(graph)[newRecovered]$currentState <- 2\n V(graph)[newRecovered]$SIduration <- round(t, 6)\n }\n \n }\n \n V(graph)$SIduration[V(graph)$currentState == 0] <- round(t, 6)\n V(graph)$Sduration[V(graph)$currentState == 0] <- round(t, 6)\n \n # Set some additional simulation properties\n graph <- set_vertex_attr(graph, \"maxTime\", value = round(t, 6))\n percentageInfected <-\n round(100 * sum(V(graph)$wasInfected) / vcount(graph), digits = 2)\n graph <-\n set_vertex_attr(graph, \"percentageInfected\", value = percentageInfected)\n \n \n return(graph)\n}\n", "meta": {"hexsha": "a14c811d7573c8093b87c964119cfc604f660099", "size": 2530, "ext": "r", "lang": "R", "max_stars_repo_path": "infection.r", "max_stars_repo_name": "J-JJJJJJ/sentinelClassification", "max_stars_repo_head_hexsha": "6b6cffdeec6a00303542612bf1227b29189cd281", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "infection.r", "max_issues_repo_name": "J-JJJJJJ/sentinelClassification", "max_issues_repo_head_hexsha": "6b6cffdeec6a00303542612bf1227b29189cd281", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "infection.r", "max_forks_repo_name": "J-JJJJJJ/sentinelClassification", "max_forks_repo_head_hexsha": "6b6cffdeec6a00303542612bf1227b29189cd281", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 32.4358974359, "max_line_length": 90, "alphanum_fraction": 0.6553359684, "num_tokens": 709, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7634837743174788, "lm_q2_score": 0.6688802537704063, "lm_q1q2_score": 0.5106792207150628}} {"text": "collatz <- function(n)\n{\n c <- 0\n while (n != 1)\n {\n if (n %% 2 == 0)\n {\n n = n / 2\n }\n else\n {\n n = n * 3 + 1\n }\n c = c + 1\n }\n return(c)\n}\n\nmain <- function()\n{\n f <- 0\n\n for (j in 0:100)\n {\n for (i in 1:100000)\n {\n f = f + collatz(i)\n }\n }\n\n print(f)\n}\n\nmain()\n", "meta": {"hexsha": "57e0ae73b488e840e9683d4d3e8787fe5387065f", "size": 398, "ext": "r", "lang": "R", "max_stars_repo_path": "src/R/collatz/col.r", "max_stars_repo_name": "vonNiklasson/KTH-programming-benchmark", "max_stars_repo_head_hexsha": "c08a43d11278bb34dac92ecae0f99979e9bfd29c", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/R/collatz/col.r", "max_issues_repo_name": "vonNiklasson/KTH-programming-benchmark", "max_issues_repo_head_hexsha": "c08a43d11278bb34dac92ecae0f99979e9bfd29c", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/R/collatz/col.r", "max_forks_repo_name": "vonNiklasson/KTH-programming-benchmark", "max_forks_repo_head_hexsha": "c08a43d11278bb34dac92ecae0f99979e9bfd29c", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 11.3714285714, "max_line_length": 30, "alphanum_fraction": 0.2713567839, "num_tokens": 136, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.7154239957834733, "lm_q2_score": 0.7122321842389469, "lm_q1q2_score": 0.5095479951738182}} {"text": "\\name{Newman}\n\\alias{Newman}\n%- Also NEED an '\\alias' for EACH other topic documented here.\n\\title{\n Corrected Newman's method for estimating the preferential attachment function\n}\n\\description{\nThis function implements a correction proposed in [1] of the original Newman's method in [2] to estimate the preferential attachment function. \n}\n\\usage{\n Newman(net_object , \n net_stat = get_statistics(net_object), \n start = 1 , \n interpolate = FALSE)\n}\n%- maybe also 'usage' for other objects documented here.\n\\arguments{\n \\item{net_object}{\n an object of class \\code{PAFit_net} that contains the network.\n}\n \\item{net_stat}{\n An object of class \\code{PAFit_data} which contains summerized statistics needed in estimation. This object is created by the function \\code{\\link{get_statistics}}. Default value is \\code{ get_statistics(net_object)}.\n }\n \\item{start}{Positive integer. The starting time from which the method is applied. Default value is \\eqn{1}.}\n \\item{interpolate}{\n Logical. If \\code{TRUE} then all the gaps in the estimated PA function are interpolated by linear interpolating in logarithm scale. Default value is \\code{FALSE}.\n }\n}\n\\value{\n Outputs an \\code{PA_result} object which contains the estimated attachment function. In particular, it contains the following field:\n \\itemize{\n \\item \\code{k} and \\code{A}: a degree vector and the estimated PA function.\n \n \\item \\code{center_k} and \\code{theta}: when we perform binning, these are the centers of the bins and the estimated PA values for those bins. \n \\item \\code{g}: the number of bins used.\n \\item \\code{alpha} and \\code{ci}: \\code{alpha} is the estimated attachment exponenet \\eqn{\\alpha} (when assume \\eqn{A_k = k^\\alpha}), while \\code{ci} is the mean plus/minus two-standard-deviation interval.\n \\item \\code{loglinear_fit}: this is the fitting result when we estimate \\eqn{\\alpha}. \n}\n}\n\\author{\n Thong Pham \\email{thongpham@thongpham.net}\n}\n\\references{\n 1. Pham, T., Sheridan, P. & Shimodaira, H. (2015). PAFit: A Statistical Method for Measuring Preferential Attachment in Temporal Complex Networks. PLoS ONE 10(9): e0137796. doi:10.1371/journal.pone.0137796 (\\url{http://dx.doi.org/10.1371/journal.pone.0137796}).\n \n 2. Newman, M.. Clustering and preferential attachment in growing networks. Physical Review E. 2001;64(2):025102 (\\url{https://journals.aps.org/pre/abstract/10.1103/PhysRevE.64.025102}).\n}\n\\seealso{\n\nSee \\code{\\link{get_statistics}} for how to create summerized statistics needed in this function.\n \nSee \\code{\\link{Jeong}}, \\code{\\link{only_A_estimate}} for other methods to estimate the attachment function in isolation.\n}\n\\examples{\n library(\"PAFit\")\n net <- generate_net(N = 1000 , m = 1 , mode = 1 , alpha = 1 , s = 0)\n net_stats <- get_statistics(net)\n result <- Newman(net, net_stats)\n summary(result)\n # true function\n true_A <- result$center_k\n #plot the estimated attachment function\n plot(result , net_stats)\n lines(result$center_k, true_A, col = \"red\") # true line\n legend(\"topleft\" , legend = \"True function\" , col = \"red\" , lty = 1 , bty = \"n\")\n}\n\n\\concept{preferential attachment}\n\\concept{attachment function}\n", "meta": {"hexsha": "b9fbbdc79320a0ebd39f79f8e4a1fc533ad2d71e", "size": 3277, "ext": "rd", "lang": "R", "max_stars_repo_path": "man/Newman.rd", "max_stars_repo_name": "eddelbuettel/PAFit", "max_stars_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "man/Newman.rd", "max_issues_repo_name": "eddelbuettel/PAFit", "max_issues_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "man/Newman.rd", "max_forks_repo_name": "eddelbuettel/PAFit", "max_forks_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 46.8142857143, "max_line_length": 263, "alphanum_fraction": 0.7076594446, "num_tokens": 884, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7279754371026367, "lm_q2_score": 0.6992544273261176, "lm_q1q2_score": 0.5090400473786844}} {"text": "# /* !/usr/bin/Rscript */\n\n#' ## Model fitting\n#' Given problem is of a classification nature, various ML algos apply.\n#' Start with a basic CART model to view the major splits of a decision\n#' tree which will hint at important variables, then use Caret to train\n#' and iterate upon a number of different models. Caret's unified\n#' interface is especially a time-saver in this regard.\n#'\n#' One important note: The SEED is set before every model so that it is\n#' consistent when models are run out of order during development of the\n#' analysis.\n\n# /*\nwriteLines(\"\\n-------------\");\nwriteLines(\"Modelling: Glm logistic, CART, Boosting and Ensembles\")\n# */\n\n#' ### Define `caret` training parameters\n#' Mitigate overfitting (high variance) by resampling the training\n#' samples to cross-validate fitted model's performance on future data.\n#' This is applied as a K-fold CV where K=10 repeated 3 times.\n#'\n#' The comparison performance metric of choice is ROC (AUC). To retain\n#' the necessary output in the models the `twoClassSummary()` fn (from\n#' `caret`) is specified as the summary function in order to to compute\n#' the sensitivity, specificity and AUC.\n#'\n#' The training control function is also configured to compute the\n#' class probabilities for held-out samples during the resample for\n#' the performance comparison.\nfitControl <- trainControl(\n method = \"repeatedcv\",\n number = 10,\n repeats = 3,\n summaryFunction = twoClassSummary,\n classProbs = TRUE\n)\n\n#' ### Basic CART model\n#' Fit a classification tree to view important splits.\nset.seed(SEED)\n\ntreeFit = tree(Fate ~ .,\n data=tr\n)\nplot(treeFit)\ntext(treeFit)\nsummary(treeFit)\n\n#' ### Logistic regression models\n#' To start the modelling proper, iterate through a number of logistic\n#' regression model trained with caret (as Generalised Linear Models -\n#' GLM).\n#'\n#' Initially all features are added, then removed or combined as\n#' interactive covariates, with interesting interactions captured in\n#' binary features via class compression.\n#'\n#' The impact in changes to the input formula were measured by each\n#' model's deviance (a measure of lack of fit between model and data).\n#' For this metric larger is worse. This said, lower deviance is also\n#' indicative to an overfitted model (high bias).\n#'\n#' It was noted that this method raised \"rank-deficient\" errors, likely\n#' due to the low number of observations in the dataset. A GLM with all\n#' features in the training set scored 0.74163 on Kaggle.\n#'\n#' #### GLM all features\nset.seed(SEED)\n\nglmFit1 <- train(Fate ~ .,\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n\n#' #### Basic GLM\n#' Try a model with the selected variables in inferred order of\n#' importance.\nset.seed(SEED)\n\nglmFit2 <- train(Fate ~ Sex + Class + Age + FamilyCount + Embarked +\n FareLog + BoatPriority,\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n\n# /*\n# Model: Logistic Regression with interaction variables\nset.seed(SEED)\n\nglmFit3 <- train(Fate ~ Sex + Class + Title + Sex*Class +\n Sex*Class*Title,\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n# */\n\n#' #### GLM with interaction variables\n#' Used to discern key interactions for potential class compression.\n#' In the following model the interaction between\n#' *ClassLower:PcSurvived* was deemed significant.\n#' `tr[tr$PcSurvived == 1, c(\"Fate\", \"Class\")]` reveals that most\n#' people travelling in lower class with a Parent/child who\n#' survived, also survived (in the training set).\nset.seed(SEED)\n\nglmFit4 <- train(Fate ~ Class + Sex + Title + Sex*Embarked +\n Class*PcSurvived*SbSurvived,\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n\n# /*\n# Model: Logistic Regression with class compression\nset.seed(SEED)\n\nglmFit5 <- train(Fate ~ Sex + Class + Title +\n I(Title==\"Mr\"&Class==\"Lower\"),\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n\n# Model: Logistic Regression with class compression\nset.seed(SEED)\n\nglmFit6 <- train(Fate ~ Sex + Class + Title +\n I(Title==\"Mr\"&Class==\"Lower\") +\n I(Class==\"Middle\"&Title==\"Master\") +\n I(Sex==\"female\"&Embarked==\"S\") +\n I(Class==\"Lower\"&PcSurvived==1),\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n\n# Model: Logistic Regression adding PcSurvived and SbSurvived\nset.seed(SEED)\n\nglmFit7 <- train(Fate ~ Sex + Class + Title + PcSurvived + SbSurvived +\n I(Title==\"Mr\"&Class==\"Lower\") +\n I(Class==\"Middle\"&Title==\"Master\") +\n I(Sex==\"female\"&Embarked==\"S\") +\n I(Class==\"Lower\"&PcSurvived==1),\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n# */\n\n#' #### GLM with class compression\n#' In this model a number of important features have been added as\n#' covariates alongside some engineered features capturing various\n#' important interactions identified previously. This submission\n#' scored 0.79904\nset.seed(SEED)\n\nglmFit8 <- train(Fate ~ Sex + Class + Title + PcSurvived + SbSurvived +\n FareLog + FamilyCount +\n I(Title==\"Mr\"&Class==\"Lower\") +\n I(Class==\"Middle\"&Title==\"Master\") +\n I(Sex==\"female\"&Embarked==\"S\") +\n I(Class==\"Lower\"&PcSurvived==1),\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n\n# /*\n# Model: Logistic Regression with selected variables\n# - Submission scored 0.79904\nset.seed(SEED)\n\nglmFit9 <- train(Fate ~ Class + Sex + Title + Age + Embarked + FamilyCount + PcSurvived + SbSurvived + Side,\n data = tr,\n method = \"glm\",\n metric = \"ROC\",\n trControl = fitControl\n)\n# */\n\n#' ### Tree Models\n#' After training increasingly complex GLM models next comes a very\n#' popular type of model for classification problems - tree-based\n#' models. Although the first CART model didn't yield any great\n#' insight, other tree-based methods are far more sophisticated in\n#' nature, able to run thousands of permutations and combine them to\n#' create a highly optimised and complex models.\n#'\n#' To keep it fair between different model types (not too mention\n#' simple!) the same formula was used to fit each model:\n#' `Fate ~ Class + Sex + Title + Age + Embarked + FamilyCount + PcSurvived + SbSurvived + FareLog'`\n#'\n#' The features listed as co-variates were derived from the best of\n#' the GLM models. Those excluded were deemed superfluous due to\n#' their lack of effect on the deviance change in the models they\n#' were included.\n#'\n#' Finally, in all cases the same training control function used to\n#' train the GLM logistic regression models was re-purposed.\n#'\n#' ---\n#' NOTE: These models can take a while to train, hang in there...\n#'\n#' #### Ada boosted tree model\n#' Boosting constitutes an ensemble model as an iterative algo that\n#' combines simple classification rules with 'mediocre' performance\n#' in terms of misclassification error rate to produce highly\n#' accurate classification rule.\n#'\n#' Ada is statistical boosting based on additive logistic regression.\n#' Here classification trees are generated via rpart and the caret\n#' package constitute the base classifiers.\n#'\n#' For this model a tuning grid configured for varying iterations\n#' (25 & 50), maxdepth (4 & 6), and nu shrinkage parameter (0.1 & 1)\n#' was created. All permutations were exercised.\n#'\n#' One can also visualise feature variance with `varplot()` from the\n#' ada package. The model reported an accuracy of 0.84, and scored\n#' 0.80383 on Kaggle's leaderboard.\nset.seed(SEED)\n\nadaGrid <- expand.grid(\n .iter = c(25, 50),\n .maxdepth = c(4, 6),\n .nu = c(0.1, 1)\n)\n\nadaFit <- train(Fate ~ Class + Sex + Title + Age + Embarked +\n FamilyCount + PcSurvived + SbSurvived + FareLog,\n data = tr,\n method = \"ada\",\n metric = \"ROC\",\n tuneGrid = adaGrid,\n trControl = fitControl\n)\n\nvarplot(adaFit$finalModel)\nprint(adaFit)\n\n#' #### Random forest\n#' The (in)famous rf model. This is an ensemble method similar to\n#' bagged trees (performed with boostrap aggregation), whereby\n#' observations are resampled and predictions are recalculated, with\n#' majority vote dictating the best model.\n#'\n#' Random forests also bootstrap the variables at each split in the\n#' tree and use the Out-of-Bag Error rate to evaluate strength of\n#' model, correlation between models, and variable importance.\n#'\n#' For this model a tuning grid with mtry set to 3 (as per Strobl et al\n#' advice that this value should equate to the sq.root of number of\n#' covariates - 9) was created. Variable importance was determined\n#' using the `importance()` fn from the random forest package. They\n#' are ranked by the Gini purity metric. This model scored 0.81340\nset.seed(SEED)\n\nrfGrid <- expand.grid(.mtry = c(3))\n\nrfFit <- train(Fate ~ Class + Sex + Title + Age + Embarked + FamilyCount + PcSurvived + SbSurvived + FareLog,\n data = tr,\n method = \"rf\",\n metric = \"ROC\",\n tuneGrid = rfGrid,\n trControl = fitControl\n)\n\nimportance(rfFit$finalModel)\nvarImpPlot(rfFit$finalModel)\nprint(rfFit)\n\n#' #### Conditional forest\n#' An alternative to random forests which handle factor co-variates\n#' with many levels better than their better known brethren. A tuning\n#' grid with the same tuning grid as the random forest was created,\n#' and the same variables used to discern important variables.\n#'\n#' Ultimately it scored 0.80861, lower than the random forest model.\nset.seed(SEED)\n\ncfGrid <- expand.grid(.mtry = c(3))\n\ncfFit <- train(Fate ~ Class + Sex + Title + Age + Embarked + FamilyCount + PcSurvived + SbSurvived + FareLog,\n data = tr,\n method = \"cforest\",\n metric = \"ROC\",\n tuneGrid = cfGrid,\n trControl = fitControl\n)\n\nprint(cfFit)\n\n#' ### Support vector machine\n#'\n#' SVM build a model that assigns observations into one category or the other\n#' based on a kernel method (comparator function). It is an example of a\n#' non-probabilistic binary linear classifier.\n#'\n#' For this model it was specified that the data should be pre-processed by\n#' centering and scaling it, as SVM is sensitive, especially when using a\n#' Gaussian kernel.\n#'\n#' Furthermore, the `tuneLength` which determines the number of levels for\n#' each tuning parameters generated by train() was set to 9. This produces a\n#' tuning grid of c(0.25, 0.5, 1, 2, 4, 8, 16, 32, 64), which corresponds to\n#' the value passed in for the `C` parameter of the kernel method. In turn,\n#' this controls the \"penalty\" for misclassified training examples.\n#'\n#' In this case the Sigma value determined by Caret. The submission scored a\n#' lowly 0.78469.\nset.seed(SEED)\n\nsvmFit <- train(Fate ~ Class + Sex + Title + Age + Embarked + FamilyCount + PcSurvived + SbSurvived + Side + FareLog + BoatPriority,\n data = tr,\n method = \"svmRadial\",\n metric = \"ROC\",\n tuneLength = 9,\n preProcess = c(\"center\", \"scale\"),\n trControl = fitControl\n)\n\nprint(svmFit)\n\n#' Meta-ensemble model\n#' This type of model combines models of different types (or the same even)\n#' in the same vein as ensemble models such as random forest, with majority\n#' vote scoring (hence the number of models should be odd).\n#'\n#' Although this could be done crudely by simply running a prediction for\n#' the desired records and comparing the results to generate the mode for\n#' each, using `caretEnsemble` means a uses a glm is used instead to create\n#' a simple linear blend of models with the same re-sampling parameters.\n#' Ideally the methods used produce models with low correlation in an effort\n#' to cover more of the variance within a given dataset.\n#'\n#' To use `caretEnsemble` another training control function needs to be\n#' created, which closely mirrors that used to construct the independent\n#' models above, but which also specifies the resample index callback. This\n#' ensures they are the same across all models.\n#'\n#' Equally, the same tuning grids are used to re-train ada, rf, and cf\n#' models, as well as the same formula. This ensemble scored 0.80383, which\n#' just goes to show more complex models don't always beat simple ones.\nfitEnsembleControl <- trainControl(\n method = \"repeatedcv\",\n number = 10,\n repeats = 3,\n savePredictions = TRUE,\n classProbs = TRUE,\n index = createResample(tr$Fate, 10),\n summaryFunction = twoClassSummary\n)\n\nset.seed(SEED)\nensembleList <- caretList(Fate ~ Class + Sex + Title + Age + Embarked + FamilyCount + PcSurvived + SbSurvived + FareLog,\n data = tr,\n metric = \"ROC\",\n trControl = fitEnsembleControl,\n methodList=c(\"glm\", \"svmRadial\"),\n tuneList = list(\n ada = caretModelSpec(method=\"ada\", tuneGrid=adaGrid),\n rf = caretModelSpec(method=\"rf\", tuneGrid=rfGrid),\n cf = caretModelSpec(method=\"cforest\", tuneGrid=cfGrid)\n )\n)\n\nensembleFit <- caretEnsemble(ensembleList)\n\n# /*\nwriteLines(\"\\n-------------\")\nwriteLines(\"Modelling: DONE\")\n# */\n\n", "meta": {"hexsha": "07dd12e19a638e7bb63168c151cd98d1a7a4404c", "size": 12719, "ext": "r", "lang": "R", "max_stars_repo_path": "src/model.r", "max_stars_repo_name": "andybeeching/kaggle-titanic", "max_stars_repo_head_hexsha": "df1000173aa9e7e5846611fd0be1223e709fbe06", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2018-07-13T14:20:50.000Z", "max_stars_repo_stars_event_max_datetime": "2018-07-13T14:20:50.000Z", "max_issues_repo_path": "src/model.r", "max_issues_repo_name": "andybeeching/kaggle-titanic", "max_issues_repo_head_hexsha": "df1000173aa9e7e5846611fd0be1223e709fbe06", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/model.r", "max_forks_repo_name": "andybeeching/kaggle-titanic", "max_forks_repo_head_hexsha": "df1000173aa9e7e5846611fd0be1223e709fbe06", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 32.7809278351, "max_line_length": 132, "alphanum_fraction": 0.708782137, "num_tokens": 3295, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8175744850834649, "lm_q2_score": 0.6224593312018546, "lm_q1q2_score": 0.5089068671927541}} {"text": "#' Function that propagates measurement uncertainty through model results\n#' \n#' Function to propagate combined errors on \\code{x} (= \\code{Dsam}) and\n#' \\code{y} (= \\code{Osam}) on the modeled X (= \\code{D}) and Y \n#' (= \\code{d18Oc}) values by means of projection of uncertainties\n#' through the modeled \\code{X-Y} relationship\n#'\n#' Note: projection leads to large uncertainties on shallow parts of the\n#' \\code{X–Y} curve\n#' @param x Vector of \\code{x} values of input data\n#' @param x_err Vector of uncertainties on \\code{x} values\n#' @param y Vector of \\code{y} values of input data\n#' @param y_err Vector of uncertainties on \\code{y} values\n#' @param X Vector of modeled \\code{X} values on which the uncertainty is\n#' to be projected\n#' @param Y Matrix of modeled x and \\code{Y} values\n#' @param MC Number of Monte Carlo simulations to apply for error propagation\n#' Default = 1000\n#' @return A vector listing the standard deviations of propagated errors \n#' propagated on all \\code{X} values.\n#' @examples\n#' # Create dummy data for input\n#' x <- seq(1, 40, 1)\n#' x_err <- rep(0.1, 40)\n#' y <- sin((2 * pi * (seq(1, 40, 1) - 8 + 30 / 4)) / 30)\n#' y_err <- rep(0.1, 40)\n#' X <- seq(1.5, 39.5, 1)\n#' Y <- cbind(seq(1, 39, 1), 0.9 * sin((2 * pi * (seq(1, 39, 1) - 9 +\n#' 25 / 4)) / 25))\n#' # Run function\n#' result <- mc_err_form(x, x_err, y, y_err, X, Y, 1000)\n#' @export\nmc_err_form <- function(x,\n x_err,\n y,\n y_err,\n X,\n Y,\n MC = 1000){\n \n xmat <- matrix(rnorm(MC * length(x)), nrow = length(x)) * x_err + matrix(rep(x, MC), nrow = length(x)) # Create matrix of simulated X values\n Xpos <- apply(abs(outer(xmat, X, FUN = \"-\")), c(1,2), which.min) # find closest position in X vector (day) for each simulated X value\n Xpos_stat <- cbind(rowMeans(Xpos), apply(Xpos, 1, sd)) # Find mean and standard deviation of positions in X vector (day) for each sample\n \n ymat <- matrix(rnorm(MC * length(y)), nrow = length(y)) * y_err + matrix(rep(Y[Xpos_stat[, 1], 2], MC), nrow = length(y)) # Create matrix of Monte Carlo simulated Y values projected on the X-Y curve\n Xpos_mat <- outer(round(Xpos_stat[, 1]), seq(-20, 20, 1), \"+\") %% length(D) + 1 # Create matrix of D positions around the mean position for each sample to match with simulated Y values. Window is +/- 20 positions\n Xpos_matO <- matrix(Y[Xpos_mat, 2], nrow = length(x)) # Find Y values for each position in Xpos_mat\n Xpos_matO2 <- apply(outer(Xpos_matO, ymat, FUN = \"-\"), c(2,4), diag) # Match and subtract each simulated Y value with the local environment of Y values in the X-Y curve\n Ypos <- apply(abs(Xpos_matO2), c(1, 3), which.min) + Xpos_stat[, 1] # Find the position of the closest Y value in the Y window (Xpos_matO) for each Y simulation\n Ypos_stat <- cbind(rowMeans(Ypos), apply(Ypos, 1, sd)) # Find mean and standard deviation of positions in X vector (day) for each sample\n Ypos_stat[which(Ypos_stat[, 2] == 0), 2] <- sd(seq(-20, 20, 1)) # If SD of Y falls outside the window of +/- 20 positions SD is assumed to be equal to that of the uniform distribution\n \n pos_err_comb <- sqrt(Ypos_stat[, 2] ^ 2 + Xpos_stat[, 2] ^ 2) # Find combined error on position (day)\n Xpos_minmax <- cbind(Xpos_stat[, 1] - pos_err_comb, Xpos_stat[, 1] + pos_err_comb) %% length(D) # Find min and max position values\n Xminmax <- matrix(approx(x = X, xout = Xpos_minmax, rule = 2)$y, ncol = 2) # Find associated D values (linear interpolation)\n x_err_comb <- 0.5 * ((Xminmax[,2] - Xminmax[,1]) %% X[length(X)]) # Find combined SD in D domain\n return(x_err_comb)\n}", "meta": {"hexsha": "88bf50ae687c5e4c1d29a25c6e3bbde3cab73ace", "size": 3595, "ext": "r", "lang": "R", "max_stars_repo_path": "R/mc_err_form.r", "max_stars_repo_name": "nhoeche/ShellChron.jl", "max_stars_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/mc_err_form.r", "max_issues_repo_name": "nhoeche/ShellChron.jl", "max_issues_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/mc_err_form.r", "max_forks_repo_name": "nhoeche/ShellChron.jl", "max_forks_repo_head_hexsha": "935bcde7e015581a18eee324b7a5ff2ab7f1970f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 61.9827586207, "max_line_length": 217, "alphanum_fraction": 0.659527121, "num_tokens": 1132, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.817574471748733, "lm_q2_score": 0.6224593312018546, "lm_q1q2_score": 0.5089068588924259}} {"text": "#' Creates phylogenetic tree and generates a single trait for each species.\n#'\n#' @param b Birth rate.\n#' @param d Death rate.\n#' @param n_taxa A number of species in the tree.\n#' @param model A model, under which trait should be generated (\"BM\" or \"OU\").\n#' @param sigma A standard-deviation of the random component for each branch.\n#' @param alpha A strength of the selective constraint for each branch in OU model.\n#'\n#' @return List consisting of phylogenetic tree, matrix of phylogenetic distances, trait and matrix of trait distances.\n#' @export\n#'\n#' @examples\n#' sim_tree_n_traits(b = 0.1, d = 0, n_taxa = 500, model = \"BM\", sigma = 0.1, alpha = 0.1)\n\nsim_tree_n_traits <- function(b = 0.1, d = 0, n_taxa = 500, model = \"BM\", sigma = 0.1, alpha = 0.1){\n tree <- geiger::sim.bdtree(b = b, d = d, stop = \"taxa\", n = n_taxa)\n dist.phy <- cophenetic(tree)\n colnames(dist.phy) <- rownames(dist.phy) <- 1:n_taxa\n\n tr <- ape::rTraitCont(tree, model = model, sigma = sigma, alpha = alpha)\n trait <- (tr - min(tr)) / (max(tr) - min(tr)) * 100\n names(trait) <- 1:length(trait)\n tr.dist <- as.matrix(dist(trait))\n\n res <- list(phylo = tree, phylo_dist = dist.phy, trait = trait, trait_dist = tr.dist)\n\n return(res)\n}\n", "meta": {"hexsha": "afed1598c9dccada9408c6f947d32aa189960c1f", "size": 1223, "ext": "r", "lang": "R", "max_stars_repo_path": "R/sim_tree_n_traits.r", "max_stars_repo_name": "arrirh/sim.assembly", "max_stars_repo_head_hexsha": "4ccf9947f0802bcdfff00aeb45a3072773fd4981", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/sim_tree_n_traits.r", "max_issues_repo_name": "arrirh/sim.assembly", "max_issues_repo_head_hexsha": "4ccf9947f0802bcdfff00aeb45a3072773fd4981", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/sim_tree_n_traits.r", "max_forks_repo_name": "arrirh/sim.assembly", "max_forks_repo_head_hexsha": "4ccf9947f0802bcdfff00aeb45a3072773fd4981", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 40.7666666667, "max_line_length": 119, "alphanum_fraction": 0.6647587899, "num_tokens": 383, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8080672227971211, "lm_q2_score": 0.6297746074044134, "lm_q1q2_score": 0.5089002179934315}} {"text": "\\name{only_A_estimate}\n\\alias{only_A_estimate}\n%- Also NEED an '\\alias' for EACH other topic documented here.\n\\title{\n Estimating the attachment function in isolation by PAFit method \n}\n\\description{\nThis function estimates the attachment function \\eqn{A_k} by PAFit method. The method has a hyper-parameter \\eqn{r}. It first performs a cross-validation step to select the optimal parameter \\eqn{r} for the regularization of \\eqn{A_k}, then uses that \\eqn{r} to estimate the attachment function with the full data. \n}\n\\usage{\nonly_A_estimate(net_object , \n net_stat = get_statistics(net_object), \n p = 0.75 ,\n stop_cond = 10^-8 , \n mode_reg_A = 0 ,\n MLE = FALSE ,\n ...)\n}\n%- maybe also 'usage' for other objects documented here.\n\\arguments{\n \\item{net_object}{\n an object of class \\code{PAFit_net} that contains the network.\n }\n \\item{net_stat}{\n An object of class \\code{PAFit_data} which contains summerized statistics needed in estimation. This object is created by the function \\code{\\link{get_statistics}}. The default value is \\code{get_statistics(net_object)}.\n }\n \\item{p}{Numeric. This is the ratio of the number of new edges in the learning data to that of the full data. The data is then divided into two parts: learning data and testing data based on \\code{p}. The learning data is used to learn the node fitnesses and the testing data is then used in cross-validation. Default value is \\code{0.75}.}\n\\item{stop_cond}{Numeric. The iterative algorithm stops when \\eqn{abs(h(ii) - h(ii + 1)) / (abs(h(ii)) + 1) < stop.cond} where \\eqn{h(ii)} is the value of the objective function at iteration \\eqn{ii}. We recommend to choose \\code{stop.cond} at most equal to \\eqn{10^(- number of digits of h - 2)}, in order to ensure that when the algorithm stops, the increase in posterior probability is less than 1\\% of the current posterior probability. Default is \\code{10^-8}. This threshold is good enough for most applications.}\n\n\\item{mode_reg_A}{Binary. Indicates which regularization term is used for \\eqn{A_k}:\n\\itemize{\n\\item \\code{0}: This is the regularization term used in Ref. 1 and 2. Please refer to Eq. (4) in the tutorial for the definition of the term. It approximately enforces the power-law form \\eqn{A_k = k^\\alpha}. This is the default value. \n\\item \\code{1}: Unlike the default, this regularization term exactly enforces the functional form \\eqn{A_k = k^\\alpha}. Please refer to Eq. (6) in the tutorial for the definition of the term. Its main drawback is it is significantly slower to converge, while its gain over the default one is marginal in most cases. \n}\n}\n\\item{MLE}{Logical. If \\code{TRUE}, then not perform cross-validation and estimate the PA function with \\code{r = 0}, i.e., maximum likelihood estimation. Default is \\code{FALSE}. One might want to set this option to \\code{TRUE} when one believes that there are sufficient data to get a reasonable MLE result, or when one wants to compare the default, regularized result with the MLE result.}\n \\item{\\dots}{\n %% ~~Describe \\code{\\dots} here~~\n }\n}\n\n\\value{\n Outputs a \\code{Full_PAFit_result} object, which is a list containing the following fields:\n \\itemize{\n \\item \\code{cv_data}: a \\code{CV_Data} object which contains the cross-validation data. This is the final Normally the user does not need to pay attention to this data. \\code{NULL} if \\code{MLE = TRUE}.\n \n \\item \\code{cv_result}: a \\code{CV_Result} object which contains the cross-validation result. Normally the user does not need to pay attention to this data. \\code{NULL} if \\code{MLE = TRUE}.\n \n \\item \\code{estimate_result}: this is a \\code{PAFit_result} object which contains the estimated PA function and its confidence interval. It also includes the estimated attachment exponenent \\eqn{\\alpha} (assuming the model \\eqn{A_k = k^\\alpha}) in the field \\code{alpha}, and the confidence interval of \\eqn{\\alpha} (in the field \\code{ci}) when possible. In particular, the important fields are:\n \\itemize{\n \\item \\code{ratio}: this is the selected value for the hyper-parameter \\eqn{r}.\n \\item \\code{k} and \\code{A}: a degree vector and the estimated PA function.\n \\item \\code{var_A}: the estimated variance of \\eqn{A}.\n \\item \\code{var_logA}: the estimated variance of \\eqn{log A}.\n \\item \\code{upper_A}: the upper value of the interval of two standard deviations around \\eqn{A}.\n \\item \\code{lower_A}: the lower value of the interval of two standard deviations around \\eqn{A}.\n \n \\item \\code{center_k} and \\code{theta}: when we perform binning, these are the centers of the bins and the estimated PA values for those bins. \\code{theta} is similar to \\code{A} but with duplicated values removed.\n \\item \\code{var_bin}: the variance of \\code{theta}. Same as \\code{var_A} but with duplicated values removed.\n \\item \\code{upper_bin}: the upper value of the interval of two standard deviations around \\code{theta}. Same as \\code{upper_A} but with duplicated values removed.\n \\item \\code{lower_lower}: the lower value of the interval of two standard deviations around \\code{theta}. Same as \\code{lower_A} but with duplicated values removed.\n \\item \\code{g}: the number of bins used.\n \\item \\code{alpha} and \\code{ci}: \\code{alpha} is the estimated attachment exponenet \\eqn{\\alpha} (when assume \\eqn{A_k = k^\\alpha}), while \\code{ci} is the confidence interval.\n \\item \\code{loglinear_fit}: this is the fitting result when we estimate \\eqn{\\alpha}. \n \\item \\code{objective_value}: values of the objective function over iterations in the final run with the full data.\n \\item \\code{diverge_zero}: logical value indicates whether the algorithm diverged in the final run with the full data.\n}\n}\n}\n\\author{\n Thong Pham \\email{thongpham@thongpham.net}\n}\n\\references{\n 1. Pham, T., Sheridan, P. & Shimodaira, H. (2015). PAFit: A Statistical Method for Measuring Preferential Attachment in Temporal Complex Networks. PLoS ONE 10(9): e0137796. doi:10.1371/journal.pone.0137796 (\\url{http://dx.doi.org/10.1371/journal.pone.0137796}).\n \n 2. Pham, T., Sheridan, P. & Shimodaira, H. (2016). Joint Estimation of Preferential Attachment and Node Fitness in Growing Complex Networks. Scientific Reports 6, Article number: 32558. doi:10.1038/srep32558 (\\url{http://www.nature.com/articles/srep32558}).\n}\n\\seealso{\n See \\code{\\link{get_statistics}} for how to create summerized statistics needed in this function.\n \n See \\code{\\link{Newman}} and \\code{\\link{Jeong}} for other methods to estimate the attachment function \\eqn{A_k} in isolation.\n \n}\n\n\\examples{\n\\dontrun{\n library(\"PAFit\")\n set.seed(1)\n #### Example 1: Linear preferential attachment #########\n # a network from BA model\n net <- generate_net(N = 1000 , m = 50 , mode = 1, alpha = 1, s = 0)\n \n net_stats <- get_statistics(net, only_PA = TRUE)\n result <- only_A_estimate(net, net_stats)\n \n # plot the estimated attachment function\n plot(result, net_stats)\n \n # true function\n true_A <- result$estimate_result$center_k\n lines(result$estimate_result$center_k, true_A, col = \"red\") # true line\n legend(\"topleft\" , legend = \"True function\" , col = \"red\" , lty = 1 , bty = \"n\")\n \n #### Example 2: a non-log-linear preferential attachment #########\n # A_k = alpha* log (max(k,1))^beta + 1, with alpha = 2, and beta = 2\n set.seed(1)\n net <- generate_net(N = 1000 , m = 50 , mode = 3, alpha = 2, beta = 2, s = 0)\n \n net_stats <- get_statistics(net,only_PA = TRUE)\n result <- only_A_estimate(net, net_stats)\n \n # plot the estimated attachment function\n plot(result, net_stats)\n \n # true function\n true_A <- 2 * log(pmax(result$estimate_result$center_k,1))^2 + 1 # true function\n lines(result$estimate_result$center_k, true_A, col = \"red\") # true line\n legend(\"topleft\" , legend = \"True function\" , col = \"red\" , lty = 1 , bty = \"n\")\n \n #############################################################################\n #### Example 3: another non-log-linear preferential attachment kernel ############\n set.seed(1)\n # A_k = min(max(k,1),sat_at)^alpha, with alpha = 1, and sat_at = 200\n # inverse variance of the distribution of node fitnesse = 10\n net <- generate_net(N = 1000 , m = 50 , mode = 2, alpha = 1, sat_at = 200, s = 0)\n net_stats <- get_statistics(net, only_PA = TRUE)\n \n result <- only_A_estimate(net, net_stats)\n \n \n # plot the estimated attachment function\n true_A <- pmin(pmax(result$estimate_result$center_k,1),200)^1 # true function\n plot(result , net_stats, max_A = max(true_A,result$estimate_result$theta))\n lines(result$estimate_result$center_k, true_A, col = \"red\") # true line\n legend(\"topleft\" , legend = \"True function\" , col = \"red\" , lty = 1 , bty = \"n\")\n }\n}\n\n\\concept{preferential attachment}\n\\concept{attachment function}\n\n", "meta": {"hexsha": "fa0de1707a2b88f86647008e19c2686f96ebd640", "size": 9042, "ext": "rd", "lang": "R", "max_stars_repo_path": "man/only_A_estimate.rd", "max_stars_repo_name": "eddelbuettel/PAFit", "max_stars_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "man/only_A_estimate.rd", "max_issues_repo_name": "eddelbuettel/PAFit", "max_issues_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "man/only_A_estimate.rd", "max_forks_repo_name": "eddelbuettel/PAFit", "max_forks_repo_head_hexsha": "4e362700bccb0de76778f31312b113cce85f6b4b", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 63.676056338, "max_line_length": 519, "alphanum_fraction": 0.6917717319, "num_tokens": 2439, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7520125737597972, "lm_q2_score": 0.6757646010190476, "lm_q1q2_score": 0.5081834768680965}} {"text": "#' Function for computing the posterior mode cluster assignment of subjects in test and training data sets.\n#'\n#' This function takes the \\code{cluster_inds} - which is a posterior matrix of cluster assignments output by \\code{NDPMix()}, \\code{PDPMix()}, and \\code{ZDPMix()} and computes the posterior mode cluster assignment while implementing a deterministic re-labeling of subjects.\n#' \n#' Please see \\url{https://stablemarkets.github.io/ChiRPsite/index.html} for examples and detailed model and parameter descriptions.\n#' \n#' Please see \\url{https://arxiv.org/abs/1810.09494}, section on hyperparameters and label switching for more information.\n#' \n#' @import stats\n#' \n#' @param c_shell set this to \\code{cluster_inds} (for either training or testing sets). \\code{cluster_inds} is output by \\code{NDPMix()}, \\code{PDPMix()}, and \\code{ZDPMix()}.\n#' @return This function returns a list of two objects: \\code{adjmat} and \\code{class_mem}. For \\code{n} subjects, the former is an \\code{n} by \\code{n} adjacency matrix with i-j th element giving the posterior probability of subject i and j being clustered together. This can be visualized using a network diagram (see examples on website linked in Details). \\code{class_mem} is a vector of length \\code{n} giving posterior mode clsuter membership for each of the \\code{n} subjects. This can be used for cluster-specific analysis, for example.\n#' @examples\n#' # simulate data \n#' \n#' set.seed(1)\n#' n<-200 ## generate from clustered, skewed, data distribution\n#' X11 <- rnorm(n = n, mean = 10, sd = 3)\n#' X12 <- rnorm(n = n, mean = 0, sd = 2)\n#' X13 <- rnorm(n = n, mean = -10, sd = 4)\n#' \n#' Y1 <- rnorm(n = n, mean = 100 + .5*X11, 20)*(1-rbinom(n, 1, prob = pnorm( -10 + 1*X11 ) ))\n#' Y2 <- rnorm(n = n, mean = 200 + 1*X12, 30)*(1-rbinom(n, 1, prob = pnorm( 1 + .05*X12 ) ))\n#' Y3 <- rnorm(n = n, mean = 300 + 2*X13, 40)*(1-rbinom(n, 1, prob = pnorm( -3 -.2*X13 ) ))\n#' \n#' d <- data.frame(X1=c(X11, X12, X13), Y = c(Y1, Y2, Y3))\n#' \n#' # split into training and testing\n#' ids <- sample(1:600, size = 500, replace = FALSE )\n#' \n#' d$X1 <- scale(d$X1)\n#' \n#' d_train <- d[ids,]\n#' d_test <- d[-ids, ]\n#' \n#' # run zero-inflated model #' \n#' res <- ChiRP::ZDPMix(d_train = d_train, d_test = d_test, formula = Y ~ X1,\n#' burnin=100, iter=200, init_k = 5, phi_y = c(10, 10000))\n#' \n#' # compute the posterior model cluster assignment for training subjects\n#' train_clus <- ChiRP::cluster_assign_mode(res$cluster_inds$train)\n#' @export\ncluster_assign_mode <- function(c_shell){\n c_shell <- c_shell\n iter<-ncol(c_shell)\n n <- nrow(c_shell)\n \n adjmat<-matrix(0, nrow=n, ncol=n)\n \n ## compute frequency matrix \n for(i in 1:iter){\n adjmat_i <- matrix(0, nrow=n, ncol=n)\n class <- c_shell[,i]\n \n for(j in 1:n){\n adjmat_i[j,] <- class==class[j]\n }\n adjmat <- adjmat + adjmat_i\n }\n \n ## choose best cluster assignment\n err_vec <- numeric(length = ncol(c_shell))\n for(i in 1:iter){\n adjmat_i <- matrix(0, nrow=n, ncol=n)\n class <- c_shell[,i]\n for(j in 1:n){\n adjmat_i[j,] <- class==class[j]\n } # compute L2 norm.\n err_vec[i] <- sum((adjmat_i - adjmat)^2)\n }\n \n class_mem <- c_shell[, c(1:ncol(c_shell))[ err_vec==min(err_vec) ] ]\n if(is.matrix(class_mem)) class_mem <- class_mem[,1]\n \n adjmat <- adjmat/iter\n return(list(adjmat=adjmat, class_mem=class_mem))\n}", "meta": {"hexsha": "a0768dd740ad9acb445500d2f6ca291fd3ae0ee8", "size": 3388, "ext": "r", "lang": "R", "max_stars_repo_path": "R/cluster_assign_mode.r", "max_stars_repo_name": "lagvier/ChiRP", "max_stars_repo_head_hexsha": "48504130a909a5b46105ba13da8e0d50405d5beb", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 10, "max_stars_repo_stars_event_min_datetime": "2019-02-26T03:01:37.000Z", "max_stars_repo_stars_event_max_datetime": "2021-03-10T09:19:11.000Z", "max_issues_repo_path": "R/cluster_assign_mode.r", "max_issues_repo_name": "lagvier/ChiRP", "max_issues_repo_head_hexsha": "48504130a909a5b46105ba13da8e0d50405d5beb", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 5, "max_issues_repo_issues_event_min_datetime": "2020-02-24T19:32:53.000Z", "max_issues_repo_issues_event_max_datetime": "2021-07-22T01:26:40.000Z", "max_forks_repo_path": "R/cluster_assign_mode.r", "max_forks_repo_name": "lagvier/ChiRP", "max_forks_repo_head_hexsha": "48504130a909a5b46105ba13da8e0d50405d5beb", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 3, "max_forks_repo_forks_event_min_datetime": "2019-03-03T19:25:17.000Z", "max_forks_repo_forks_event_max_datetime": "2020-11-18T18:03:51.000Z", "avg_line_length": 44.0, "max_line_length": 544, "alphanum_fraction": 0.652892562, "num_tokens": 1058, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7931059511841119, "lm_q2_score": 0.640635861701035, "lm_q1q2_score": 0.5080921144570525}} {"text": "#' mph2ms\n#' \n#' Conversion speed from mile per hour in meter per second.\n#'\n#' @param numeric mph Speed in mile per hour.\n#' @return \n#'\n#'\n#' @author Istituto di Biometeorologia Firenze Italy Alfonso Crisci \\email{a.crisci@@ibimet.cnr.it}\n#' @keywords mph2ms\n#' \n#' @export\n#'\n#'\n#'\n#'\n\nmph2ms<-function(mph)\n{\n return (mph * 0.44704);\n}", "meta": {"hexsha": "dbbaf9a9a9f7eef75b7ce0a78fb50b91bec074ea", "size": 344, "ext": "r", "lang": "R", "max_stars_repo_path": "R/mph2ms.r", "max_stars_repo_name": "alfcrisci/biometeoR", "max_stars_repo_head_hexsha": "e9ee73da6ecc515ecd471cb9ae059a4370a51493", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2019-06-13T15:54:46.000Z", "max_stars_repo_stars_event_max_datetime": "2019-06-13T15:54:46.000Z", "max_issues_repo_path": "R/mph2ms.r", "max_issues_repo_name": "alfcrisci/biometeoR", "max_issues_repo_head_hexsha": "e9ee73da6ecc515ecd471cb9ae059a4370a51493", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/mph2ms.r", "max_forks_repo_name": "alfcrisci/biometeoR", "max_forks_repo_head_hexsha": "e9ee73da6ecc515ecd471cb9ae059a4370a51493", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 16.380952381, "max_line_length": 101, "alphanum_fraction": 0.6453488372, "num_tokens": 118, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.7879311856832191, "lm_q2_score": 0.6442250996557036, "lm_q1q2_score": 0.5076050466186085}} {"text": "#Internal function that permutes the treatment indicator according to biased-coin randomization.\n#This simply flips N-many biased coins to get a new treatment indicator;\n#it ensures that treatment is not all 1s or 0s\npermuteData.biasedCoin = function(N, probs){\n\n permutation = stats::rbinom(n = N, size = 1, prob = probs)\n\n #ensure that the treatment indicator is not all 1s or 0s\n while( sum(permutation) == 0 | sum(permutation) == N ){\n permutation = stats::rbinom(n = N, size = 1, prob = probs )\n }\n\n return(permutation)\n}\n\n#Internal function for permuting an indicator\n#(instrument or exposure) within a subclass.\n#This function takes a table(subclass, indicator) object.\ngetBlockPerm = function(subclassIndicatorTable){\n #number of subclasses\n Ns = nrow(subclassIndicatorTable)\n #total number of units\n N = sum(subclassIndicatorTable)\n #for each subclass, create a permuted indicator,\n #according to the number of 1s and 0s in that subclass.\n permutedIndicator = vector()\n for(s in 1:Ns){\n permutedIndicator.s = sample(c(rep(0, subclassIndicatorTable[s,\"0\"]), rep(1, subclassIndicatorTable[s,\"1\"])))\n permutedIndicator = append(permutedIndicator, permutedIndicator.s)\n }\n return(permutedIndicator)\n}\n\n#Internal function that returns the covariate mean differences\n#across many permutations of an indicator.\ngetCompletePerms.meanDiffs = function(X, indicator, perms = 1000){\n #number of covariates\n K = ncol(X)\n\n #observed standardized covariate mean differences\n covMeanDiff.obs = getCovMeanDiffs(X, indicator)\n\n #permutations for the randomization test\n indicator.permutations = replicate(perms, sample(indicator), simplify = FALSE)\n #compute the vector of covariate mean differences for each permutation\n permutations.covMeanDiffs = matrix(nrow = perms, ncol = K)\n for(i in 1:perms){\n permutations.covMeanDiffs[i,] = getCovMeanDiffs(X, indicator.permutations[[i]])\n }\n return(permutations.covMeanDiffs)\n}\n\n#Internal function that returns the standardized covariate mean differences\n#across many permutations of an indicator.\ngetCompletePerms.balance = function(X, indicator, perms = 1000){\n #number of covariates\n K = ncol(X)\n\n #observed standardized covariate mean differences\n standardizedCovMeanDiff.obs = getStandardizedCovMeanDiffs(X, indicator)\n\n #permutations for the randomization test\n indicator.permutations = replicate(perms, sample(indicator), simplify = FALSE)\n #compute the vector of covariate mean differences for each permutation\n permutations.standardizedCovMeanDiffs = matrix(nrow = perms, ncol = K)\n for(i in 1:perms){\n permutations.standardizedCovMeanDiffs[i,] = getStandardizedCovMeanDiffs(X, indicator.permutations[[i]])\n }\n return(permutations.standardizedCovMeanDiffs)\n}\n\n#Internal function that returns the Mahalanobis distance (MD)\n#across many permutations of an indicator.\ngetCompletePerms.md = function(X, indicator, perms = 1000){\n #to efficiently compute the MD across permutations, it'll be helpful\n #to compute the inverse of the covariate covariance matrix\n #(which doesn't change across permutations)\n covX.inv = solve(as.matrix(stats::cov(X)))\n\n #permutations for the randomization test\n indicator.permutations = replicate(perms, sample(indicator), simplify = FALSE)\n #compute the vector of covariate mean differences for each permutation\n permutations.md = vector(length = perms)\n for(i in 1:perms){\n permutations.md[i] = getMD(X, indicator.permutations[[i]], covX.inv)\n }\n return(permutations.md)\n}\n\n#Internal function that returns the sum of absolute biases\n#across many permutations of an indicator.\n#Note that the bias is different for the exposure (D) than for the instrument (Z),\n#as discussed in Equation (3) of Branson and Keele (2020).\ngetCompletePerms.absBias = function(X, D = NULL, Z = NULL, perms = 1000){\n if(is.null(D) & is.null(Z)){\n print(\"Error: Need to provide D or D and Z indicators.\")\n }\n #if the exposure is provided, the absolute bias is just the absolute standardized covariate mean differences\n if(!is.null(D) & is.null(Z)){\n absCovMeanDiffs = abs(getCompletePerms.meanDiffs(X = X, indicator = D, perms = perms))\n #take the sum across covariates\n absCovMeanDiffs.sum = rowSums(absCovMeanDiffs)\n return(absCovMeanDiffs.sum)\n }\n #if the instrument is provided, the absolute bias is the standardized covariate mean difference\n #divided by the mean exposure difference\n #Thus, we also need the exposure to be provided.\n if( !is.null(Z) & is.null(D) ){\n return( \"Error: to compute instrument bias, also need D indicator.\")\n }\n else{\n #the exposure mean difference is\n DMeanDiff = mean(D[Z == 1]) - mean(D[Z == 0])\n #then, the absolute biases are\n absBias = abs(getCompletePerms.meanDiffs(X = X, indicator = Z, perms = perms)/DMeanDiff)\n absBias.sum = rowSums(absBias)\n return(absBias.sum)\n }\n}\n\n#Internal function that returns the Mahalanobis distance (MD)\n#across many block permutations of an indicator within a subclass.\ngetBlockPerms.md = function(X, indicator, subclass, perms = 1000){\n #for efficiency purposes, it'll be helpful to order the data by subclass\n\n #collect covariates, indicator, and subclass into a dataframe\n data = data.frame(X, indicator = indicator, subclass = subclass)\n\n #order the data by subclass\n data = data[order(subclass),]\n\n #Now, the new X, indicator, and subclass are\n X = as.matrix(subset(data, select = -c(indicator, subclass)))\n covX.inv = solve(as.matrix(stats::cov(X)))\n indicator = data$indicator\n subclass = data$subclass\n\n #To efficiently get block permutations,\n #we just need a table of the indicator and subclass\n subclassIndicatorTable = table(subclass, indicator)\n\n #Then, the set of block permutations is\n indicator.permutations = t(replicate(perms,\n getBlockPerm(subclassIndicatorTable), simplify = TRUE))\n #compute the vector of covariate mean differences for each permutation\n permutations.md = vector(length = perms)\n for(i in 1:perms){\n permutations.md[i] = getMD(X, indicator.permutations[i,], covX.inv)\n }\n return(permutations.md)\n}\n\n#Internal function that returns the sum of absolute biases\n#across many block permutations of an indicator within a subclass.\n#Note that the bias is different for the exposure (D) than for the instrument (Z),\n#as discussed in Equation (3) of Branson and Keele (2020).\ngetBlockPerms.absBias = function(X, D = NULL, Z = NULL, subclass = NULL, perms = 1000){\n if(is.null(subclass)){\n print(\"Error: Need to provide subclass vector for block randomization.\")\n }\n if(is.null(D) & is.null(Z)){\n print(\"Error: Need to provide D or D and Z indicators.\")\n }\n \n #permutations for the randomization test\n if(!is.null(D) & is.null(Z)){\n indicator = D\n }\n else{\n indicator = Z\n }\n #for efficiency purposes, it'll be helpful to order the data by subclass\n\n #collect covariates, indicator, and subclass into a dataframe\n data = data.frame(X, indicator = indicator, subclass = subclass)\n\n #order the data by subclass\n data = data[order(subclass),]\n\n #Now, the new X, indicator, and subclass are\n X = as.matrix(subset(data, select = -c(indicator, subclass)))\n covX.inv = solve(as.matrix(stats::cov(X)))\n indicator = data$indicator\n subclass = data$subclass\n\n #To efficiently get block permutations,\n #we just need a table of the indicator and subclass\n subclassIndicatorTable = table(subclass, indicator)\n\n #Then, the set of block permutations is\n indicator.permutations = t(replicate(perms,\n getBlockPerm(subclassIndicatorTable), simplify = TRUE))\n\n #if the exposure is provided, the absolute bias is just the absolute standardized covariate mean differences\n if(!is.null(D) & is.null(Z)){\n #compute the vector of covariate mean differences for each permutation\n permutations.covMeanDiffs = matrix(nrow = perms, ncol = ncol(X))\n for(i in 1:perms){\n permutations.covMeanDiffs[i,] = getCovMeanDiffs(X, indicator.permutations[i,])\n }\n absCovMeanDiffs = abs(permutations.covMeanDiffs)\n #take the sum across covariates\n absCovMeanDiffs.sum = rowSums(absCovMeanDiffs)\n return(absCovMeanDiffs.sum)\n }\n #if the instrument is provided, the absolute bias is the standardized covariate mean difference\n #divided by the mean exposure difference\n #Thus, we also need the exposure to be provided.\n if( !is.null(Z) & is.null(D) ){\n return( \"Error: to compute instrument bias, also need D indicator.\")\n }\n else{\n #the exposure mean difference is\n DMeanDiff = mean(D[Z == 1]) - mean(D[Z == 0])\n #then, the absolute biases are\n permutations.covMeanDiffs = matrix(nrow = perms, ncol = ncol(X))\n for(i in 1:perms){\n permutations.covMeanDiffs[i,] = getCovMeanDiffs(X, indicator.permutations[i,])\n }\n absBias = abs(permutations.covMeanDiffs/DMeanDiff)\n absBias.sum = rowSums(absBias)\n return(absBias.sum)\n }\n}\n\n#Internal function that returns the Mahalanobis distance (MD)\n#across many Bernoulli-trial permutations of an indicator.\ngetBernoulliPerms.md = function(X, indicator, perms = 1000){\n #to efficiently compute the MD across permutations, it'll be helpful\n #to compute the inverse of the covariate covariance matrix\n #(which doesn't change across permutations)\n covX.inv = solve(as.matrix(stats::cov(X)))\n\n #observed Mahalanobis distance is\n md.obs = getMD(X, indicator, covX.inv)\n\n #need to ensure that X is a matrix to compute the propensity scores\n X = as.matrix(X)\n #model for the propensity scores\n psModel = glm(indicator~X, family = \"binomial\")\n #the propensity scores\n ps = stats::predict(psModel, type = \"response\")\n\n #permutations for the randomization test\n indicator.permutations = replicate(perms,\n permuteData.biasedCoin(N = length(indicator), probs = ps),\n simplify = FALSE)\n #compute the vector of covariate mean differences for each permutation\n permutations.md = vector(length = perms)\n for(i in 1:perms){\n permutations.md[i] = getMD(X, indicator.permutations[[i]], covX.inv)\n }\n return(permutations.md)\n}\n\n#Internal function that returns the sum of absolute biases\n#across many Bernoulli-trial permutations of an indicator.\n#Note that the bias is different for the exposure (D) than for the instrument (Z),\n#as discussed in Equation (3) of Branson and Keele (2020).\ngetBernoulliPerms.absBias = function(X, D = NULL, Z = NULL, perms = 1000){\n if(is.null(D) & is.null(Z)){\n print(\"Error: Need to provide D or D and Z indicators.\")\n }\n #need to ensure that X is a matrix to compute the propensity scores\n X = as.matrix(X)\n #model for the propensity scores\n if(!is.null(D) & is.null(Z)){\n psModel = glm(D~X, family = \"binomial\")\n }\n else{\n psModel = glm(Z~X, family = \"binomial\")\n }\n #the propensity scores\n ps = stats::predict(psModel, type = \"response\")\n\n #permutations for the randomization test\n if(!is.null(D) & is.null(Z)){\n indicator.permutations = replicate(perms,\n permuteData.biasedCoin(N = length(D), probs = ps),\n simplify = FALSE)\n }\n else{\n indicator.permutations = replicate(perms,\n permuteData.biasedCoin(N = length(Z), probs = ps),\n simplify = FALSE)\n }\n #if the exposure is provided, the absolute bias is just the absolute standardized covariate mean differences\n if(!is.null(D) & is.null(Z)){\n #compute the vector of covariate mean differences for each permutation\n permutations.covMeanDiffs = matrix(nrow = perms, ncol = ncol(X))\n for(i in 1:perms){\n permutations.covMeanDiffs[i,] = getCovMeanDiffs(X, indicator.permutations[[i]])\n }\n absCovMeanDiffs = abs(permutations.covMeanDiffs)\n #take the sum across covariates\n absCovMeanDiffs.sum = rowSums(absCovMeanDiffs)\n return(absCovMeanDiffs.sum)\n }\n #if the instrument is provided, the absolute bias is the standardized covariate mean difference\n #divided by the mean exposure difference\n #Thus, we also need the exposure to be provided.\n if( !is.null(Z) & is.null(D) ){\n return( \"Error: to compute instrument bias, also need D indicator.\")\n }\n else{\n #the exposure mean difference is\n DMeanDiff = mean(D[Z == 1]) - mean(D[Z == 0])\n #then, the absolute biases are\n permutations.covMeanDiffs = matrix(nrow = perms, ncol = ncol(X))\n for(i in 1:perms){\n permutations.covMeanDiffs[i,] = getCovMeanDiffs(X, indicator.permutations[[i]])\n }\n absBias = abs(permutations.covMeanDiffs/DMeanDiff)\n absBias.sum = rowSums(absBias)\n return(absBias.sum)\n }\n}", "meta": {"hexsha": "991f2465007f6774e398586515699bee5b51e9f8", "size": 12485, "ext": "r", "lang": "R", "max_stars_repo_path": "R/internalFuncBransonKeele.r", "max_stars_repo_name": "hyunseungkang/ivmodel", "max_stars_repo_head_hexsha": "322a6759f381d59a061cd6c39883b2f851c30029", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 8, "max_stars_repo_stars_event_min_datetime": "2018-09-22T13:38:52.000Z", "max_stars_repo_stars_event_max_datetime": "2021-06-23T02:04:23.000Z", "max_issues_repo_path": "R/internalFuncBransonKeele.r", "max_issues_repo_name": "cran/ivmodel", "max_issues_repo_head_hexsha": "ed4b9ddab888f520dddbb15334044481c0a2c507", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2020-11-09T19:20:30.000Z", "max_issues_repo_issues_event_max_datetime": "2022-02-01T13:44:24.000Z", "max_forks_repo_path": "R/internalFuncBransonKeele.r", "max_forks_repo_name": "hyunseungkang/ivmodel", "max_forks_repo_head_hexsha": "322a6759f381d59a061cd6c39883b2f851c30029", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-06-01T16:33:38.000Z", "max_forks_repo_forks_event_max_datetime": "2020-06-01T16:33:38.000Z", "avg_line_length": 39.3848580442, "max_line_length": 113, "alphanum_fraction": 0.7245494594, "num_tokens": 3253, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8056321796478255, "lm_q2_score": 0.6297746004557471, "lm_q1q2_score": 0.5073666840520019}} {"text": ".bw.scott.product <- function(x) {\n sds <- apply(x, 2, sd)\n n <- nrow(x)\n d <- ncol(x)\n H <- diag(sds) * n ^ (-1 / (d + 4))\n return(H)\n}\n\n#' Fit a kernel density estimate\n#'\n#' description\n#' @param x Matrix or vector of samples. For matrices, rows are samples and columns are variables.\n#' @param adjust Scalar multiplication of the bandwidth\n#' @param bw.fn Function used to calculate the bandwidth. If diagonal = T, bw.fn should accept a vector of 1-dimensional values. If diagonal = F, bw.fn should accept a matrix of samples.\n#' @param diagonal If true, estimate a diagonal bandwidth matrix; otherwise estimate a full bandwidth matrix.\n#' @param verbose No effect.\n#' @param ... Further arguments are passed to bw.fn\n#' @export\n#' @examples\n#' x <- mvtnorm::rmvnorm(50, c(0, 0), rbind(c(1, 0.5), c(0.5, 1)))\n#' fit <- fit.kde(x)\n#' fit <- fit.kde(x, bw.fn = ks::Hpi, diagonal = FALSE)\nfit.kde <- function(x, adjust = 1, diagonal = ncol(x) > 6,\n bw.fn = if(ncol(x) < 6) { if(diagonal) ks::Hpi.diag else ks::Hpi } else { if(diagonal) .bw.scott.product else NULL},\n verbose = FALSE, ...)\n{\n if (is.null(bw.fn)) {\n if (!diagonal && ncol(x) > 6) {\n warning(\"A non-diagonal KDE is selected for >6 dimensions, for which a bandwidth selection method is not implemented. Please supply a custom bandwidth selection method or choose a diagonal bandwidth.\")\n } else {\n warning(\"NULL bandwidth function\")\n }\n return(NULL)\n }\n n <- nrow(x)\n d <- ncol(x)\n \n result <- list()\n result$type <- \"kde\"\n result$dim <- ncol(x)\n result$x <- x\n \n if (diagonal) {\n result$H <- .bw.scott.product(x)\n } else {\n result$H <- adjust * bw.fn(x, ...)\n }\n return(structure(result, class = \"mvd.density\"))\n}\n\n#' Fit a kernel density estimate after transforming the variables to an unbounded domain.\n#'\n#' The following transformations are used\n#' [0,inf] -> log(x)\n#' [-inf,0] -> log(-x)\n#' [0,1] -> logit(x) = log(x / (1 - x))\n#' [a,b] -> scaled logit(x) = log((a - x) / (x - b)) \n#' @param x Matrix or vector of samples. For matrices, rows are samples and columns are variables.\n#' @param bounds Dx2 matrix specifying the lower and upper bound for each variable.\n#' @param adjust Scalar multiplication of the bandwidth\n#' @param bw.fn Function used to calculate the bandwidth. If diagonal = T, bw.fn should accept a vector of 1-dimensional values. If diagonal = F, bw.fn should accept a matrix of samples.\n#' @param diagonal If true, estimate a diagonal bandwidth matrix; otherwise estimate a full bandwidth matrix.\n#' @param verbose No effect.\n#' @export\n#' @examples\n#' x <- exp(mvtnorm::rmvnorm(50, c(0, 0), rbind(c(1, 0.5), c(0.5, 1))))\n#' fit <- fit.kde(x, rbind(c(0, -Inf), c(0, Inf)))\n#' fit <- fit.kde(x, rbind(c(0, -Inf), c(0, Inf)), bw.fn = ks::Hpi, diagonal = FALSE)\nfit.kde.transformed <- function(x, bounds, adjust = 1, bw.fn = bw.SJ, diagonal = TRUE, verbose = FALSE)\n{\n result <- list()\n result$type <- \"kde.transformed\"\n result$transform.bounds <- bounds\n transformed <- mvd.transform_to_unbounded(x, bounds)\n result$kde <- fit.kde(transformed, adjust = adjust, bw.fn = bw.fn, diagonal = diagonal)\n return(structure(result, class = \"mvd.density\"))\n}\n\n.evaluate.kde <- function(fit, x, log = FALSE)\n{\n tp <- rep(NA, nrow(x))\n for (i in 1:nrow(x)) {\n p <- mvtnorm::dmvnorm(fit$x, x[i,], fit$H)\n if (log) {\n tp[i] <- log(sum(p)) - log(nrow(fit$x))\n } else {\n tp[i] <- sum(p) / nrow(fit$x)\n }\n }\n return(tp)\n}\n", "meta": {"hexsha": "ce696c03bb586469d7d6ef4c3a112ce6cf48bd47", "size": 3512, "ext": "r", "lang": "R", "max_stars_repo_path": "R/kde.r", "max_stars_repo_name": "bramthijssen/mvdens", "max_stars_repo_head_hexsha": "27a31caef525dcdd97eaa89e6664f60036b1320c", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/kde.r", "max_issues_repo_name": "bramthijssen/mvdens", "max_issues_repo_head_hexsha": "27a31caef525dcdd97eaa89e6664f60036b1320c", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2020-04-21T12:45:03.000Z", "max_issues_repo_issues_event_max_datetime": "2020-04-21T12:45:03.000Z", "max_forks_repo_path": "R/kde.r", "max_forks_repo_name": "NKI-CCB/mvdens", "max_forks_repo_head_hexsha": "27a31caef525dcdd97eaa89e6664f60036b1320c", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 38.1739130435, "max_line_length": 207, "alphanum_fraction": 0.6332574032, "num_tokens": 1046, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8198933359135361, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.50733395402027}} {"text": "# [Code Pertama, Hello World!]\r\n\"Hello World\"\r\n1 + 5\r\n\r\n# [Teks, Angka dan Rumus Perhitungan]\r\n9\r\n\"Budi\"\r\n9 * 3\r\n\r\n# [Menampilkan dengan Fungsi Print]\r\nprint(\"Hello World\")\r\nprint(3 + 4)\r\n\r\n# [Huruf Besar, Huruf Kecil dan Format Angka]\r\n01\r\n1\r\n\"01-01-1980\"\r\n\"1-1-1980\"\r\n\"Budi\"\r\n\"BUDI\"\r\n\r\n# [Function]\r\nc(5:10)\r\n\r\n# [Variable]\r\nbudi_berat_kg <- 68\r\nsanti_berat_kg <- 54.5\r\nbudi_berat_kg\r\nsanti_berat_kg\r\npi <- 3.14\r\npi\r\n\r\n# [Comment pada R]\r\n2 + 2 # Ini adalah baris komentar\r\n\r\n# [Vector](\r\n# Ini adalah contoh vector untuk angka numerik dengan 3 data c(4, 5, 6)\r\nc(4, 5, 6)\r\n# Variable bernama angka dengan input berupa vector\r\nangka <- c(4, 5, 6)\r\n# Tampilkan isi variable angka dengan fungsi print\r\nprint(angka)\r\n\r\n\r\n# [Deretan Nilai dengan Operator :]\r\nangka1 <- c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10)\r\nprint(angka1)\r\nangka2 <- c(1:10)\r\nprint(angka2)\r\n\r\n\r\n# [Vector dengan Isi Teks]\r\n# Variable nama_mahasiswa dengan input character\r\nnama_mahasiswa <- c(\"Amira\",\"Budi\",\"Charlie\")\r\nprint(nama_mahasiswa)\r\n\r\n# [Index dan Accessor pada Vector]\r\n# Buat vector variable bernama angka yang isinya 20 s/d 30\r\nangka <- c(20:30)\r\nprint(angka)\r\n# Tampilkan isi variable angka pada posisi ke 3\r\nprint(angka[3])\r\n# Tampilkan isi variable angka pada posisi ke 5\r\nprint(angka[[5]])\r\n# Tampilkan isi variable angka pada posisi ke 4 s/d 6\r\nprint(angka[4:6])\r\n# Buat vector teks dengan nama kode_prodi yang diisi sesuai petunjuk soal\r\nkode_prodi <- c(\"DKV\",\"ILKOM\",\"ICT\")\r\n# Tampilkan isi indeks ketiga dari kode_prodi\r\nprint(kode_prodi[3])\r\n\r\n\r\n# [Named Vector]\r\n# Membuat named vector dengan nama nilai\r\nnilai <- c(statistik = 89, \r\n fisika = 95, \r\n ilmukomunikasi = 100)\r\n# Menampilkan isi variable nilai\r\nprint(nilai)\r\n# Menampilkan isi dengan nama fisika\r\nprint(nilai[\"fisika\"])\r\n# Buat variable profil sesuai permintaan soal\r\nprofil <- c(nama = \"Budi\", \r\n tempat_tinggal = \"Jakarta\", \r\n tingkat_pendidikan = \"S1\")\r\n# Tampilkan variable profil\r\nprint(profil)\r\n\r\n\r\n# [List]\r\n# List disimpan dalam variable dengan nama list_random\r\nlist_random <- list(2, \"Budi\", 4)\r\n# Menampilkan isi list\r\nlist_random \r\n# List disimpan dalam variable dengan nama dati2\r\ndati2 <- list(nama = \"Denpasar\", \r\n propinsi = \"Bali\")\r\n# Menampilkan isi list dati2\r\ndati2 \r\n# Buat variable kota sesuai permintaan soal\r\nkota <- list(nama_kota = \"Makassar\", \r\n propinsi = \"Sulawesi Selatan\", \r\n luas_km2 = 199.3)\r\n# Tampilkan isi variable list kota\r\nprint(kota)\r\n\r\n\r\n# [List Index]\r\n# Membentuk list dengan 2 angka dan 1 character\r\nlist_saya <- list(2, \"Budi\", 4)\r\n# Menampilkan index kedua dengan aksesor kurung siku tunggal \r\nlist_saya[2]\r\n# Menampilkan index kedua dengan aksesor kurung siku ganda\r\nlist_saya[[2]]\r\n# Menampilkan index kedua s/d ketiga\r\nlist_saya[2:3]\r\n\r\nlist_satu <- list(1, \"Online\", TRUE)\r\nlist_satu[1]\r\n\r\n\r\n# [Data Frame]\r\n# Membuat dua variable vector\r\nfakultas <- c(\"Bisnis\", \"D3 Perhotelan\", \"ICT\", \"Ilmu Komunikasi\", \"Seni dan Desain\")\r\njumlah_mahasiswa <- c(260, 28, 284, 465, 735)\r\n# Membuat data frame dari kedua vector di atas\r\ninfo_mahasiswa <- data.frame(fakultas, jumlah_mahasiswa)\r\n# Melihat isi data frame\r\ninfo_mahasiswa\r\n# Buat vector baru sebagai representasi akreditasi\r\nakreditasi <- c(\"A\",\"A\",\"B\",\"A\",\"A\")\r\n# Buat data frame dari ketiga vector di atas\r\ninfo_mahasiswa <- data.frame(info_mahasiswa, akreditasi)\r\ninfo_mahasiswa\r\n\r\n\r\n# [Cara Akses Data Frame]\r\n# Membuat tiga variable vector\r\nfakultas <- c(\"Bisnis\", \"D3 Perhotelan\", \"ICT\", \"Ilmu Komunikasi\", \"Seni dan Desain\")\r\njumlah_mahasiswa <- c(260, 28, 284, 465, 735)\r\nakreditasi <- c(\"A\",\"A\",\"B\",\"A\",\"A\")\r\n# Membuat data frame dari kedua vector di atas\r\ninfo_mahasiswa <- data.frame(fakultas, jumlah_mahasiswa, akreditasi)\r\n# Menampilkan kolom jumlah_mahasiswa\r\ninfo_mahasiswa$jumlah_mahasiswa\r\n# Menampilkan kolom fakultas\r\ninfo_mahasiswa$fakultas\r\n\r\n\r\n# [Package ggplot2]\r\nfakultas <- c(\"Bisnis\", \"D3 Perhotelan\", \"ICT\", \"Ilmu Komunikasi\", \"Seni dan Desain\")\r\njumlah_mahasiswa <- c(260, 28, 284, 465, 735)\r\nakreditasi <- c(\"A\",\"A\",\"B\",\"A\",\"A\")\r\ninfo_mahasiswa <- data.frame(fakultas, jumlah_mahasiswa, akreditasi)\r\ninfo_mahasiswa\r\n# Menggunakan package ggplot2\r\nlibrary(\"ggplot2\")\r\n# Membuat kanvas\r\ngambar <- ggplot(info_mahasiswa, aes(x=fakultas, y=jumlah_mahasiswa, fill=fakultas)) + \r\n geom_bar(width=1, stat=\"identity\")\r\ngambar\r\n\r\n\r\n# [Membuat Grafik Sebaran Mahasiswa (1)]\r\n# Membuat dua vector\r\nfakultas <- c(\"Bisnis\", \"D3 Perhotelan\", \"ICT\", \"Ilmu Komunikasi\", \"Seni dan Desain\")\r\njumlah_mahasiswa <- c(260, 28, 284, 465, 735)\r\nakreditasi <- c(\"A\",\"A\",\"B\",\"A\",\"A\")\r\n# Buat data frame dari ketiga vector di atas\r\ninfo_mahasiswa <- data.frame(fakultas, jumlah_mahasiswa, akreditasi)\r\ninfo_mahasiswa\r\n# Menggunakan package ggplot2\r\nlibrary(ggplot2)\r\n# Membuat kanvas\r\ngambar <- ggplot(info_mahasiswa, aes(x=fakultas, y=jumlah_mahasiswa, fill=fakultas))\r\n# Menambahkan objek bar chart, simpan kembali sebagai variable gambar\r\ngambar <- gambar + geom_bar(width=1, stat=\"identity\")\r\n# Menambahkan judul grafik\r\ngambar <- gambar + ggtitle(\"Jumlah Mahasiswa per Fakultas\")\r\n# Menambahkan caption pada sumbu x\r\ngambar <- gambar + xlab(\"Nama Fakultas\")\r\n# Menambahkan caption pada sumbu y\r\ngambar <- gambar + ylab(\"Jumlah Mahasiswa\")\r\n# Menggambar grafik\r\ngambar\r\n\r\n\r\n# [Membaca File Excel]\r\n# Menggunakan package ggplot2\r\nlibrary(ggplot2)\r\n# Menggunakan package openxlsx\r\nlibrary(openxlsx)\r\n# Membaca file mahasiswa.xlsx\r\nmahasiswa <- read.xlsx(\"https://academy.dqlab.id/dataset/mahasiswa.xlsx\", sheet = \"Sheet 1\")\r\n# Menampilkan data\r\nprint(mahasiswa)\r\n# Menampilkan kolom Prodi\r\nprint(mahasiswa$Prodi)\r\n\r\n\r\n# [Membuat Grafik Sebaran Mahasiswa (2)]\r\nlibrary(ggplot2)\r\n# Menggunakan package openxlsx\r\nlibrary(openxlsx)\r\n# Membaca file mahasiswa.xlsx\r\nmahasiswa <- read.xlsx(\"https://academy.dqlab.id/dataset/mahasiswa.xlsx\", sheet = \"Sheet 1\")\r\n# Membuat kanvas\r\ngambar <- ggplot(mahasiswa, aes(x=Fakultas, y=JUMLAH, fill=Fakultas))\r\n# Menambahkan objek bar chart, simpan kembali sebagai variable gambar\r\ngambar <- gambar + geom_bar(width=1, stat=\"identity\")\r\n# Menggambar grafik\r\ngambar\r\n\r\n\r\n# [Trend Jumlah Mahasiswa dari Tahun ke Tahun]\r\nlibrary(ggplot2)\r\n# Menggunakan package openxlsx\r\nlibrary(openxlsx)\r\n# Membaca file mahasiswa.xlsx\r\nmahasiswa <- read.xlsx(\"https://academy.dqlab.id/dataset/mahasiswa.xlsx\", sheet = \"Sheet 1\")\r\n# Menghitung Jumlah Data by Fakultas\r\nsummarybyfakultas <- aggregate(x=mahasiswa$JUMLAH, \r\n by=list(Kategori=mahasiswa$Fakultas, Tahun=mahasiswa$ANGKATAN), \r\n FUN=sum)\r\nsummarybyfakultas <- setNames(summarybyfakultas, c(\"fakultas\",\"tahun\", \"jumlah_mahasiswa\"))\r\nsummarybyfakultas\r\n\r\nsummarybyfakultas$tahun = as.factor(summarybyfakultas$tahun)\r\n\r\nggplot(summarybyfakultas, aes(x=fakultas, y=jumlah_mahasiswa)) + \r\n geom_bar(stat=\"identity\", aes(fill=tahun), width=0.8, position=position_dodge(width=0.8)) + \r\n theme_classic()\r\n\r\n\r\n# [Pie Chart]\r\nlibrary(ggplot2)\r\nlibrary(openxlsx)\r\n# Membaca file mahasiswa.xlsx\r\nmahasiswa <- read.xlsx(\"https://academy.dqlab.id/dataset/mahasiswa.xlsx\", sheet = \"Sheet 1\")\r\n# Menghitung Jumlah Data by Fakultas\r\nsummarybyfakultas <- aggregate(x=mahasiswa$JUMLAH, \r\n by=list(Kategori=mahasiswa$Fakultas), \r\n FUN=sum)\r\nsummarybyfakultas <- setNames(summarybyfakultas, \r\n c(\"fakultas\",\"jumlah_mahasiswa\"))\r\n\r\npiechart<- ggplot(summarybyfakultas, aes(x=\"\", y=jumlah_mahasiswa, fill=fakultas)) + \r\n geom_bar(width = 1, stat = \"identity\")\r\n\r\npiechart <- piechart + coord_polar(\"y\", start=0)\r\npiechart <- piechart + ggtitle(\"Disribusi Mahasiswa per Fakultas\")\r\npiechart <- piechart + scale_fill_brewer(palette=\"Blues\") + theme_minimal()\r\npiechart <- piechart + guides(fill=guide_legend(title=\"Fakultas\"))\r\npiechart <- piechart + ylab(\"Jumlah Mahasiswa\") \r\npiechart\r\n\r\n\r\n# [Filtering]\r\nlibrary(\"ggplot2\")\r\nlibrary(\"openxlsx\")\r\n# Membaca file mahasiswa.xlsx\r\nmahasiswa <- read.xlsx(\"https://academy.dqlab.id/dataset/mahasiswa.xlsx\", sheet = \"Sheet 1\")\r\n# Menghitung Jumlah Data by Fakultas\r\nsummarybyfakultas <- aggregate(x=mahasiswa$JUMLAH, \r\n by=list(Kategori=mahasiswa$Fakultas, Tahun=mahasiswa$ANGKATAN), \r\n FUN=sum)\r\nsummarybyfakultas <- setNames(summarybyfakultas, c(\"fakultas\",\"tahun\", \"jumlah_mahasiswa\"))\r\nsummarybyfakultas\r\n\r\nsummarybyfakultas$tahun = as.factor(summarybyfakultas$tahun)\r\nsummarybyfakultas[summarybyfakultas$fakultas %in% c(\"ICT\", \"Ilmu Komunikasi\"),]\r\n\r\nggplot(summarybyfakultas[summarybyfakultas$fakultas %in% c(\"ICT\", \"Ilmu Komunikasi\"),], \r\n aes(x=fakultas, y=jumlah_mahasiswa)) + \r\n geom_bar(stat = \"identity\", aes(fill = tahun), width=0.8, position = position_dodge(width=0.8)) + \r\n theme_classic() ", "meta": {"hexsha": "add6da2bc2f53631fcc10b57e7c0492e9da778f3", "size": 8847, "ext": "r", "lang": "R", "max_stars_repo_path": "R/Kelas Persiapan/R Fundamental for Data Science.r", "max_stars_repo_name": "Miadwicynthia/DQLab", "max_stars_repo_head_hexsha": "711e9650224683844a67963e76b10c1d4a38ebf4", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/Kelas Persiapan/R Fundamental for Data Science.r", "max_issues_repo_name": "Miadwicynthia/DQLab", "max_issues_repo_head_hexsha": "711e9650224683844a67963e76b10c1d4a38ebf4", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/Kelas Persiapan/R Fundamental for Data Science.r", "max_forks_repo_name": "Miadwicynthia/DQLab", "max_forks_repo_head_hexsha": "711e9650224683844a67963e76b10c1d4a38ebf4", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 32.0543478261, "max_line_length": 101, "alphanum_fraction": 0.6958290946, "num_tokens": 2800, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.6334102636778401, "lm_q2_score": 0.8006920044739461, "lm_q1q2_score": 0.5071665336785806}} {"text": "s = 0\n\ni = 1\n\nwhile(i<11){\n s = s + i\n i = i + 1\n}\ncat(\"sum is\",s)", "meta": {"hexsha": "1926bc559fc01da29b6ec1a91bd46dfc56969361", "size": 68, "ext": "r", "lang": "R", "max_stars_repo_path": "Basics/control_structures/while.r", "max_stars_repo_name": "DivyaMaddipudi/R-Programming-Basics", "max_stars_repo_head_hexsha": "28546e2f159b98bd8e94503b2a2e07aef68e24f9", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Basics/control_structures/while.r", "max_issues_repo_name": "DivyaMaddipudi/R-Programming-Basics", "max_issues_repo_head_hexsha": "28546e2f159b98bd8e94503b2a2e07aef68e24f9", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Basics/control_structures/while.r", "max_forks_repo_name": "DivyaMaddipudi/R-Programming-Basics", "max_forks_repo_head_hexsha": "28546e2f159b98bd8e94503b2a2e07aef68e24f9", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 7.5555555556, "max_line_length": 15, "alphanum_fraction": 0.3970588235, "num_tokens": 38, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.6959583250334526, "lm_q2_score": 0.727975460709318, "lm_q1q2_score": 0.5066405823007128}} {"text": "## estimate redshift offsets\n\ngetdz <- function(gdat, lib, snrthresh=5, nlthresh=2000, dzlim=0.003, searchinterval=1e-4) {\n nr <- nrow(gdat$flux)\n dz <- numeric(nr)\n dz.err <- numeric(nr)\n z <- gdat$meta$z\n lambda <- gdat$lambda\n fmla <- as.formula(paste(\"f ~\", paste(names(lib)[-1], collapse=\"+\")))\n fn <- function(x) {\n lambda.out <- lambda/(1+z+x)\n lib.out <- regrid(lambda.out, lib)\n lfit <- lm(fmla, data=lib.out, weights=iv)\n deviance(lfit)\n }\n fn_v <- Vectorize(fn)\n z_search <- seq(-dzlim, dzlim, by=searchinterval)\n \n for (i in 1:nr) {\n if (is.na(gdat$snr[i]) || gdat$snr[i]<=snrthresh) {\n dz[i] <- NA\n dz.err[i] <- NA\n next\n }\n f <- gdat$flux[i, ]\n if (length(which(!is.na(f))) < nlthresh) {\n dz[i] <- NA\n dz.err[i] <- NA\n next\n }\n iv <- gdat$ivar[i, ]\n dev_grid <- fn_v(z_search)\n z0 <- z_search[which.min(dev_grid)]\n bestz <- Rsolnp::solnp(pars=z0, fun=fn, \n LB=z0-searchinterval, UB=z0+searchinterval, \n control=list(trace=0))\n dz[i] <- bestz$pars\n dz.err[i] <- sqrt(2/bestz$hessian)\n }\n list(dz=dz, dz.err=dz.err)\n}\n\n## Moore-Penrose pseudo inverse (needed for uncertainty estimates in fit.nn)\n\nmpinv <- function(X) {\n S <- svd(X)\n eps <- .Machine$double.eps * max(dim(X)) * S$d[1]\n dinv <- numeric(length(S$d))\n dinv[S$d >= eps] <- 1/S$d[S$d >= eps]\n tcrossprod(S$v %*% diag(dinv), S$u)\n}\n\n\n## nnls fits to manga data cube or rss file\n\nnnfitmanga <- function(gdat, dz,\n nz=length(Z), nt=length(ages),\n snrthresh=5, tsf=0.1, rlaw=calzetti, dlogl=1.e-4, \n starts = c(0.25, 1., 1.), lb=c(0, 0.7, 0), ub=c(3., 5., 5.),\n flux.em.bad=1.e5,\n which.lick=c(1, 13:15, 20),\n PLOT=TRUE) {\n require(nnls)\n require(cosmo)\n if (PLOT) {\n require(ggplot2)\n require(reshape2)\n options(\"warn\" = -1)\n }\n \n ## variables and function definition for nonlinear fit to get tau, vdisp.st, vdisp.em\n \n fit.nn <- NULL\n x.st <- NULL\n x.em <- NULL\n in.em <- NULL\n allok <- NULL\n fn <- function(pars) {\n tauv <- pars[1]\n vdisp.em <- 100*pars[2]\n vdisp.st <- 100*pars[3]\n x.st <<- blur.lib(lib.ssp, vdisp.st)\n allok <<- complete.cases(flux, ivar, x.st)\n lib.em <- make_emlib(lambda.em, vdisp.em, logl, allok)\n x.em <<- lib.em$x_em\n in.em <<- lib.em$in_em\n att <- as.vector(rlaw(lambda.rest[allok], tauv))\n if (is.null(in.em)) {\n fit.nn <<- nnls(x.st[allok,]*att*sqrt(ivar[allok]), flux[allok]*sqrt(ivar[allok]))\n } else {\n fit.nn <<- nnls(cbind(x.st[allok,]*att, x.em[allok, ])*sqrt(ivar[allok]),\n flux[allok]*sqrt(ivar[allok]))\n }\n fit.nn$deviance/2\n }\n \n dz <- dz$dz\n nr <- length(dz)\n lib.ssp$lambda <- airtovac(lib.ssp$lambda)\n olib.ssp <- lib.ssp\n tauv_err <- rep(NA, nr)\n vdisp.em_err <- rep(NA, nr)\n vdisp.st_err <- rep(NA, nr)\n tauv <- rep(NA, nr)\n vdisp.em <- rep(NA, nr)\n vdisp.st <- rep(NA, nr)\n \n T.gyr <- 10^(ages-9)\n isf <- which.min(abs(tsf-T.gyr))\n lambda.em <- lambda_em\n n.em <- length(lambda.em)\n ##needed to correct for log lambda grid\n em.mult <- lambda.em * log(10)/10000\n n.st <- ncol(lib.ssp)-1\n \n nnfits <- matrix(NA, nrow=nr, ncol=n.em+n.st)\n log_lik <- rep(NA, nr)\n tbar <- rep(NA, nr)\n tbar.lum <- rep(NA, nr)\n zbar <- rep(NA, nr)\n Mstar <- rep(NA, nr)\n gri <- matrix(NA, nrow=nr, ncol=nrow(gri.ssp))\n d4000_n <- rep(NA, nr)\n d4000_n_err <- rep(NA, nr)\n lick <- matrix(NA, nrow=nr, ncol=length(which.lick))\n lick.err <- matrix(NA, nrow=nr, ncol=length(which.lick))\n flux.em <- matrix(NA, nrow=nr, ncol=n.em)\n flux.em.err <- matrix(NA, nrow=nr, ncol=n.em)\n \n for (i in 1:nr) {\n if (is.na(gdat$snr[i]) || gdat$snr[i]<=snrthresh || is.na(dz[i])) {\n next\n }\n lambda <- gdat$lambda\n flux <- gdat$flux[i,]\n ivar <- gdat$ivar[i,]\n z <- gdat$meta$z+dz[i]\n lambda.rest <- lambda/(1+z)\n logl <- log10(lambda.rest)\n lib.ssp <- regrid(lambda.rest, olib.ssp)\n fitij <- Rsolnp::solnp(pars=starts, fn, LB=lb, UB=ub)\n tauv[i] <- fitij$pars[1]\n vdisp.em[i] <- 100*fitij$pars[2]\n vdisp.st[i] <- 100*fitij$pars[3]\n errs <- tryCatch(sqrt(diag(solve(fitij$hessian))), error=function(e) rep(NA, 3))\n tauv_err[i] <- errs[1]\n vdisp.em_err[i] <- 100*errs[2]\n vdisp.st_err[i] <- 100*errs[3]\n \n b <- fit.nn$x\n b.st <- b[1:n.st]\n b.em <- b[(n.st+1):length(b)]\n nnfits[i, 1:length(b)] <- b\n ni.em <- length(b)-n.st\n nl <- length(lambda.rest[allok])\n fitted <- cbind(x.st*as.vector(rlaw(lambda.rest,tauv[i])), x.em) %*% b\n fitted.em <- x.em %*% b.em\n gflux.net <- flux-fitted.em\n residual <- (flux-fitted)*sqrt(ivar)\n \n ## basic measures of goodness of fit \n log_lik[i] <- fit.nn$deviance/2\n \n ## plot spectrum and fit\n tdat <- data.frame(lambda=lambda.rest, gflux=flux, fitted=fitted,\n residual=residual)\n tlong <- melt(tdat, id.vars=\"lambda\")\n val <- c(rep(\" obs\",length(lambda.rest)), rep(\"fitted\",length(lambda.rest)),\n rep(\"residual\",length(lambda.rest)))\n tlong <- cbind(tlong, val = val)\n tlong$variable[tlong$variable==\"gflux\"] <- \"fitted\"\n base <- qplot(lambda, value, data=tlong, geom=\"line\", xlab=expression(lambda), \n ylab=\"\", col=val, \n main=paste(\"(\",i,\") log_lik= \", \n format(log_lik[i], digits=0), sep=\"\"))\n add.resid <- facet_grid(variable ~ ., scale=\"free_y\")\n plot(base+add.resid)\n \n ## emission line fluxes and errors\n flux.em[i,in.em] <- b.em*em.mult[in.em]\n nzeros <- sort(fit.nn$passive)\n X <- cbind(x.st*as.vector(rlaw(lambda.rest,tauv[i])), x.em)[allok,nzeros]\n V <- max(log_lik[i]*2/nl, 1)*mpinv(crossprod(X, diag(ivar[allok])) %*% X)\n sd.b <- rep(NA, n.st+ni.em)\n sd.b[nzeros] <- sqrt(diag(V))\n flux.em.err[i,in.em] <- sd.b[(n.st+1):(n.st+ni.em)]*em.mult[in.em]\n \n ## gri magnitudes\n gri[i,] <- -2.5*log10(gri.ssp %*% b.st) + 20.092\n \n ## summaries from estimated sfh\n tbar[i] <- log10(sum(rep(T.gyr, nz)*b.st)/sum(b.st))+9\n tbar.lum[i] <- log10(sum(rep(T.gyr, nz) * b.st * gri.ssp[\"r\",])/\n sum(b.st * gri.ssp[\"r\",]))+9\n m.st <- cosmo::lum.sol(1, z)*b.st\n Mstar[i] <- log10(sum(m.st*mstar))\n \n ##d4000 and lick indices\n d4 <- d4000n(lambda.rest, gflux.net, ivar)\n d4000_n[i] <- d4$d4000_n\n d4000_n_err[i] <- d4$d4000_n_err\n \n d4 <- lickew(lambda.rest, gflux.net, ivar, which.index=which.lick)\n lick[i,] <- d4[1:length(which.lick)]\n lick.err[i,] <- d4[(length(which.lick)+1):length(d4)] \n } \n \n flux.em[flux.em > flux.em.bad] <- NA\n flux.em.err[!is.finite(flux.em)] <- NA\n dimnames(lick)[[2]] <- as.list(names(d4)[1:length(which.lick)])\n dimnames(lick.err)[[2]] <- as.list(names(d4)[(length(which.lick)+1):length(d4)])\n dimnames(gri)[[2]] <- dimnames(gri.ssp)[[1]]\n dimnames(flux.em)[[2]] <- as.list(names(lambda.em))\n dimnames(flux.em.err)[[2]] <- as.list(paste(names(lambda.em), \"_err\", sep=\"\"))\n options(\"warn\"=0)\n returns <- list(tauv=tauv, vdisp.em=vdisp.em, vdisp.st=vdisp.st, \n tauv_err=tauv_err, vdisp.em_err=vdisp.em_err, vdisp.st_err=vdisp.st_err,\n d4000_n=d4000_n, d4000_n_err=d4000_n_err, \n lick=lick, lick.err=lick.err, \n flux.em=flux.em, flux.em.err=flux.em.err, gri=gri, \n tbar=tbar, tbar.lum=tbar.lum, Mstar=Mstar,\n log_lik=log_lik, nnfits=nnfits)\n returns\n}\n \n \nreplot <- function(gdat, dz, nnfit, lib.ssp, \n which.spax=gdat$meta$cpix, rlaw=calzetti,\n title=NULL) {\n require(ggplot2)\n require(reshape2)\n options(\"warn\" = -1)\n \n i <- which.spax[1]\n j <- which.spax[2]\n flux <- gdat$flux[i, j, ]\n ivar <- gdat$ivar[i, j, ]\n lambda <- gdat$lambda\n lambda.rest <- lambda/(1+gdat$meta$z+dz[i,j])\n logl <- log10(lambda.rest)\n lib.ssp$lambda <- airtovac(lib.ssp$lambda)\n lib.ssp <- regrid(lambda.rest, lib.ssp)\n x.st <- blur.lib(lib.ssp, nnfit$vdisp.st[i,j])\n n.st <- ncol(x.st)\n allok <- complete.cases(flux, ivar, x.st)\n x.st <- x.st[allok, ]\n lambda.rest <- lambda.rest[allok]\n lambda.em <- lambda_em\n x.em <- make_emlib(lambda.em, nnfit$vdisp.em[i,j], allok, logl)\n in.em <- x.em$in.em\n x.em <- x.em$x.em\n b <- nnfit$nnfits[i, j, ]\n b <- b[!is.na(b)]\n b.st <- b[1:n.st]\n b.em <- b[(n.st+1):length(b)]\n fitted <- cbind(x.st*as.vector(rlaw(lambda.rest,nnfit$tauv[i,j])), x.em[allok,]) %*% b\n fitted.em <- x.em[allok,] %*% b.em\n gflux.net <- flux[allok]-fitted.em\n residual <- (flux[allok]-fitted)*sqrt(ivar[allok])\n tdat <- data.frame(lambda=lambda.rest, flux=flux[allok], fitted=fitted,\n fitted.em=fitted.em, residual=residual)\n tlong <- melt(tdat, id.vars=\"lambda\")\n val <- c(rep(\" obs\",length(lambda.rest)), rep(\"fitted\",length(lambda.rest)),\n rep(\"em\", length(lambda.rest)), rep(\"residual\",length(lambda.rest)))\n tlong <- cbind(tlong, val = val)\n tlong$variable[tlong$variable !=\"residual\"] <- \"flux\"\n base <- qplot(lambda, value, data=tlong, geom=\"line\", xlab=expression(lambda), \n ylab=\"\", col=val)\n if (!is.null(title)) {\n base <- base + ggtitle(title)\n }\n add.resid <- facet_grid(variable ~ ., scale=\"free_y\")\n g1 <- base+add.resid\n plot(g1)\n options(\"warn\" = 0)\n g1\n}\n\ngetbpt <- function(nnfits, snthresh=3, PLOT=TRUE) {\n flux.em <- nnfits$flux.em\n flux.em.err <- nnfits$flux.em.err\n dims <- dim(flux.em)\n nr <- dims[1]\n nc <- dims[2]\n \n n2halpha <- as.vector(log10(flux.em[,,'nii_6584']/flux.em[,,'h_alpha']))\n sn.halpha <- as.vector(flux.em[,,'h_alpha']/flux.em.err[,,'h_alpha_err'])\n sn.nii <- as.vector(flux.em[,,'nii_6584']/flux.em.err[,,'nii_6584_err'])\n n2halpha[!is.finite(n2halpha) | sn.halpha 0) {\n loglike = loglike + lfactorial(eta - 1)\n }# END: eta check\n \n }# END: i loop\n \n for(i in 1 : N) {\n loglike = loglike - log(alpha + i - 1)\n }\n \n loglike = loglike + muPrior(mu, K, mu0, P0)\n \n return(loglike)\n}#END: prior\n\nzPrior <- cmpfun(zPrior)\n\n#---PROPOSAL---#\n# zProposal <- function(z, K, N, mu, P, mu0, P0, alpha) {\n# # random walk (symmetric) integer proposal\n# index = sample( c(1 : N), 1)\n# value = sample( c(1 : K), 1 )\n# \n# r.cand = z\n# r.cand[index] = value\n# \n# # on the log scale\n# d.cand = 0\n# d.curr = 0\n# \n# return(list(r.cand = r.cand, d.cand = d.cand, d.curr = d.curr))\n# }# END: proposal\n\nzProposal <- function(z, K, N, mu, P, mu0, P0, alpha) {\n # gibbs proposal (algorithm 2 from Neal 2000)\n r.cand = z\n for(index in 1 : N) {\n \n occupancy = matrix(NA, ncol = K, dimnames = list(NULL, c(1 : K) ) )\n zi = r.cand[ - index]\n theTable = table(zi)\n \n for(i in 1 : K) {\n colname <- colnames(occupancy)[i]\n value <- theTable[which(names(theTable) == colname)]\n value <- ifelse(is.numeric(value), value, 0)\n occupancy[, i] <- ifelse(is.na(value), 0, value)\n }#END: i loop\n \n probs = matrix(NA, ncol = K, dimnames = list(NULL, c(1 : K) ) )\n for(i in 1 : K) {\n \n if(occupancy[i] == 0) {# draw new\n \n # likelihood for unrepresented class: / P(x[index] | mu[i]) * P(mu[i]) dm[i]\n # M-H for poor people: sample from prior for mu (base model), evaluate at likelihood\n # TODO: loop over all existing mus?\n \n like = 0\n M = 1\n for(m in 1 : M) {\n mu.cand = muRand(mu0, P0);#-2.6\n like = like + partialLoglike( x[index], mu.cand, P ) \n }\n \n like = like / M\n prob = ((alpha) / (N - 1 + alpha));\n probs[i] = log( prob ) + like\n \n if(DEBUG) {\n# cat(paste(\"probability for new:\", prob, \"\\n\"));\n cat(paste(\"data:\", x[index], \"\\n\"));\n cat(paste(\"mu candidate:\", mu.cand, \"\\n\"));\n cat(paste(\"stdev:\", P, \"\\n\"));\n cat(paste(\"loglikelihood for new:\", like, \"\\n\"));\n cat(\"\\n\")\n# cat(paste(\"logprobability\", probs[i], \"\\n\\n\"));\n }\n \n } else {# draw existing\n \n # likelihood for components with observations other than x_i currently associated with them is N(mu_j, P)\n like = partialLoglike( x[index], mu[i], P )\n prob = ((occupancy[i]) / (N - 1 + alpha))\n probs[i] = log( prob ) + like\n \n if(DEBUG) {\n# cat(paste(\"probability for existing:\", prob, \"\\n\"));\n cat(paste(\"data:\", x[index], \"\\n\"));\n cat(paste(\"mu[i]:\", mu[i], \"\\n\"));\n cat(paste(\"stdev:\", P, \"\\n\"));\n cat(paste(\"loglikelihood for existing:\", like, \"\\n\"));\n cat(\"\\n\")\n# cat(paste(\"logprobability\", probs[i], \"\\n\\n\"));\n }\n \n }#END: occupation check\n \n }#END: i loop\n \n # rescale to improve accuracy\n # max = max(probs)\n # for(i in 1 : K) {\n # probs[i] = probs[i] - max\n # }\n # \n # # normalize probs (b in Neal 2000)\n # norm = 0;\n # for(i in 1 : K) {\n # norm = norm + probs[i]^2\n # }\n # norm = sqrt( norm )\n # \n # for(i in 1 : K) {\n # probs[i] = probs[i] / norm \n # }\n # \n probs = exp(probs)\n \n value = sample(c(1 : K), size = 1, prob = probs)\n r.cand[index] = value\n }#END: index loop\n \n # on log scale\n d.cand = 0 \n d.curr = 0 \n \n return(list(r.cand = r.cand, d.cand = d.cand, d.curr = d.curr))\n}#END: proposal\n\nzProposal <- cmpfun(zProposal)\n\n###############\n#---SAMPLER---#\n###############\nmetropolisHastings <- function(loglikelihood, prior, proposal, data, zstartvalue, mustartvalue, P, mu0, P0, alpha, Nsim) {\n \n N <- length(zstartvalue)\n K <- length(mustartvalue)\n \n muchain = array(dim = c(Nsim, K))\n muchain[1, ] = mustartvalue\n \n zchain = array(dim = c(Nsim, N))\n zchain[1, ] = zstartvalue\n for (i in 1 : (Nsim - 1)) {\n \n zcandidate = zProposal(z = zchain[i, ], K, N, mu = muchain[i, ], P, mu0, P0, alpha)\n r.zcandidate = zcandidate$r.cand\n d.zcandidate = zcandidate$d.cand\n d.zcurr = zcandidate$d.curr\n \n mucandidate = muProposal(muchain[i, ], operate = TRUE) \n \n r.mucandidate = mucandidate$r.cand\n d.mucandidate = sum(mucandidate$d.cand)\n d.mucurr = sum(mucandidate$d.curr)\n \n probab = exp(\n \n ( loglikelihood(mu = r.mucandidate, z = r.zcandidate, P = P, data) + \n zPrior(r.zcandidate, K, N, r.mucandidate, mu0, P0, alpha) + d.zcandidate + \n muPrior(r.mucandidate, K, mu0, P0) + d.mucandidate\n ) -\n \n ( loglikelihood(mu = muchain[i, ], z = zchain[i, ], P = P, data) + \n zPrior(zchain[i, ], K, N, muchain[i, ], mu0, P0, alpha) + d.zcurr + \n muPrior(muchain[i, ], K, mu0, P0) + d.mucurr\n )\n \n )\n \n if (runif(1) < probab) {\n zchain[i + 1, ] = r.zcandidate\n muchain[i + 1, ] = r.mucandidate\n } else {\n zchain[i + 1, ] = zchain[i, ]\n muchain[i + 1, ] = muchain[i, ]\n }#END: accept check\n \n }#END: iterations loop\n \n return(list(zchain = zchain, muchain = muchain))\n}#END: metropolisHastings\n\nmetropolisHastings <- cmpfun(metropolisHastings)\n\n\n############\n#---MCMC---#\n############\nnum.mode <- function(x) {\n as.numeric(names(which(table(x) == max(table(x)))))\n}\n\nCI <- function (x, ci = 0.95) {\n a = mean(x)\n s = sd(x)\n n = length(x)\n error = qt(ci + (1 - ci)/2, df = n - 1) * s/sqrt(n)\n \n return(c(upper = a + error, mean = a, lower = a - error))\n}\n\nrun <- function() {\n \n Nsim <- 10^3\n N <- length(x)\n P <- 1\n alpha <- 0.01\n z <- rep(1, N)\n K <- 2\n mu <- c(-4, 2)\n mu0 <- mean(x) \n P0 <- sd(x)\n \n chain = metropolisHastings(loglikelihood, prior, proposal, data = x, zstartvalue = z, mustartvalue = mu, P, mu0, P0, alpha, Nsim)\n \n muchain = chain$muchain\n zchain = chain$zchain\n \n ### getting the posterior past burnin\n burnin <- 200\n postMode = apply(zchain[burnin : Nsim, ], 2, num.mode)\n postMean = apply(muchain[burnin : Nsim, ], 2, mean)\n \n probs = rep(NA, length(postMean))\n for(i in 1 : K) {\n probs[i] = sum(postMode == i) / N\n }\n \n grid = seq(min(x) - 1, max(x) + 1, length = 500)\n dens = rep(NA, length = length(grid))\n \n for(i in 1 : length(grid)) {\n dens[i] = sum(probs * dnorm(grid[i], postMean, P))\n }\n \n hist(x, freq = FALSE)\n lines(grid, dens, col = 'red', lwd = 2)\n \n print(postMode)\n print(kmeans(x, centers = c(-4, 2))$cluster)\n \n for(i in 1 : K) {\n print(CI(muchain[burnin : Nsim, i], ci = 0.95))\n }#END: i loop\n \n}#END: run\n\nrun()\n", "meta": {"hexsha": "d3129d501288e3948adf96be85ea1ff4e6e658fb", "size": 11718, "ext": "r", "lang": "R", "max_stars_repo_path": "posterior.r", "max_stars_repo_name": "fbielejec/dppmixtures", "max_stars_repo_head_hexsha": "5d2da3a577fdf8ae40228fba91927b9ef9ed6e62", "max_stars_repo_licenses": ["Beerware"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2015-01-06T13:03:21.000Z", "max_stars_repo_stars_event_max_datetime": "2015-01-06T13:03:21.000Z", "max_issues_repo_path": "posterior.r", "max_issues_repo_name": "fbielejec/dppmixtures", "max_issues_repo_head_hexsha": "5d2da3a577fdf8ae40228fba91927b9ef9ed6e62", "max_issues_repo_licenses": ["Beerware"], "max_issues_count": 3, "max_issues_repo_issues_event_min_datetime": "2015-04-20T07:47:22.000Z", "max_issues_repo_issues_event_max_datetime": "2015-04-20T07:47:22.000Z", "max_forks_repo_path": "posterior.r", "max_forks_repo_name": "fbielejec/dppmixtures", "max_forks_repo_head_hexsha": "5d2da3a577fdf8ae40228fba91927b9ef9ed6e62", "max_forks_repo_licenses": ["Beerware"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 27.8337292162, "max_line_length": 131, "alphanum_fraction": 0.5456562553, "num_tokens": 4227, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8652240825770432, "lm_q2_score": 0.5851011542032312, "lm_q1q2_score": 0.5062436093602598}} {"text": "model_tempmin <- function (Sdepth_cm = 0.0,\n prof = 0.0,\n tmin = 0.0,\n tminseuil = 0.0,\n tmaxseuil = 0.0){\n #'- Name: TempMin -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: Minimum temperature calculation\n #' * Author: STICS\n #' * Reference: doi:http://dx.doi.org/10.1016/j.agrformet.2014.05.002\n #' * Institution: INRA\n #' * Abstract: recalculation of minimum temperature\n #'- inputs:\n #' * name: Sdepth_cm\n #' ** description : snow depth\n #' ** inputtype : variable\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 500.0\n #' ** unit : cm\n #' ** uri : \n #' * name: prof\n #' ** description : snow cover threshold for snow insulation \n #' ** inputtype : parameter\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 1000\n #' ** unit : cm\n #' ** uri : \n #' * name: tmin\n #' ** description : current minimum air temperature\n #' ** inputtype : variable\n #' ** variablecategory : auxiliary\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 100.0\n #' ** unit : degC\n #' ** uri : \n #' * name: tminseuil\n #' ** description : minimum temperature when snow cover is higher than prof\n #' ** inputtype : parameter\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : 0.0\n #' ** max : 5000.0\n #' ** unit : degC\n #' ** uri : \n #' * name: tmaxseuil\n #' ** description : maximum temperature when snow cover is higher than prof\n #' ** inputtype : parameter\n #' ** parametercategory : constant\n #' ** datatype : DOUBLE\n #' ** default : 0.0\n #' ** min : \n #' ** max : \n #' ** unit : degC\n #' ** uri : \n #'- outputs:\n #' * name: tminrec\n #' ** description : recalculated minimum temperature\n #' ** variablecategory : state\n #' ** datatype : DOUBLE\n #' ** min : 0.0\n #' ** max : 500.0\n #' ** unit : degC\n #' ** uri : \n tminrec <- tmin\n if (Sdepth_cm > prof)\n {\n if (tmin < tminseuil)\n {\n tminrec <- tminseuil\n }\n else\n {\n if (tmin > tmaxseuil)\n {\n tminrec <- tmaxseuil\n }\n }\n }\n else\n {\n if (Sdepth_cm > 0.0)\n {\n tminrec <- tminseuil - ((1 - (Sdepth_cm / prof)) * (abs(tmin) + tminseuil))\n }\n }\n return (list('tminrec' = tminrec))\n}", "meta": {"hexsha": "e1dc455e946f3106ef4dd18ba945ff4e87b84826", "size": 4023, "ext": "r", "lang": "R", "max_stars_repo_path": "src/r/STICS_SNOW/Tempmin.r", "max_stars_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_stars_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/r/STICS_SNOW/Tempmin.r", "max_issues_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_issues_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/r/STICS_SNOW/Tempmin.r", "max_forks_repo_name": "Crop2ML-Catalog/STICS_SNOW", "max_forks_repo_head_hexsha": "26ed2a9a30e068d72d5589b1dc64916c1b07fa09", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 41.90625, "max_line_length": 104, "alphanum_fraction": 0.314193388, "num_tokens": 867, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7490872243177518, "lm_q2_score": 0.6757645879592642, "lm_q1q2_score": 0.5062066194866345}} {"text": "#########################################################################################################\n# Empirical Power simulation Codes of New Combined Threshold Unit Root Test introduced in\n# \n# \"Threshold Unit Root Tests with Smooth Transitions\"\n# \n# written by \n# \n# Dr. Mehmet ÖZCAN\n# mehmetozcan@kmu.edu.tr\n#\n# *NOTES*\n# 1) Caner & Hansen (2001) Threshold Unit Root test and Threshold Effect test\n# programs are taken from Hansen's web page http://www.ssc.wisc.edu/~bhansen/ .\n# However, unit root test program are adjusted to test unit root null hypothesis \n# without any deterministic component.\n#\n# 2) There are two indicator which come from Caner & Hansen (2001)'s codes. Caner &\n# Hansen investigate different type of threshol variable type and different trimming\n# intervals. In this study, lagged difference threshold variable and [0.15, 0.85] \n# trimming interval are used for all simulation experiments and empirical application.\n# \n# thresh Type of threshold variable\n# \t\t 1 for long difference, z(t-1)=y(t-1)-y(t-(m-1))\n# \t\t 2 for lagged difference, z(t-1)=y(t-m)-y(t-m-1)\n# \t\t 3 for lagged level, z(t-1)=y(t-m)\n# trim Trimming Range (normalized threshold, lambda)\n# \t\t 1 for [.15,.85]\n# \t \t 2 for [.10,.90]\n# \t\t 3 for [.05,.95]\n# \n# Other important imputs:\n# rep Monte-Carlo replication \n# p Autoregressive Lag Order \n# mmin Minimal Delay Order for testing\n# mmax Maximal Delay Order for testing\n# m\t\t Delay Order (fixed) for estimation. Set to 0 for m to be estimated by maximizing Wald RT statistics.\n#\n# 3) Results are printed to the screen.\n#\n# 4) All steps of simulation process are explained below. Please check them before run the codes.\n#\n# *Instructions*\n#\n# 1) First, please install required R packcages stated below.\n# 2) You could simulate different parameter settings. Please check \"simulation settings\" part blow.\n# 3) This codes includes critical values for T=100 and T=200.\n# 4) Run all codes\n#\n# \n#########################################################################################################\n# Required Packcages \nlibrary(nloptr)\n#########################################################################################################\n# Required Functions\n#########################################################################################################\n\n#Power DGP\npwDGP <- function(n,m,p1,p2){\n n <- n+2\n et<- as.matrix(rnorm(n))\n et<- et-mean(et)\n\n yt <- matrix(0,n,1)\n for(i in 3:n) {\n if((yt[i-1]-yt[i-2])<0){yt[i]<--m+(1+p1)*yt[i-1]+et[i]}else{yt[i]<-m+(1+p2)*yt[i-1]+et[i]}\n }\n yt<-as.matrix(yt[3:n])\n}\n\n# LNV DGP \ndgpLNV<-function(X, bA, bB, bC, vb){\ntt<-length(vb)\nX <-X[1:tt,]\nstaba<-function(b){(b[1]*X[,1]+b[2]*(1/(1+exp(-(b[3])*((X[,2])-b[4]*tt))))+vb)}\nyAb<-staba(bA)\n\nstabb<-function(b){(b[1]*X[,1]+b[2]*X[,2]+b[3]*(1/(1+exp(-(b[4])*((X[,2])-b[5]*tt))))+vb)}\nyBb<-stabb(bB)\n\nstabc<-function(b){(b[1]*X[,1]+b[2]*X[,2]+b[3]*(1/(1+exp(-(b[5])*((X[,2])-b[6]*tt))))+\nb[4]*(X[,2])*(1/(1+exp(-(b[5])*((X[,2])-b[6]*tt))))+vb)}\nyCb<-stabc(bC)\nlist(yAb=yAb, yBb=yBb, yCb=yCb)\n}\n\n# Starting value estimation for LNV(1998) Models (A, B & C) #\nstartLNV<-function(dta){\ntt<-length(dta)\ntr<-as.matrix(c(1:tt))\ncc<-as.matrix(rep(1,tt))\n\ntauGs<-c()\nfor(kk in 1:tt){ tauGs[kk]<-(kk/tt) }\ngamaGs<-seq(1,10,1)\n\nSSRA<-matrix(0,length(tauGs),length(gamaGs))\nSSRB<-matrix(0,length(tauGs),length(gamaGs))\nSSRC<-matrix(0,length(tauGs),length(gamaGs))\n\nfor(vk in 1:3){\nfor(vi in 1:(length(gamaGs))){\nfor(vj in 1:(length(tauGs))){\nst<-as.matrix((1/(1+exp(-(gamaGs[vi])*((tr)-(tauGs[vj]*tt))))))\nif(vk==1){X<-as.matrix(cbind(cc,st))}\nif(vk==2){X<-as.matrix(cbind(cc,tr,st))}\nif(vk==3){X<-as.matrix(cbind(cc,tr,st,(tr*st)))} \nXX<-(MASS:::ginv(t(X)%*%X, tol=1e-25))\nbetas<-XX%*%(t(X)%*%dta)\nYhat<-t((t(betas)%*%t(X)))\nehat<-(dta-Yhat)\nif(vk==1){SSRA[vj,vi]<-sum((ehat)^2)}\nif(vk==2){SSRB[vj,vi]<-sum((ehat)^2)}\nif(vk==3){SSRC[vj,vi]<-sum((ehat)^2)} \n}\n}\n}\n\nfor(vs in 1:3){\nif(vs==1){posA<-which(SSRA == min(SSRA), arr.ind = TRUE);\nst<-as.matrix((1/(1+exp(-(gamaGs[posA[2]])*((tr)-(tauGs[posA[1]]*tt))))));\nX<-as.matrix(cbind(cc,st))}\nif(vs==2){posB<-which(SSRB == min(SSRB), arr.ind = TRUE);\nst<-as.matrix((1/(1+exp(-(gamaGs[posB[2]])*((tr)-(tauGs[posB[1]]*tt))))));\nX<-as.matrix(cbind(cc,tr,st))}\nif(vs==3){posC<-which(SSRC == min(SSRC), arr.ind = TRUE);\nst<-as.matrix((1/(1+exp(-(gamaGs[posC[2]])*((tr)-(tauGs[posC[1]]*tt))))));\nX<-as.matrix(cbind(cc,tr,st,(tr*st)))}\n\nXX<-(MASS:::ginv(t(X)%*%X, tol=1e-25))\n\nif(vs==1){betasA<-XX%*%(t(X)%*%dta);\nbetasA<-rbind(betasA,gamaGs[posA[2]],tauGs[posA[1]])}\nif(vs==2){betasB<-XX%*%(t(X)%*%dta);\nbetasB<-rbind(betasB,gamaGs[posB[2]],tauGs[posB[1]])}\nif(vs==3){betasC<-XX%*%(t(X)%*%dta);\nbetasC<-rbind(betasC,gamaGs[posC[2]],tauGs[posC[1]])}\n}\nlist(betasA=betasA, betasB=betasB, betasC=betasC)\n}\n\n# LNV(1998) Models (A, B & C) Estimation with Sequential Quadratic Programing\nestLNV<-function(yt, X, b0a, b0b, b0c){\ntt<-length(yt)\nstabA<-function(b){sum((yt-(b[1]*X[,1]+b[2]*(1/(1+exp(-(b[3])*((X[,2])-b[4]*tt))))))^2)}\nstabB<-function(b){sum((yt-(b[1]*X[,1]+b[2]*X[,2]+b[3]*(1/(1+exp(-(b[4])*((X[,2])-b[5]*tt))))))^2)}\nstabC<-function(b){sum((yt-(b[1]*X[,1]+b[2]*X[,2]+b[3]*(1/(1+exp(-(b[5])*((X[,2])-b[6]*tt))))+\nb[4]*(X[,2])*(1/(1+exp(-(b[5])*((X[,2])-b[6]*tt))))))^2)}\n\nlba<-c(-Inf,-Inf,0,0)\nuba<-c(Inf,Inf,Inf,1)\n\nlbb<-c(-Inf,-Inf,-Inf,0,0)\nubb<-c(Inf,Inf,Inf,Inf,1)\n\nlbc<-c(-Inf,-Inf,-Inf,-Inf,0,0)\nubc<-c(Inf,Inf,Inf,Inf,Inf,1)\n\nSA<-nloptr:::slsqp(b0a,fn=stabA,lower=lba,upper=uba, control=list(xtol_rel=1e-25))\nSB<-nloptr:::slsqp(b0b,fn=stabB,lower=lbb,upper=ubb, control=list(xtol_rel=1e-25))\nSC<-nloptr:::slsqp(b0c,fn=stabC,lower=lbc,upper=ubc, control=list(xtol_rel=1e-25))\n\nstaba<-function(b){(b[1]*X[,1]+b[2]*(1/(1+exp(-(b[3])*((X[,2])-b[4]*tt)))))}\nyhatA<-staba(SA$par)\n\nstabb<-function(b){b[1]*X[,1]+b[2]*X[,2]+b[3]*(1/(1+exp(-(b[4])*((X[,2])-b[5]*tt))))}\nyhatB<-stabb(SB$par)\n\nstabc<-function(b){(b[1]*X[,1]+b[2]*X[,2]+b[3]*(1/(1+exp(-(b[5])*((X[,2])-b[6]*tt))))+\nb[4]*(X[,2])*(1/(1+exp(-(b[5])*((X[,2])-b[6]*tt)))))}\nyhatC<-stabc(SC$par)\n\nlist(ehat=as.matrix(cbind((yt-yhatA),(yt-yhatB),(yt-yhatC))), yhat=as.matrix(cbind(yhatA,yhatB,yhatC)),BetaA=as.matrix(SA$par),BetaB=as.matrix(SB$par),BetaC=as.matrix(SC$par))\n}\n\n# ADF Estimator\nlinear<-function(dat,p,cons,trend){\n t <- nrow(dat)\n n <- t - 1 - p\n dy <- as.matrix(dat[2:t]-dat[1:(t-1)])\n y <- as.matrix(dy[(1+p):(t-1)])\n x <- cbind(dat[(1+p):(t-1)], dy[p:(t-2)])\n\n if(p>=2){for (k in 2:p) x <- cbind(x,dy[(p+1-k):(t-1-k)])}\n if(cons==1 & trend==0) x <- cbind(rep(1,n), x)\n if(cons==1 & trend==1) x <- cbind(rep(1,n),seq(1,n,1),x)\n\n kx <- ncol(x)\n xx <- solve(t(x)%*%x, tol=1e-25)\n bols <- xx%*%(t(x)%*%y) \n eols <- y - x%*%bols\n sols <- t(eols)%*%eols\n sigols <- sols/(n-kx)\n seols <- sqrt(diag(xx)%*%sigols)\n rho <- bols[1+cons+trend]\n tadf <- rho/seols[1+cons+trend]\n ar0 <- as.matrix(bols[(cons+trend+2):(cons+trend+1+p)])\n bic<-log((sum(eols^2))/t)+(length(bols)*(log(t)/t))\n aic<-log((sum(eols^2))/t)+(length(bols)*(2/t))\n list(ar0=ar0,eols=eols,aic=aic,bic=bic,tadf=tadf,rho=rho)\n}\n\n# Caner & Hansen (2001) Estimator\ntur_est2 <- function(dat,p,mmin,mmax,m,constant,trend){\n t <- nrow(dat)\n n <- t - 1 - p\n mn <- mmax-mmin+1\n ms <- as.matrix(seq(mmin,mn,1))\n dy <- as.matrix(dat[2:t]-dat[1:(t-1)])\n y <- as.matrix(dy[(1+p):(t-1)])\n if(constant==1) x <- matrix(1,n,1)\n if (trend==1) x <- cbind(x,as.matrix(seq(1,n,1)))\n if(constant==1){x <- cbind(x,dat[(1+p):(t-1)],dy[p:(t-2)])}else{\n x <- cbind(dat[(1+p):(t-1)],dy[p:(t-2)])}\n if(p>1) for (ki in 2:p) x <- cbind(x,dy[(p+1-ki):(t-1-ki)])\n\n if (sum(coef)==0){\n xi_ns <- as.matrix(seq(1,p,1)+1+constant+trend)\n }else{\n idex <- colSums(as.matrix(coef%*%matrix(1,1,p))==(matrix(1,nrow(coef),1)%*%seq(1,p,1)))\n xi_ns <- seq(1,p,1)\n xi_ns <- as.matrix(xi_ns[idex==0]+1+constant+trend)\n }\n\n if (trend==0){\n if(constant==0){\n\t if (sum(coef)==0){\n xi_s <- rbind(1)\n }else{xi_s <- rbind(1,(coef+1))\n }\n }\n }\n if (trend==0){\n if(constant==1){\n\tif (sum(coef)==0){\n xi_s <- rbind(1,2)\n }else{\n xi_s <- rbind(1,2,(coef+2))\n }\n }\n }\n if (trend==1){\n if(constant==1){\n\tif (sum(coef)==0){\n xi_s <- rbind(1,2,3)\n }else{\n xi_s <- rbind(1,2,3,(coef+3))\n }\n }\n }\n\n xs <- as.matrix(x[,xi_s])\n xns <- as.matrix(x[,xi_ns])\n\n if (thresh==1){\n qs <- dat[(1+p):(t-1)]-dat[(1+p-mmin):(t-1-mmin)]\n qs <- as.matrix(qs)\n if(p>1){\n\t for (mi in (mmin+1):mmax){\n qs <- cbind(qs,(dat[(1+p):(t-1)]-dat[(1+p-mi):(t-1-mi)]))\n }\n\t }\n }\n if (thresh==2){\n qs <- dat[(1+p-mmin+1):(t-1-mmin+1)]-dat[(1+p-mmin):(t-1-mmin)]\n qs <- as.matrix(qs)\n if(p>1){\n\t for (mi in (mmin+1):mmax){\n qs <- cbind(qs,(dat[(1+p-mi+1):(t-1-mi+1)]-dat[(1+p-mi):(t-1-mi)]))\n }\n\t }\n }\n if (thresh==3){\n qs <- dat[(1+p-mmin+1):(t-mmin)]\n qs <- as.matrix(qs)\n if(p>1){\n\t for (mi in (mmin+1):mmax){\n qs <- cbind(qs,(dat[(1+p-mi+1):(t-mi)]))\n }\n\t }\n }\n kx <- ncol(x)\n ks <- nrow(coef)+1+constant+trend\n xx <- solve(t(x)%*%x, tol=1e-25)\n bols <- xx%*%(t(x)%*%y) \n eols <- y - x%*%bols\n sols <- t(eols)%*%eols\n wwss <- matrix(0,mn,1)\n smins <- matrix(0,mn,1)\n lams <- matrix(0,mn,1)\n ws <- matrix(0,mn,1)\n r1 <- matrix(0,mn,1)\n r2 <- matrix(0,mn,1)\n t1 <- matrix(0,mn,1)\n t2 <- matrix(0,mn,1)\n rur <- rbind(1,(ks+1))+trend+constant\n indx <- 1:(n)\n pi1 <-.2-.05*trim\n pi2 <-.8+.05*trim\n for (mi in 1:mn){\n q <- qs[,mi]\n qq <- unique(q)\n qq <- as.matrix(sort(qq))\n qq <- as.matrix(qq[(floor(n*pi1)+1):(ceiling(n*pi2)-1)])\n qn <- nrow(qq)\n s <- matrix(0,qn,1)\n wws<- matrix(0,qn,1)\n for (qi in 1:qn){\n d1 <- as.matrix((qlam)\nlist(mhat=mhat,lam=lam,e=e,bs_s=bs_s,ar01=ar01,ar02=ar02,b1=b1,b2=b2,r1=r1,r2=r2,t1=t1,t2=t2,ts=ts,ws=ws,w=w,\nwws=wws, wwss=wwss,leng=leng,vd=vd)\n}\n\n###########################################################\n# Caner & Hansen (2001) TAR Model Estimation Settings #####\nthresh <- 2 \ntrim <- 1 \np <- 1\nm <- 0\nmmin <- 1\nmmax <- p\ncoef <- matrix(c(1:p),p,1)\n###########################################################\n# Simulation Settings: ####################################\nrep <- 10000\nn <- 100\n# Fixed Parameters: #######################################\np2 <- c(-0.1 , -0.3, -0.9)\n# Smooth Transtion Model A Parameters: ####################\nLNVA_1 <- matrix(c(1, 2, 0.5, 0.5), 4,1)\nLNVA_2 <- matrix(c(1, 2, 5 , 0.5), 4,1)\nLNVA_3 <- matrix(c(1, 10, 0.5, 0.5), 4,1)\nLNVA_4 <- matrix(c(1, 10, 5 , 0.5), 4,1)\n# Smooth Transtion Model B Parameters: ####################\nLNVB_1 <- matrix(c(1, 0.5, 2, 0.5 ,0.5), 5,1)\nLNVB_2 <- matrix(c(1, 0.5, 2, 5 ,0.5), 5,1)\nLNVB_3 <- matrix(c(1, 0.5, 10, 0.5 ,0.5), 5,1)\nLNVB_4 <- matrix(c(1, 0.5, 10, 5 ,0.5), 5,1)\n# Smooth Transtion Model C Parameters: ####################\nLNVC_1 <- matrix(c(1, 0.5, 2, 0.5, 0.5, 0.5), 6,1)\nLNVC_2 <- matrix(c(1, 0.5, 2, 0.5, 5 , 0.5), 6,1)\nLNVC_3 <- matrix(c(1, 0.5, 10, 0.5, 0.5, 0.5), 6,1)\nLNVC_4 <- matrix(c(1, 0.5, 10, 0.5, 5 , 0.5), 6,1)\n# Smooth Transiton Model Parameter Cases: #################\nlnvcase <- 4\n# TAR Parameter Cases: ####################################\n#p1 <-c(-0.1, -0.3, -0.9) # Case-1\np1 <-c(0, 0, 0) # Case-2\n#p1 <-c(-0.1, -0.1, -0.1) # Case-3\n###########################################################\n# 5% Criticals T=100 ######################################\n R1A <- 21.15948\n R2A <- 21.27445\n R1B <- 27.69705\n R2B <- 27.77254\n R1C <- 30.93129\n R2C <- 30.95712\n LNVa<- -4.232 # LNV(1998) Critical Values from Table I \n LNVb<- -4.771\n LNVc<- -5.011\n# 5% Criticals T=200 ######################################\n#R1A <- 20.82907\n#R2A <- 21.07073\n#R1B <- 26.67934\n#R2B <- 26.80033\n#R1C <- 29.65022\n#R2C <- 29.69269\n#LNVa<- -4.161 # LNV(1998) Critical Values from Table I \n#LNVb<- -4.629\n#LNVc<- -4.867\n###########################################################\n# Simulation Process\n\ntr<-as.matrix(c(1:n))\ncc<-as.matrix(rep(1,n))\n\nApwr1<-matrix(0,3,1); Apwr2<-matrix(0,3,1); \nBpwr1<-matrix(0,3,1); Bpwr2<-matrix(0,3,1);\nCpwr1<-matrix(0,3,1); Cpwr2<-matrix(0,3,1);\nlnvA <-matrix(0,3,1); lnvB <-matrix(0,3,1);\nlnvC <-matrix(0,3,1); \n\nset.seed(012345)\nptm <- proc.time()\n\nif(lnvcase==1){ bA<-LNVA_1; bB<-LNVB_1; bC<-LNVC_1}\nif(lnvcase==2){ bA<-LNVA_2; bB<-LNVB_2; bC<-LNVC_2}\nif(lnvcase==3){ bA<-LNVA_3; bB<-LNVB_3; bC<-LNVC_3}\nif(lnvcase==4){ bA<-LNVA_4; bB<-LNVB_4; bC<-LNVC_4}\n\nfor(h in 1:length(p2)){\nAszr1<- matrix(0,rep,1); Aszr2<- matrix(0,rep,1);\nBszr1<- matrix(0,rep,1); Bszr2<- matrix(0,rep,1);\nCszr1<- matrix(0,rep,1); Cszr2<- matrix(0,rep,1);\nlnvAs<- matrix(0,rep,1); lnvBs<- matrix(0,rep,1);\nlnvCs<- matrix(0,rep,1); \n\nfor(i in 1:rep){\nX <-as.matrix(cbind(cc,tr))\nvb <-as.matrix(pwDGP(n, 0, p1[h], p2[h]))\ndat <-dgpLNV(X, bA, bB, bC, vb)\n\nSLNVA<- startLNV(dat$yAb)\nSLNVB<- startLNV(dat$yBb)\nSLNVC<- startLNV(dat$yCb)\nb0a <- SLNVA$betasA\nb0b <- SLNVB$betasB\nb0c <- SLNVC$betasC\n\nELNVA<-estLNV(dat$yAb, X, b0a, b0b, b0c)\nELNVB<-estLNV(dat$yBb, X, b0a, b0b, b0c)\nELNVC<-estLNV(dat$yCb, X, b0a, b0b, b0c)\n\nDA<-as.matrix(ELNVA$ehat)\nDB<-as.matrix(ELNVB$ehat)\nDC<-as.matrix(ELNVC$ehat)\n\ntarA <-tur_est2(as.matrix(DA[,1]),p,mmin,mmax,m,0,0)\ntarB <-tur_est2(as.matrix(DB[,2]),p,mmin,mmax,m,0,0)\ntarC <-tur_est2(as.matrix(DC[,3]),p,mmin,mmax,m,0,0)\nlarA <-linear(as.matrix(DA[,1]),1,0,0)\nlarB <-linear(as.matrix(DB[,2]),1,0,0)\nlarC <-linear(as.matrix(DC[,3]),1,0,0) \n\nif(tarA$r1[tarA$mhat].\n#\n#@author: Pablo Carbonell, SYNBIOCHEM\n#@description: Design of experiments routines for combinatorial assembly using R libraries\n#\n # Use always multiples of 2\n\nrepo <- \"https://cloud.r-project.org\"\nif (!require('DoE.base')) {\n if (!require('gmp',quietly=TRUE)) {install.packages('gmp', repos=repo, verbose=FALSE)}\n if (!require('gmp', quietly=TRUE)) {install.packages('numbers', repos=repo, verbose=FALSE)}\n if (!require('crossdes', quietly=TRUE)) {install.packages('crossdes', repos=repo, verbose=FALSE)}\n if (!require('combinat', quietly=TRUE, warn.conflicts = FALSE)) {install.packages('combinat', repos=repo, verbose=FALSE)}\n if (!require('planor', quietly=TRUE)) {install.packages('planor', repos=repo, verbose=FALSE)}\n if (!require('R.utils', quietly=TRUE)) {install.packages('R.utils', repos=repo, verbose=FALSE)}\n if (!require('DoE.base', quietly=TRUE)) {install.packages('DoE.base', repos=repo, verbose=FALSE)}\n}\n\nlatin.augment <- function(m1, m2) {\n ms <- list()\n for (x in seq(1, max(m1))) {\n ms[[x]] <- m2 + (x-1)*max(m2)\n }\n m4 <- NULL\n for (i in seq(1, ncol(m1))) {\n m3 <- NULL\n for (j in seq(1, ncol(m1))) {\n if (is.null(m3)) {\n m3 <- ms[[ m1[i,j] ]]\n } else {\n m3 <- cbind( m3, ms[[ m1[i,j] ]] )\n }\n }\n if (is.null(m4)) {\n m4 <- m3\n } else {\n m4 <- rbind(m4, m3)\n }\n \n }\n return(m4)\n}\n\nlatin <- function(ntotal) {\n if (!require('gmp', quietly=TRUE)) {\n require('numbers', quietly=TRUE)\n useGMP <- FALSE\n } else {\n useGMP <- TRUE\n }\n require('crossdes', quietly=TRUE)\n # Find all prime factors\n if (useGMP) {\n\tif (ntotal > 1) {\n \tm <- factorize(ntotal)\n \tm <- as.integer(m)\n\t} else {\n\t\tm <- c(1)\n\t}\n } else {\n m <- primeFactors(ntotal)\n }\n sq <- list()\n # Join any 2 * 2 as 4 (orthogonal latin squares only for n >= 3)\n fours <- c()\n while ( (length(m) >=2) && (m[2] == 2) ) {\n if (length(m) > 2) {\n m <- m[3:length(m) ]\n fours <- c(fours, 4)\n } else {\n m <- c(4)\n }\n }\n m <- c(fours, m)\n # Compute orthogonal squares of each factor\n for (n in m) {\n if (n > 2) {\n x <- des.MOLS(n,n)\n } else {\n if (n ==2 ) {\n x <- matrix(c(1,2,2,1), nrow=2)\n } else {\n x <- matrix(c(1))\n }\n }\n sq[[n]] <- x\n \n }\n # Augment the matrix with all prime factors\n m1 <- sq[[ m[1] ]]\n if (length(m) == 1) {\n return( m1 )\n } else {\n m2 <- sq[[ m[2] ]]\n mx <- latin.augment(m1, m2)\n if (length(m) > 2) {\n for (i in seq(3, length(m))) {\n m3 <- sq[[ m[i] ]]\n mx <- latin.augment(mx, m3)\n }\n }\n }\n \n return(mx)\n}\n\n\n\nlatin.old <- function(n, nrand = 20) {\n library('crossdes')\n if (n > 2) {\n x <- des.MOLS(n,n)[1:n, 1:n]\n } else {\n if (n ==2 ) {\n x <- matrix(c(1,2,2,1), nrow=2)\n } else {\n x <- matrix(c(1))\n }\n }\n \n# x <- matrix(1:n, n, n)\n# x <- t(x)\n# for (i in 2:n) {\n# x[i, ] <- x[i, c(i:n, 1:(i-1))]\n# }\n if (n > 2) {\n if (nrand > 0) {\n for (i in 1:nrand) {\n x <- x[sample(n), ]\n x <- x[,sample(n)]\n }\n }\n x <- x[, order(x[1,])]\n }\n return(x)\n}\n\n# Generate full permutation or latin square\npermut <- function(n, ptype='latin', randomize=TRUE) {\n require('combinat', quietly=TRUE, warn.conflicts = FALSE)\n if (ptype == 'latin') {\n if (randomize) {\n return(latin(n))\n } else {\n return(latin(n))\n }\n } else {\n return(permn(n))\n }\n}\n\n# Regular factorial design with implicit S model\ndoe1 <- function(factors, nlevels, timeout=5) {\n print(factors)\n require('planor', quietly=TRUE)\n require('R.utils', quietly=TRUE)\n require('combinat', quietly=TRUE)\n construct <- planor.factors(\n factors=factors, \n nlevels=nlevels\n )\n const.ffd <- list()\n # We check if we have enough resolution for the model \n HAVERESOLUTION <- TRUE\n # Start with the simplest model: ~S\n res <- 3\n const.total <- prod(construct@fact.info$nlev)\n n <- 1\n const.ffd[[n]] <- NULL\n # Change to verify if we have enough resolution\n units <- 0\n while (units < const.total) {\n # Start with full combinations\n k <- 1\n des <- 0\n units <- const.total\n while (!is.null(des)) {\n des <- NULL\n try(\n evalWithTimeout(\n des <- regular.design(factors=construct, resolution=res, nunits=const.total/k),\n timeout=timeout, onTimeout='silent'\n )\n )\n # Simplify S4 structure\n if (!is.null(des)) {\n const.ffd[[n]] <- list(design=slot(des, 'design'), resolution=res)\n units <- dim(const.ffd[[n]]$design)[1]\n }\n k <- k*2\n }\n # Increase to the next model\n n <- n+1\n res <- res+2\n }\n return(const.ffd)\n}\n\n# Orthogonal arrays\ndoe2 <- function(factors, nlevels, timeout=5, seed=888) {\n require('DoE.base', quietly=TRUE)\n const.oa <- list()\n des <- oa.design(factor.names=factors, nlevels=nlevels, seed=seed)\n const.oa[[1]] <- list(design=des, resolution=NULL)\n return(const.oa)\n}\n", "meta": {"hexsha": "986f005c060795e36a56ee2a024b67e64b13979b", "size": 5863, "ext": "r", "lang": "R", "max_stars_repo_path": "mydeo.r", "max_stars_repo_name": "pablocarb/sbc-doe", "max_stars_repo_head_hexsha": "3467a42765bae03aedfd24924e4c18213753d27a", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "mydeo.r", "max_issues_repo_name": "pablocarb/sbc-doe", "max_issues_repo_head_hexsha": "3467a42765bae03aedfd24924e4c18213753d27a", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "mydeo.r", "max_forks_repo_name": "pablocarb/sbc-doe", "max_forks_repo_head_hexsha": "3467a42765bae03aedfd24924e4c18213753d27a", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 27.6556603774, "max_line_length": 125, "alphanum_fraction": 0.4857581443, "num_tokens": 1698, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7606506635289835, "lm_q2_score": 0.6654105454764747, "lm_q1q2_score": 0.5061449729358634}} {"text": "##https://github.com/jrfaulkner/spmrf\n\n#' Extract latent process parameters and associated Bayesian credible intervals\n#'\n#' Extract theta (latent process) parameters from an object with posterior draws and associated Bayesian credible intervals.\n#' @param mfit An object containing posterior draws from a \\code{spmrf} model fit or \\code{stan} model fit. The object can be of class \\code{stanfit}, \\code{array}, \\code{matrix}, or \\code{data.frame}.\n#' @param obstype Character string with the name of the probability distribution of the observations. This controls the back-transformation of the process parameters. Possible values for \\code{obstype} are 'normal', 'poisson', or 'binomial'.\n#' @param alpha Controls level for 100*(1-\\code{alpha})\\% Bayesian credible intervals. Values must be 0 < \\code{alpha} < 1.\n#' @return Returns a list with the posterior median and posterior (1-\\code{alpha}) quantiles of the theta parameter vector.\n#' @seealso \\code{\\link[rstan]{stan}}, \\code{\\link[rstan]{as.array.stanfit}}, \\code{\\link[rstan]{as.matrix.stanfit}}, \\code{\\link[rstan]{as.data.frame.stanfit}}, \\code{\\link{spmrf}}\n#' @export\n\nextract_theta <- function(mfit, obstype=\"normal\", alpha=0.05){\n\n if (missing(mfit)) stop(\"Must specify object with posterior draws from a spmrf or stan model fit.\")\n if ( !(class(mfit)[1] %in% c(\"array\", \"matrix\", \"data.frame\", \"stanfit\") ) ) stop(\"Object must be of class 'stanfit', 'array', 'matrix', or 'data.frame'. Object must be or be generated from a stan model fit object.\")\n if ( !(obstype %in% c(\"normal\", \"poisson\", \"binomial\") ) ) stop(\"Argument 'obstype' must be 'normal', 'poisson', or 'binomial'.\")\n if (!(0 < alpha & alpha < 1)) stop(\"Must specify 'alpha' between 0 and 1.\")\n\n if (class(mfit)[1]==\"stanfit\") {\n \t if (!requireNamespace(\"rstan\", quietly = TRUE)) {\n \t\t\tstop(\"Package 'rstan' needed for this function to work. Please install it.\", call. = FALSE)\n \t }\n \ttmp.th <- rstan::extract(mfit, \"theta\")[[1]]\n }\n if (class(mfit)[1]==\"array\") {\n nca <- dim(mfit)[2]\n\t tmp.th1 <- mfit[ , 1, ]\n\t if (nca > 1){\n\t for (jj in 2:nca){\n\t tmp.th1 <- rbind(tmp.th1, mfit[ ,jj,])\n\t }\n\t }\n\t ath <- grep(x=dimnames(tmp.th1)[[2]], pattern=\"theta\")\n\t zth <- grep(x=dimnames(tmp.th1)[[2]], pattern=\"ztheta\")\n\t thind <- setdiff(ath, zth)\n\t tmp.th <- tmp.th1[ , thind]\n }\n if (class(mfit)[1]==\"matrix\"){\n \tath <- grep(x=colnames(mfit), pattern=\"theta\")\n \tzth <- grep(x=colnames(mfit), pattern=\"ztheta\")\n \tthind <- setdiff(ath, zth)\n \ttmp.th <- mfit[ , thind]\n }\n if (class(mfit)[1]==\"data.frame\") {\n \tath <- grep(x=names(mfit), pattern=\"theta\")\n \tzth <- grep(x=names(mfit), pattern=\"ztheta\")\n \tthind <- setdiff(ath, zth)\n \ttmp.th <- as.matrix(mfit[ , thind])\n }\n plow <- alpha/2\n phigh <- 1 - alpha/2\n\n if (obstype==\"normal\"){\n tmp.md <- apply(tmp.th, 2, median)\n tmp.l <- apply(tmp.th, 2, quantile, probs=plow)\n tmp.u <- apply(tmp.th, 2, quantile, probs=phigh)\n }\n if (obstype==\"poisson\"){\n tmp.md <- exp(apply(tmp.th, 2, median))\n tmp.l <- exp(apply(tmp.th, 2, quantile, probs=plow))\n tmp.u <- exp(apply(tmp.th, 2, quantile, probs=phigh))\n }\n if (obstype==\"binomial\"){\n tmp.md <- plogis(apply(tmp.th, 2, median))\n tmp.l <- plogis(apply(tmp.th, 2, quantile, probs=plow))\n tmp.u <- plogis(apply(tmp.th, 2, quantile, probs=phigh))\n }\n out <- list(postmed=tmp.md, bci.lower=tmp.l, bci.upper=tmp.u)\n out\n\n}\n\n", "meta": {"hexsha": "766cd5496cf8e926289daf7ee0aa856e29cd01c7", "size": 3427, "ext": "r", "lang": "R", "max_stars_repo_path": "src/R/extract_theta.r", "max_stars_repo_name": "xiaobw95/Time_Series_Decomposition_Based_on_Gaussian_Markov_Random_Fields", "max_stars_repo_head_hexsha": "569a663e3fe0c19c31baaa698e269debf8f33bf3", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/R/extract_theta.r", "max_issues_repo_name": "xiaobw95/Time_Series_Decomposition_Based_on_Gaussian_Markov_Random_Fields", "max_issues_repo_head_hexsha": "569a663e3fe0c19c31baaa698e269debf8f33bf3", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/R/extract_theta.r", "max_forks_repo_name": "xiaobw95/Time_Series_Decomposition_Based_on_Gaussian_Markov_Random_Fields", "max_forks_repo_head_hexsha": "569a663e3fe0c19c31baaa698e269debf8f33bf3", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 46.3108108108, "max_line_length": 243, "alphanum_fraction": 0.6454625036, "num_tokens": 1076, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7879312056025699, "lm_q2_score": 0.640635868562172, "lm_q1q2_score": 0.5047769922684417}} {"text": "#!/usr/bin/Rscript\n\n# Bhishan Poudel\n# Jan 16, 2016\n\n# set up a device driver to plot\n#postscript(file='~/Copy/Programming/R/rprograms/plotting/legends/legend3.eps')\n\npar(mar = c(4, 4, 2, 0.1))\nplot(rnorm(100), rnorm(100),\n xlab = expression(hat(mu)[0]), ylab = expression(alpha^beta),\n main = expression(paste(\"Plot of \", alpha^beta, \" versus \", hat(mu)[0])))\n\n# turn off device driver\n#dev.off()", "meta": {"hexsha": "4828b454bc9501544e7fa959706703ddea148e75", "size": 405, "ext": "r", "lang": "R", "max_stars_repo_path": "AstroSeminar2019Spring/ModernStatistics_R/chap0/plotting/legends/legend3.r", "max_stars_repo_name": "bhishanpdl/AstroSeminar_OU", "max_stars_repo_head_hexsha": "3181fb74b3ac67a86c37683ddb0d48355a084495", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "AstroSeminar2019Spring/ModernStatistics_R/chap0/plotting/legends/legend3.r", "max_issues_repo_name": "bhishanpdl/AstroSeminar_OU", "max_issues_repo_head_hexsha": "3181fb74b3ac67a86c37683ddb0d48355a084495", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "AstroSeminar2019Spring/ModernStatistics_R/chap0/plotting/legends/legend3.r", "max_forks_repo_name": "bhishanpdl/AstroSeminar_OU", "max_forks_repo_head_hexsha": "3181fb74b3ac67a86c37683ddb0d48355a084495", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 27.0, "max_line_length": 79, "alphanum_fraction": 0.6641975309, "num_tokens": 137, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES\n\n", "lm_q1_score": 0.6791786991753929, "lm_q2_score": 0.7431680143008301, "lm_q1q2_score": 0.5047438852215975}} {"text": "rawData <- read.csv(\"data/train.csv\")\ntrain <- rawData[1:(nrow(rawData) * 0.7), ]\ncrossValid <- rawData[(nrow(train) + 1):(nrow(rawData)), ]\ntest <- read.csv(\"data/test.csv\")\n\ncleaningData <- function(original) {\n if (\"Survived\" %in% colnames(original)) {\n #usefulCol <- c(\"Survived\", \"Pclass\", \"Sex\", \"Age\", \"SibSp\",\"Parch\",\"Fare\",\"Embarked\")\n usefulCol <- c(\"Survived\", \"Pclass\", \"Sex\", \"Age\")\n } else {\n #usefulCol <- c(\"Pclass\", \"Sex\", \"Age\", \"SibSp\",\"Parch\",\"Fare\",\"Embarked\")\n usefulCol <- c(\"Pclass\", \"Sex\", \"Age\")\n }\n\n data <- data.matrix(fixNa(original)[, usefulCol])\n cbind(matrix(1, nrow(data), 1), data)\n #fixNa(original)[, usefulCol]\n}\n\nfixNa <- function(data) {\n temp <- data\n temp$Age[is.na(temp$Age)] <- mean(temp$Age, na.rm=T)\n temp$Fare[is.na(temp$Fare)] <- mean(temp$Fare, na.rm=T)\n temp$Pclass[is.na(temp$Pclass)] <- mean(temp$Pclass, na.rm=T)\n temp$SibSp[is.na(temp$SibSp)] <- mean(temp$SibSp, na.rm=T)\n temp$SibSp[is.na(temp$Parch)] <- mean(temp$Parch, na.rm=T)\n temp\n}\n\nmySigmoid <- function(z) {\n 1.0 / (1.0 + exp(-z));\n}\n\ncostFunc <- function(theta, X, y) {\n sum(cbind(y, 1 - y) * cbind(log(sigmoid(X %*% theta)), log(1 - sigmoid(X %*% theta)))) / -nrow(X)\n}\n\ngradient <- function(theta, X, y) {\n (t(X) %*% (sigmoid(X %*% theta) - y)) / nrow(X);\n}\n\nlearning <- function(train, alpha, iterNum) {\n cleanTrain <- cleaningData(train)\n X <- cleanTrain[, -2]\n y <- cleanTrain[, 2]\n theta <- matrix(0, ncol(X), 1)\n for (i in 1:iterNum) {\n theta <- theta - alpha * gradient(theta, X, y)\n }\n theta\n}\n\nbuiltInLearning <- function(train) {\n glm(Survived~., family = binomial, data=train)\n}\n\ncrossValidate <- function(validateSet, theta) {\n cleanedData <- cleaningData(validateSet)\n X <- cleanedData[, -2]\n y <- cleanedData[, 2]\n m <- nrow(X)\n p <- rep(c(0), m)\n correct <- 0\n t_neg <- 0\n t_pos <- 0\n f_neg <- 0\n f_pos <- 0\n for (i in 1:m) {\n rate <- sigmoid(X[i,] %*% theta)\n \n if (rate >= 0.5 && y[i] == 1) {\n t_pos <- t_pos + 1\n } else if (rate < 0.5 && y[i] == 1) {\n t_neg <- t_neg + 1\n } else if (rate >= 0.5 && y[i] == 0) {\n f_neg <- f_neg + 1\n } else if (rate < 0.5 && y[i] == 0) {\n f_pos <- f_pos + 1\n } \n \n if ((rate >= 0.5 && y[i] == 1) || (rate < 0.5 && y[i] == 0)) {\n correct <- correct + 1\n }\n\n }\n \n precision <- (t_pos / (t_pos + f_pos))\n recall <- (t_pos / (t_pos + f_neg))\n f1 <- (precision * recall) / (precision + recall)\n #acc <- (t_pos + t_neg) / (t_pos + t_neg + f_pos + f_neg)\n acc <- correct / m\n \n list(precision = precision, recall = recall, f1 = f1, acc = acc)\n}\n\nmyPredict <- function(test, theta) {\n X <- cleaningData(test)\n m <- nrow(X)\n p <- rep(c(0), m)\n\n for (i in 1:m) {\n rate <- sigmoid(X[i,] %*% theta)\n\n if (rate >= 0.5) {\n p[i] <- 1\n } else {\n p[i] <- 0\n }\n\n }\n\n data.frame(PassengerId=test[,\"PassengerId\"], Survived=p)\n}", "meta": {"hexsha": "3832ce72becfcab6282f99a0ca76492a7289483b", "size": 3132, "ext": "r", "lang": "R", "max_stars_repo_path": "titanic/scripts.r", "max_stars_repo_name": "tenggyut/r_work", "max_stars_repo_head_hexsha": "b963109fa4bea67f5d808e60b18cfd732761604e", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "titanic/scripts.r", "max_issues_repo_name": "tenggyut/r_work", "max_issues_repo_head_hexsha": "b963109fa4bea67f5d808e60b18cfd732761604e", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "titanic/scripts.r", "max_forks_repo_name": "tenggyut/r_work", "max_forks_repo_head_hexsha": "b963109fa4bea67f5d808e60b18cfd732761604e", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 27.7168141593, "max_line_length": 101, "alphanum_fraction": 0.5102171137, "num_tokens": 1073, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8354835371034369, "lm_q2_score": 0.6039318337259583, "lm_q1q2_score": 0.5045751046107283}} {"text": "# 4. faza: Napredna analiza podatkov\n\n\n# CLUSTERING ###################################################################\n\n#podatki za zdravstvo po letu 2011 vedno slabši zato je najbolj optimalno leto za analizo 2011\n\n\n# tabela:\n\nX <- zivljenje %>% filter(leto == \"2011\") %>% filter(!drzava %in% c(\"Bulgaria\",\"Luxembourg\"))\nX$drzava[X$drzava == \"Czech Republic\"] <- \"Czech Rep.\"\nX.norm <- X %>% dplyr::select(pricakovana.starost, kg.na.osebo, odstotek.BDP.ki.gre.v.zdravstvo, tveganje.revscine)%>% \n scale()\nrownames(X.norm) <- X$drzava\nX.norm <- X.norm[1:29, -1]\n\n\n\n## HIERARHIČNO\n\n#funkcije s predavanj:\n\n\nhc.kolena = function(dendrogram, od = 1, do = NULL, eps = 0.5) {\n # število primerov in nastavitev parametra do\n n = length(dendrogram$height) + 1\n if (is.null(do)) {\n do = n - 1\n }\n # k.visina je tabela s štirimi stolpci\n # (1) k, število skupin\n # (2) višina združevanja\n # (3) sprememba višine pri združevanju\n # (4) koleno: ali je točka koleno?\n k.visina = tibble(\n k = as.ordered(od:do),\n visina = dendrogram$height[do:od]\n ) %>%\n # sprememba višine\n mutate(\n dvisina = visina - lag(visina)\n ) %>%\n # ali se je intenziteta spremembe dovolj spremenila?\n mutate(\n koleno = lead(dvisina) - dvisina > eps\n )\n k.visina\n}\n\n# iz tabele k.visina vrne seznam vrednosti k,\n# pri katerih opazujemo koleno\nhc.kolena.k = function(k.visina) {\n k.visina %>%\n filter(koleno) %>%\n dplyr::select(k) %>%\n unlist() %>%\n as.character() %>%\n as.integer()\n}\n\n# narišemo diagram višin združevanja\ndiagram.kolena = function(k.visina) {\n k.visina %>% ggplot() +\n geom_point(\n mapping = aes(x = k, y = visina),\n color = \"red\"\n )+\n geom_line(\n mapping = aes(x = as.integer(k), y = visina),\n color = \"red\"\n )+\n geom_point(\n data = k.visina %>% filter(koleno),\n mapping = aes(x = k, y = visina),\n color = \"blue\", size = 2\n )+\n ggtitle(paste(\"Kolena:\", paste(hc.kolena.k(k.visina), collapse = \", \"))) +\n xlab(\"število skupin (k)\") +\n ylab(\"razdalja pri združevanju skupin\") +\n theme_classic()\n}\n\ndiagram.skupine = function(podatki, oznake, skupine, k) {\n podatki = podatki %>%\n bind_cols(skupine) %>%\n rename(skupina = ...2)\n \n d = podatki %>%\n ggplot(\n mapping = aes(\n x = x, y = y, color = skupina\n )\n ) +\n geom_point() +\n geom_label(label = oznake, size = 2) +\n scale_color_hue() +\n theme_classic()\n \n for (i in 1:k) {\n d = d + geom_encircle(\n data = podatki %>%\n filter(skupina == i)\n )\n }\n d\n}\n\n\nlibrary(rgeos)\nlibrary(ggalt)\n\n\ndendrogram <- dist(X.norm) %>% hclust(method = \"ward.D\")\n\nr = hc.kolena(dendrogram)\ndiagram.kolena(r)\n# -> 4 skupine\n\n\n#zakomentirano da ne izpisuje v rmd\n#plot(dendrogram, hang=-1, cex=0.4, main = \"drzava\", labels = X$drzava)\n#rect.hclust(dendrogram,k=4,border=\"red\")\n\nskupine = dendrogram %>% cutree(k = 4) %>% as.ordered()\nskupine\n\ndiagram.skupine(X.norm, colnames(X.norm), skupine, 2 )\n\n#drzave <- attributes(X.norm)$dimnames[[1]] # dobimo imena, ki jih spremenimo potem v stolpce (to ni df)\n#drzave.x.y = as_tibble(X.norm %>% cmdscale(k = 2)) #%>% bind_cols(drzave) %>%\n# rename(drzava = ...3, x = V1, y = V2)\n\n\n# Narišemo države tako, da jih pobarvamo glede na pripadnost skupini\n# diagram.skupine(drzave.x.y, drzave.x.y$drzava, skupine, 4)\n\n\ndata(\"World\")\nevropa <- World %>% filter (continent == 'Europe')\nskupina <- data.frame(name = X$drzava, skupine=factor(skupine))\nzem <- merge(x = evropa, y = skupina)\ngrupiranje.map <- tm_shape(zem) + tm_polygons(\"skupine\", palette = \"Pastel1\", popup.vars = c(\"Skupina: \" = \"skupine\"))+tmap_mode(\"view\")\n\ngrupiranje.map\n\n\n# NAPOVED ######################################################################\n\n# za napoved pricakovane zivljenjske dobe v Sloveniji za leto 2020 \n# uvozila nove podatke, z daljšim časovnim obdobjem\n\npricakovana.slo <- read_csv(\"podatki/HFA_43_EN.csv\",\n skip = 25,\n na = \"-\",\n locale = locale(encoding = \"Windows-1250\"))%>% \n filter(COUNTRY == \"SVN\")%>%\n dplyr::select(YEAR, VALUE )\n\nlibrary(ranger)\n\nLag <- function(x, n){c(rep(NA, n), x)[1:length(x)]}\n\nnaredi.df <- function(x){\n data.frame(pricakovana = x,\n pricakovana1 = Lag(x, 1),\n pricakovana2 = Lag(x, 2),\n pricakovana3 = Lag(x, 3),\n pricakovana4 = Lag(x, 4))\n \n}\n\ndf <- naredi.df(pricakovana.slo$VALUE)\nmodel = ranger(formula = pricakovana ~ ., data = df %>% drop_na())\n\nn = nrow(df)\n\ndf2 <- naredi.df(c(pricakovana.slo$VALUE, NA))\nnapoved <- predict(model, data = df2[n+1,])$predictions\ndf2[n+1,1] = napoved\n\npricakovanje <- as_tibble(data.frame(\n c(1985:2020),\n df2$pricakovana))\n\n\nnapoved.graf <- ggplot(pricakovanje) + geom_line(mapping = aes(x = c.1985.2020., y = df2.pricakovana), color = \"red\")+\n geom_line(mapping = aes(x = c.1985.2020., y = df2.pricakovana, color = c.1985.2020. <= 2019), show.legend = FALSE)+\n labs(\n x = \"Leto\",\n y = \"Pričakovana življenjska doba\",\n title = \"Gibanje pričakovane življenjske dobe v Sloveniji in napoved za leto 2020\"\n )\n\n\n\n\n#### \n#samo da ugotovim korelacijo:\nggpairs(data.frame(X.norm))\n", "meta": {"hexsha": "d7ce484fe136c6ec09203f616978cee15f13e64d", "size": 5227, "ext": "r", "lang": "R", "max_stars_repo_path": "analiza/analiza.r", "max_stars_repo_name": "nikapavlic/APPR-2021-22", "max_stars_repo_head_hexsha": "38dfc42b1871282fcde385d38b079ae1def1f82e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "analiza/analiza.r", "max_issues_repo_name": "nikapavlic/APPR-2021-22", "max_issues_repo_head_hexsha": "38dfc42b1871282fcde385d38b079ae1def1f82e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2022-01-20T18:56:29.000Z", "max_issues_repo_issues_event_max_datetime": "2022-02-04T12:22:38.000Z", "max_forks_repo_path": "analiza/analiza.r", "max_forks_repo_name": "nikapavlic/APPR-2021-22", "max_forks_repo_head_hexsha": "38dfc42b1871282fcde385d38b079ae1def1f82e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 26.0049751244, "max_line_length": 136, "alphanum_fraction": 0.5982399082, "num_tokens": 1875, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7248702880639791, "lm_q2_score": 0.6959583313396339, "lm_q1q2_score": 0.5044795161186867}} {"text": "#\r\n# R-Code zu NOT-Statistik 2. Auflage\r\n#\r\n\r\n# Stand 30.01.2021\r\n\r\n## Kapitel 3 ##\r\n#\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 51: Daten in R einlesen\r\n# Vorbereitung: Daten-Datei \"PLZ_0.CSV\" lokal speichern\r\nPLZ.0 = read.csv2(file.choose(), stringsAsFactors = TRUE)\r\n# erstellt Daten-Container \"PLZ.0\"\r\n# read.csv2() liest eine csv-Datei mit Komma als Dezimaltrenner in \"PLZ.0\" ein\r\n# file.choose() öffnet den Explorer zur Dateiauswahl, ausgewählt wird \"PLZ_0.CSV\"\r\n# stringsAsFactors = TRUE importiert Buchstaben/Zeichenketten als Faktoren\r\n\r\nPLZ.0$Beginn=as.Date(PLZ.0$Beginn, \"%d.%m.%Y\")\r\nPLZ.0$Ende=as.Date(PLZ.0$Ende, \"%d.%m.%Y\")\r\n# PLZ.0$[Spaltenname] wählt Spalte aus\r\n# Datumsformatierung für Spalten \"Beginn\" und \"Ende\"\r\n\r\nstr(PLZ.0) # zeigt eine Übersicht der PLZ.0-Daten und Formate an\r\n\r\nattach(PLZ.0) # PLZ.0-Merkmale so direkt über Spaltennamen auswählbar\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 51: Balkendiagramm\r\n# plot() zeichnet Balkendiagramm für attributive Merkmale\r\nplot(Lieferant)\r\nplot(Kunde)\r\nplot(Methode)\r\nplot(Review)\r\nplot(Komplexität)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 51: Einzelwertdiagramme\r\n# stripchart() zeichnet Einzelwertdiagramm\r\nstripchart(PL, vertical=TRUE, method=\"stack\", pch=20, main=\"PL\")\r\n# vertical=TRUE: Punkte werden vertikal eingezeichnet\r\n# \"stack\": gleiche Werte stapeln\r\n# pch: point character, 20 ist ein ausgefüllter Kreis\r\n# main: setzt Titel über die Grafik\r\nstripchart(PMa, vertical=TRUE, method=\"stack\", pch=20, main=\"PMa\")\r\nstripchart(Abteilungen, vertical=TRUE, method=\"stack\", pch=20, main=\"Abteilungen\")\r\nstripchart(Laufzeit, vertical=TRUE, method=\"stack\", pch=20, main=\"Laufzeit\")\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 52: Zeitreihendiagramme für 1 Merkmal\r\n# plot() zeichnet Zeitreihendiagramm für variable Merkmale\r\nplot(PL, type=\"b\")\t# type=\"b\": both, d. h. Linien und Punkte zeichnen\r\nplot(PMa, type=\"b\")\r\nplot(Abteilungen, type=\"b\")\r\nplot(Laufzeit, type=\"b\")\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 52: Zeitreihendiagramm gemeinsam für Beginn und Ende\r\nplot(Beginn, type=\"b\", pch=20, xlab=\"Index\")\r\nlines(Ende, type=\"b\", col=\"blue\", pch=20)\r\n# Hinzufügen der Linien und Punkte für 'Ende'\r\n# col: color, Farbe für Linie und Punkte\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 56: Zeilennummern für Einträge finden\r\n# which() liefert Zeilennummern mit vorgegebenen Kriterien\r\nwhich(Lieferant == \"Ja\")\r\nwhich(Kunde == \"nein \")\r\nwhich(Review == \"eventuell\" | Review == \"vielleicht\")\r\n# | ist das Zeichen für ODER\r\nwhich(PL > 50)\r\nwhich(Laufzeit < 0)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 57: Einlesen der überarbeiteten Daten\r\nsetwd(\"~/Arbeit/Buch/SPM/R\")\r\n\r\ndetach(PLZ.0)\r\n# Die Merkmalsnamen im R-Speicher müssen eindeutig sein. Da die Spaltennamen in \"PLZ_0.CSV\" und \"PLZ_1.CSV\"\r\n# identisch sind, wird zuerst die alte Daten-Datei aus dem direkten R-Speicher entfernt. Sie sind mit der\r\n# Angabe \"PLZ.0$[Spaltenname]\" weiter verfügbar.\r\n\r\n# Vorbereitung: Daten-Datei \"PLZ_1.CSV\" lokal speichern\r\nPLZ.1 = read.csv2(file.choose(), stringsAsFactors = TRUE)\r\n# erstellt Daten-Container \"PLZ.1\"\r\n# eingelesen in \"PLZ.1\" wird jetzt die Datei \"PLZ_1.CSV\"\r\n\r\n# Formatierung des Datums\r\nPLZ.1$Beginn=as.Date(PLZ.1$Beginn, \"%d.%m.%Y\")\r\nPLZ.1$Ende=as.Date(PLZ.1$Ende, \"%d.%m.%Y\")\r\n\r\nstr(PLZ.1) # Übersicht Einträge in \"PLZ.1\"\r\nattach(PLZ.1) # Daten-Container \"PLZ.1\" im direkten R-Speicher verfügbar machen\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 57: Erstellen der Grafiken mit PLZ.1\r\n# Balkendiagramme\r\nplot(Lieferant)\r\nplot(Kunde)\r\nplot(Review)\r\n\r\n# Einzelwertdiagramme\r\nstripchart(PL, vertical=TRUE, method=\"stack\", pch=20, main=\"PL\")\r\nstripchart(Laufzeit, vertical=TRUE, method=\"stack\", pch=20, main=\"Laufzeit\")\r\n\r\n# Zeitreihendiagramme\r\nplot(PL, type=\"b\")\r\nplot(Laufzeit, type=\"b\")\r\n\r\n\r\n## Kapitel 4 ##\r\n#\r\n# Grafiken\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 67: Boxplots\r\n# Daten PLZ_1.CSV sind eingelesen und mit attach(PLZ.1) verfügbar gemacht\r\n\r\nboxplot(Laufzeit ~ Lieferant, pch=8, col=\"lightgrey\", lty=1)\r\n# erstellt Boxplot für Laufzeit nach Lieferant\r\n# pch=8: Extremwerte werden als Stern angezeigt\r\n# col=\"lightgrey\": Kästen werden hellgrau ausgefüllt\r\n# lty=1: Linien werden durchgezogen gezeichnet\r\nboxplot(Laufzeit ~ Kunde, pch=8, col=\"lightgrey\", lty=1)\r\nboxplot(Laufzeit ~ Methode, pch=8, col=\"lightgrey\", lty=1)\r\nboxplot(Laufzeit ~ Review, pch=8, col=\"lightgrey\", lty=1)\r\nboxplot(Laufzeit ~ Komplexität, pch=8, col=\"lightgrey\", lty=1)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 69: Haupteffekte-Diagramm\r\n# Definieren einer Haupteffekte-Diagramm-Funktion: HEdiagramm\r\nHEdiagramm = function(x,y){\r\n mw = aggregate(y ~ x, data=PLZ.1, FUN=mean)\r\n # erstellt Variable mw\r\n # speichert die Mittelwerte (mw) von y je Stufe von x aus den Daten PLZ.1 über die Funktion (FUN) mean darin\r\n plot(mw, lty=\"blank\", xlab=deparse(substitute(x)), ylab=deparse(substitute(y)))\r\n # zeichnet ein leeres Diagramm mit Achsen und Achsen-Beschriftung\r\n lines(mw) # fügt Verbindungslinie der Mittelwerte hinzu\r\n points(mw, pch=19) # zeichnet Punkte für die Mittelwerte ein\r\n}\r\nHEdiagramm(Lieferant,Laufzeit)\r\nHEdiagramm(Kunde,Laufzeit)\r\nHEdiagramm(Methode,Laufzeit)\r\nHEdiagramm(Review,Laufzeit)\r\nHEdiagramm(Komplexität,Laufzeit)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 72: Streudiagramm-Matrix (im Buch: Wechselwirkungs-Diagramm)\r\nrequire(lattice)\r\n# für die Streudiagramm-Matrix muss das Package \"lattice\" installiert sein\r\n# Pakete in R installieren (Internetverbindung notwendig!):\r\n# Pakete > Installiere Pakete > OK > Paketnamen in der Liste finden > OK\r\n\r\nPLZ.1.variabel = subset(PLZ.1, select=c(PL,PMa,Abteilungen,Laufzeit))\r\n# subset() erstellt Teilmenge aus PLZ.1, die alle vier variablen Merkmale PL, PMa, Abteilungen und Laufzeit enthält\r\nsplom(~PLZ.1.variabel)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 76: Wechselwirkungs-Diagramm\r\ninteraction.plot(Lieferant,Kunde,Laufzeit, type=\"b\", col=1:nlevels(Kunde))\r\n# interaction.plot ist Funktion für Wechselwirkungs-Diagramme\r\n# hier: Lieferant auf x-Achse, Kunde für unterschiedliche Linien\r\n# type=\"b\": Linien und Punkte (\"both\") zeichnen\r\n# col=1:nlevels(Kunde): Farben (\"colors\") der Linien und Punkte\r\ninteraction.plot(Lieferant,Methode,Laufzeit, type=\"b\", col=1:nlevels(Methode))\r\ninteraction.plot(Lieferant,Review,Laufzeit, type=\"b\", col=1:nlevels(Review))\r\ninteraction.plot(Lieferant,Komplexität,Laufzeit, type=\"b\", col=1:nlevels(Komplexität))\r\ninteraction.plot(Kunde,Methode,Laufzeit, type=\"b\", col=1:nlevels(Methode))\r\ninteraction.plot(Kunde,Review,Laufzeit, type=\"b\", col=1:nlevels(Review))\r\ninteraction.plot(Kunde,Komplexität,Laufzeit, type=\"b\", col=1:nlevels(Komplexität))\r\ninteraction.plot(Methode,Review,Laufzeit, type=\"b\", col=1:nlevels(Review))\r\ninteraction.plot(Methode,Komplexität,Laufzeit, type=\"b\", col=1:nlevels(Komplexität))\r\ninteraction.plot(Review,Komplexität,Laufzeit, type=\"b\", col=1:nlevels(Komplexität))\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 81: 3D-Streudiagramme\r\ncloud(Laufzeit ~ PL + PMa, scales = list(arrows=FALSE))\r\n# erstellt ein 3D-Streudiagramm (Punktewolke / cloud)\r\n# für Laufzeit (Z-Achse) mit PL (Y-Achse) und PMa (X-Achse)\r\n# scales = list(arrows=FALSE): Achsenbeschriftung (Skala) mit Zahlen statt Pfeilen\r\ncloud(Laufzeit ~ PL + Abteilungen, scales = list(arrows=FALSE))\r\ncloud(Laufzeit ~ PL + PMa, scales = list(arrows=FALSE))\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 84: Matrixplot mit attributiven Merkmalen\r\nPLZ.1.variabel = subset(PLZ.1, select=c(PL,PMa,Abteilungen,Laufzeit))\r\n# subset() erstellt Teilmenge aus PLZ.1, die alle variablen Merkmale enthält\r\n\r\nsymbole = trellis.par.get(\"superpose.symbol\")\r\n# Eigenschaften von Gruppierungs-Symbolen in \"symbole\" gespeichert\r\n\r\nsplom(~PLZ.1.variabel, # Streudiagramm-Matrix für alle variablen Merkmale\r\n\tgroups=Komplexität, # unterschieden nach Komplexitäts-Stufen\r\n\tkey=list(columns=3,\r\n\t\tpoints=list(pch=symbole$pch[1:3], col=symbole$col[1:3]),\r\n\t\ttext=list(levels(Komplexität)))\r\n\t# key definiert die Legende (Liste mit Vorgaben):\r\n\t# columns=3: Anzahl Legenden-Spalten über der Grafik 3\r\n\t# points=... : Vorgabe der Symbole und Farben\r\n\t# text=... : Für jedes Legenden-Symbol wird die Bezeichnung/Level von Komplexität eingefügt\r\n\t)\r\n\r\n\r\n\r\n## Kapitel 5 ##\r\n#\r\n# Erstes statistisches Prozess-Modell\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 98: Erstes SPM\r\n# Daten PLZ_1.CSV sind eingelesen und mit attach(PLZ.1) verfügbar gemacht\r\n\r\nrequire(car)\r\n# für die hier verwendeten Kontraste und Varianzanalyse (ANOVA) wird das Paket car benötigt\r\n\r\nlm.1 = lm(Laufzeit ~ PL + PMa + Abteilungen + Lieferant + Kunde + Methode + Review + Komplexität\r\n + PL:PMa + PL:Abteilungen + PL:Lieferant + PL:Kunde + PL:Methode + PL:Review + PL:Komplexität\r\n + PMa:Abteilungen + PMa:Lieferant + PMa:Kunde + PMa:Methode + PMa:Review + PMa:Komplexität\r\n + Abteilungen:Lieferant + Abteilungen:Kunde + Abteilungen:Methode + Abteilungen:Review + Abteilungen:Komplexität\r\n + Lieferant:Kunde + Lieferant:Methode + Lieferant:Review + Lieferant:Komplexität\r\n + Kunde:Methode + Kunde:Review + Kunde:Komplexität\r\n + Methode:Review + Methode:Komplexität\r\n + Review:Komplexität\r\n + I(PL^2) + I(PMa^2),\r\n contrasts=list(Lieferant=contr.Sum,Kunde=contr.Sum,Methode=contr.Sum,Review=contr.Sum,Komplexität=contr.Sum),\r\n data=PLZ.1)\r\n# in Container lm.1 sind die Ergebnisse des ersten SPMs gespeichert\r\n# lm: linear model für Zielgröße Laufzeit (vor der Tilde ~)\r\n# PL:PMa (z. B.) Wechselwirkung, durch \":\" angegeben\r\n# I(PL^2) quadratischer Effekt von PL\r\n# contrasts: Art der Effekt-Berechnung von attributiven Merkmalen\r\n\r\nAnova(lm.1, type=3)\r\n# gibt eine Varianzanalyse-Tabelle mit dem Berechnungstyp 3 aus\r\n\r\n\r\n\r\n## Kapitel 6 ##\r\n#\r\n# SPM vereinfachen\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 103: Auswahl Effekt-Kodierung\r\ngetOption(\"contrasts\")\r\n# zeigt die Einstellungen für die Kontraste an:\r\n# erster Eintrag für Faktoren ohne Rangfolge (Voreinstellung: contr.treatment)\r\n# zweiter Eintrag für Faktoren mit Rangfolge (Voreinstellung: contr.poly)\r\n\r\noptions(contrasts=c(\"contr.Sum\",\"contr.poly\"))\r\n# setzt die Kontraste global für alle Modelle\r\n# contr.Sum: Effekt-Kodierung für attributive Merkmale ohne Rangfolge der Stufen\r\n# contr.poly: polynomiale Kodierung für attributive Merkmale mit Rangfolge der Stufen\r\ngetOption(\"contrasts\")\r\n# geänderte Kontrast-Einstellung: \"contr.Sum\" und \"contr.poly\"\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 120: Unwichtige Einflüsse ausschließen (1/2)\r\n# Daten PLZ_1.CSV sind eingelesen\r\n# Das SPM lm.1 ist berechnet\r\n\r\nlm.1.step = step(lm.1, direction=\"both\")\r\n# step: schrittweise Auswahl wichtiger Einflüsse über AIC\r\n# direction=\"both\": Ausschluss und Hinzunahme in der Auswahl-Strategie\r\n\r\nrequire(car) # Paket car laden\r\n\r\nAnova(lm.1.step, type=3)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 121: Unwichtige Einflüsse ausschließen (2/2)\r\n# Daten PLZ_1.CSV sind eingelesen\r\n\r\nlm.2 = lm(Laufzeit ~ PL + PMa + Abteilungen + Lieferant + Kunde + Methode + Review + Komplexität\r\n + I(PMa^2)\r\n + PL:PMa\r\n + Abteilungen:Methode\r\n + Lieferant:Kunde\r\n + Kunde:Komplexität\r\n + Methode:Review, \r\n data = PLZ.1)\r\n\r\nAnova(lm.2, type=3)\r\n\r\n\r\n\r\n## Kapitel 7 ##\r\n#\r\n# Modell-Qualität prüfen\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 133: Modell-Qualitäts-Kennzahlen R², R²(prog) und S\r\n# lm.2 wurde berechnet\r\n\r\nSST = sum((Laufzeit-mean(Laufzeit, na.rm=TRUE))^2)\r\n# berechnet SST und speichert Wert in \"SST\"\r\nSSE = sum(resid(lm.2)^2)\r\n# berechnet SSE und speichert Wert in \"SSE\"\r\nR2 = 1-SSE/SST # berechnet R² und speichert Wert in \"R2\"\r\nR2 # zeigt Wert von \"R2\" an, hier: 0.8744451 = 87,44%\r\n\r\nPRESS = sum((resid(lm.2)/(1-hatvalues(lm.2)))^2)\r\n# berechnet PRESS und speichert Wert in \"PRESS\"\r\nPRESS # zeigt Wert von \"PRESS\" an\r\nR2.prog = 1-PRESS/SST # berechnet R²_prog und speichert Wert in \"R2.prog\"\r\nR2.prog # zeigt Wert von \"R2.prog\" an, hier: 0.8366248 = 83,66%\r\n\r\nS.spm = sqrt(SSE/df.residual(lm.2))\r\n# berechnet Modell-Standardabweichung S und speichert Wert in \"S.spm\"\r\n# df.residual ermittelt die Freiheitsgrade der Residuen, hier: n-k=115\r\nS.spm # zeigt Wert von \"S.spm\" an, hier: 8.533396 = 8,53 Tage\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 141: Residuendiagramme: Wahrscheinlichkeitsnetze\r\n# lm.2 wurde berechnet\r\n\r\nqqnorm(residuals(lm.2), xlab=\"Normalverteilungs-Quantile\", ylab=\"Residuen\",\r\n\tmain=\"Wahrscheinlichkeitsnetz\", pch=20)\r\nqqline(residuals(lm.2), col=\"blue\")\r\n# qqnorm zeichnet Wahrscheinlichkeitsnetz für Normalverteilung\r\n# residuals liefert Residuen\r\n# xlab, ylab, main: Beschriftung x-Achse, y-Achse und Titel\r\n# qqline zeichnet Ideallinie (hier: blau)\r\n\r\nqqnorm(rstandard(lm.2), xlab=\"Normalverteilungs-Quantile\", ylab=\"standardisierte Residuen\",\r\n\tmain=\"W.-Netz\", pch=20)\r\nqqline(rstandard(lm.2), col=\"blue\")\r\n# rstandard liefert standardisierte Residuen\r\n\r\nqqnorm(rstudent(lm.2), xlab=\"Normalverteilungs-Quantile\", ylab=\"studentisierte Residuen\",\r\n\tmain=\"W.-Netz\", pch=20)\r\nqqline(rstudent(lm.2), col=\"blue\")\r\n# rstudent liefert studentisierte Residuen\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 142: Residuendiagramme: Anpassungen vs. Residuen und Beobachtungsnr. vs. Residuen\r\n# lm.2 wurde berechnet\r\n\r\nplot(fitted(lm.2), residuals(lm.2), xlab=\"Angepasste Werte für Laufzeit\", ylab=\"Residuen\", main=\"Angepasste Werte vs. Residuen\", pch=20)\r\nabline(h=0, col=\"blue\")\r\n# fitted liefert die angepassten Werte für die Laufzeit\r\n# abline zeichnet eine Referenzlinie ein, h: horizontal\r\n\r\nplot(residuals(lm.2), type=\"b\", xlab=\"Beobachtungsnummer\", ylab=\"Residuen\", main=\"Beobachtungsnummer vs. Residuen\", pch=20)\r\nabline(h=0, col=\"blue\")\r\n# type=\"b\" zeichnet Punkte und Linien (\"both\")\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 143: Residuendiagramme: Auffällige Residuen finden\r\n# lm.2 wurde berechnet\r\n\r\nPLZ.1.SPM = cbind(PLZ.1,Fit=fitted(lm.2),Resi=residuals(lm.2))\r\n# cbind: column-bind, hängt Spalten aneinander\r\n# erzeugt Container PLZ.1.SPM mit den Daten PLZ.1 sowie den\r\n# Anpassungen (Fit) und Residuen (Resi)\r\n\r\nPLZ.1.SPM[which(residuals(lm.2) < -20),]\r\n# zeigt die Zeile(n) aus PLZ.1.SPM an, für die die Residuen kleiner -20 sind\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 158: Modell-Qualitäts-Kennzahlen: Test auf Normalverteilung für Residuen\r\n# lm.2 wurde berechnet\r\n\r\nrequire(nortest)\r\n# für den Anderson-Darling-Test wird das Paket \"nortest\" benötigt\r\nad.test(residuals(lm.2))\r\n# berechnet den Anderson-Darling-Test auf Normalverteilung (ad.test) und gibt die Werte aus\r\n\r\n\r\n\r\n## Kapitel 8 ##\r\n#\r\n# Modell-Interpretation und Prognose\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 166: Effektediagramme\r\n# lm.2 wurde berechnet\r\n\r\nrequire(effects) # für die Effektdiagramme muss das Paket \"effects\" installiert sein\r\n\r\n# Haupteffekte-Diagramme mit großem E bei Effect\r\nplot(Effect(\"PL\", lm.2, ylim=c(91,156)))\r\n\t# ylim gibt Wertebereich der y-Achse an\r\nplot(Effect(\"PMa\", lm.2, ylim=c(91,156)))\r\nplot(Effect(\"Abteilungen\", lm.2, ylim=c(91,156)))\r\nplot(Effect(\"Lieferant\", lm.2, ylim=c(91,156)))\r\nplot(Effect(\"Kunde\", lm.2, ylim=c(91,156)))\r\nplot(Effect(\"Methode\", lm.2, ylim=c(91,156)))\r\nplot(Effect(\"Review\", lm., ylim=c(91,156),2))\r\nplot(Effect(\"Komplexität\", lm.2, ylim=c(91,156)))\r\n\r\n\r\n# Wechselwirkungsdiagramme mit kleinem e bei effect\r\nplot(effect(\"PL:PMa\", lm.2, ylim=c(72,168)))\r\nplot(effect(\"Abteilungen:Methode\", lm.2, ylim=c(72,168)))\r\nplot(effect(\"Lieferant:Kunde\", lm.2, ylim=c(72,168)))\r\nplot(effect(\"Kunde:Komplexität\", lm.2, ylim=c(72,168)))\r\nplot(effect(\"Methode:Review\", lm.2, ylim=c(72,168)))\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 168: Quantile und Auswahl von Projekten\r\nPLZ.1[PL >= quantile(PL, .75) \r\n & PMa >= quantile(PMa, .75) \r\n & Laufzeit >= quantile(Laufzeit, .75),]\r\n# [... ,] gibt Auswahlkriterium an\r\n# quantile(x, .75) berechnet 75%-Quantil von x\r\n# Auswahl PL >= Q_75(PL)\r\n# und (&) PMa >= Q_75(PMa)\r\n# und (&) Laufzeit >= Q_75(Laufzeit)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 174: Kontur- und Wirkungsflächendiagramme Laufzeit\r\n# lm.2 wurde berechnet\r\n\r\nrequire(lattice) # Pakte lattice laden\r\n\r\n# Gitter-Stützpunkte berechnen\r\nPL.p = seq(min(PL), max(PL), length.out=100)\r\nPMa.p = seq(min(PMa), max(PMa), length.out=100)\r\n# seq() erstellt eine Sequenz von Werten\r\nAbteil.p = c(1,3,6)\r\npunkte = list(PL=PL.p, PMa=PMa.p, Abteilungen=Abteil.p)\r\n# list() erstellt eine Liste\r\ngitter = expand.grid(punkte)\r\n# erstellt das Gitter für die Stützpunkte\r\n\r\ngitter.niedrig = cbind(gitter,\r\n\tdata.frame(Lieferant=\"nein\", Kunde=\"nein\",\r\n\tMethode=\"SiSi\", Review=\"nein\",\r\n\tKomplexität=\"niedrig\"))\r\n# data.frame() erstellt eine Datenstruktur\r\ngitter.niedrig[, \"Prognose\"] = c(predict(lm.2, gitter.niedrig))\r\n# mit [,...] wird an den vorhandenen data.frame eine Spalte angefügt\r\n# predict(modell,daten) berechnet Prognosen für \"daten\", hier für gitter.niedrig\r\n\r\ngitter.hoch = cbind(gitter,\r\n\tdata.frame(Lieferant=\"ja\", Kunde=\"ja\",\r\n\tMethode=\"PDCA\", Review=\"ja\",\r\n\tKomplexität=\"hoch\"))\r\ngitter.hoch[, \"Prognose\"] = c(predict(lm.2, gitter.hoch))\r\n\r\n# Konturdiagramme\r\ncontourplot(Prognose ~ PL*PMa | Abteilungen,\r\n\tdata=gitter.niedrig, main=\"Kontur niedrig\",\r\n\tregion=TRUE,cuts=10^3, contour=FALSE, \r\n\tcol.regions=rainbow(2000))\r\n# contourplot() erstellt das Konturdiagramm\r\n# Prognose ~ PL*PMa | Abteilungen: Höhenlinien aus \"Prognose\", \r\n# x- und y-Achse PMa und PL,\r\n# unterschiedliche Grafiken je Wert in \"Abteilungen\"\r\n\r\ncontourplot(Prognose ~ PL*PMa | Abteilungen,\r\n\tdata=gitter.hoch, main=\"Kontur hoch\",\r\n\tregion=TRUE,cuts=10^3, contour=FALSE,\r\n\tcol.regions=rainbow(2000))\r\n\r\n# Wirkungsflächendiagramme\r\nwireframe(Prognose ~ PL*PMa | Abteilungen,\r\n\tdata=gitter.niedrig, main=\"Wirkungsfläche niedrig\",\r\n\tzlab=\"\", colorkey=FALSE, shade=TRUE, drape=TRUE,\r\n\tscales=list(arrows=FALSE), light.source = c(0,2,2))\r\n# wireframe() erstellt das Wirkungsflächendiagramm\r\n# Achsen und Grafiken wie in countourplot (s. o.)\r\n\r\nwireframe(Prognose ~ PL*PMa | Abteilungen,\r\n\tdata=gitter.hoch, main=\"Wirkungsfläche hoch\", \r\n\tzlab=\"\", colorkey=FALSE, shade=TRUE, drape=TRUE,\r\n\tscales=list(arrows=FALSE), light.source = c(0,2,2))\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 187: Koeffizienten des vereinfachten SPMs ausgeben\r\n# lm.2 wurde berechnet\r\n\r\nround(lm.2$coef,2) # zeigt Koeffizienten von lm.2 an\r\n\r\nstr(PLZ.1) # zeigt Beschreibung der Daten inkl. Stufen von attributiven Einflussgrößen an\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 195 Prognosen für Laufzeit\r\n# lm.2 wurde berechnet\r\n\r\n# Einstellwerte vorgeben\r\nDaten.neu = data.frame(Lieferant=\"ja\", Kunde=\"nein\",\r\n\tMethode=\"PDCA\", Review=\"ja\",\r\n\tKomplexität=c(\"hoch\",\"mittel\",\"niedrig\"), \r\n\tPL=10, PMa=5, Abteilungen=2)\r\n# data.frame() erstellt eine Datenstruktur\r\n# Spalten werden aufgefüllt, so dass alle Spalten dieselbe Länge haben\r\n\r\n# Prognose berechnen\r\nPrognose=predict(lm.2, Daten.neu)\r\n\r\ncbind(Daten.neu,Prognose)\r\n# cbind() verbindet Spaltenweise (column-bind)\r\n# Ergebnis: Anzeige der Einstellwerte und Prognosen\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 198: Vertrauens- und Prognosebereiche für Laufzeit\r\n# lm.2 wurde berechnet\r\n\r\n# Einstellwerte vorgeben\r\nDaten.neu = data.frame(Lieferant=\"ja\", Kunde=\"nein\",\r\n\tMethode=\"PDCA\", Review=\"ja\",\r\n\tKomplexität=c(\"hoch\",\"mittel\",\"niedrig\"), \r\n\tPL=10, PMa=5, Abteilungen=2)\r\n\r\n# Prognose berechnen\r\nPrognose=round(predict(lm.2, Daten.neu),2)\r\n# round() rundet Werte, hier auf 2 Nachkommastellen\r\n\r\n# Obere Grenze 98% Vertrauensbereich berechnen\r\nOGW.KI=round(predict(lm.2, Daten.neu, interval=\"confidence\", level=0.98)[,3],2)\r\n# [,3] wählt dritte Spalte aus, die die OGW angibt\r\n\r\n# Obere Grenze 98% Prognosebereiche berechnen\r\nOGW.PI=round(predict(lm.2, Daten.neu, interval=\"prediction\", level=0.98)[,3],2)\r\n\r\ncbind(Prognose,OGW.KI,OGW.PI)\r\n# cbind() verbindet Spaltenweise (column-bind)\r\n# Ergebnis: Anzeige der Prognosen und obere Streubereichsgrenzen\r\n\r\n\r\n\r\n## Kapitel 9 ##\r\n#\r\n# Prozess-Simulation\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 206: Anzahl Werte für Lieferant\r\n# Daten PLZ_1.CSV sind eingelesen und mit attach(PLZ_1.CSV) verfügbar gemacht\r\ntable(Lieferant)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 207: Monte Carlo-Simulation\r\n# lm.2 wurde berechnet\r\n\r\nSimu=10^4 # Anzahl Simulationen in \"Simu\" speichern\r\n\r\nsimu.attr=function(Merkmal,Simu) {\r\n\tsample(rep(levels(Merkmal),rmultinom(1,Simu,prob=table(Merkmal))),Simu)\r\n\t}\r\n# Funktion \"simu.attr\" für die Einstellwerte der attributiven Merkmale:\r\n# 2 Werte: Merkmal und Anzahl (Simu)\r\n# sample() zieht eine Stichprobe aus den Werten\r\n# rep(a,b) wiederholt den Eintrag \"a\" b Mal\r\n# levels(Merkmals) Stufen des Merkmals\r\n# rmultinom: Zufallszahlen der Multinomialverteilung (2 oder mehr Gruppen im Merkmal)\r\n# rmultinom(a,b,c) erzeugt a Mal Zufallswerte, je Durchlauf b Werte mit den Wahrscheinlichkeiten c je Stufe des Merkmals\r\n# prog=table(Merkmal) gibt die Anzahl Werte je Stufe wieder, die in rmultinom automatisch in Anteile umgewandelt werden\r\n\r\nsim.Lief=simu.attr(Lieferant,Simu)\r\n# speichert Zufallswerte für Lieferant in sim.Lief\r\nsim.Kund=simu.attr(Kunde,Simu)\r\nsim.Meth=simu.attr(Methode,Simu)\r\nsim.Rev=simu.attr(Review,Simu)\r\nsim.Kompl=simu.attr(Komplexität,Simu)\r\n\r\nsim.PL=abs(rnorm(Simu, mean(PL), sd(PL)))\r\n# speichert Zufallswerte für PL in sim.PL\r\n# abs() berechnet den Absolutwert\r\nsim.PMa=ceiling(abs(rnorm(Simu, mean(PMa), sd(PMa))))\r\n# ceiling berechnet die nächstgrößere ganze Zahl\r\nsim.Abt=ceiling(abs(rnorm(Simu, mean(Abteilungen),sd(Abteilungen))))\r\n\r\n# Zusammenstellung der Daten für die Monte Carlo-Simulation\r\nsim.daten = data.frame(Lieferant=sim.Lief,Kunde=sim.Kund,Methode=sim.Meth,Review=sim.Rev,\r\n\tKomplexität=sim.Kompl,PL=sim.PL,PMa=sim.PMa,Abteilungen=sim.Abt)\r\n\r\n# Berechnung der simulierten Laufzeit\r\nLaufzeit.sim=predict(lm.2,sim.daten)\r\nMC.daten=cbind(sim.daten,Laufzeit.sim)\r\n\r\n# Boxplot simulierte Laufzeit nach Review, Methode und Komplexität\r\nboxplot(Laufzeit.sim ~ Review+Methode+Komplexität, pch=8, col=\"lightgrey\", lty=1, \r\n axes=FALSE, xlab=\"\", ylab=\"\", data=MC.daten)\r\nmtext(side=1, line=3.5, at=c(2.5,6.5,10.5), levels(Komplexität))\r\nmtext(side=1, line=3.5, at=-2.0, adj=c(0), \"Komplexität\")\r\nmtext(side=1, line=2, at=seq(1.5,11.5,by=2), levels(Methode))\r\nmtext(side=1, line=2, at=-2.0, adj=c(0), \"Methode\")\r\nmtext(side=1, line=0.5, at=1:12, levels(Review))\r\nmtext(side=1, line=0.5, at=-2.0, adj=c(0), \"Review\")\r\naxis(1, at=1:12, labels=NA)\r\naxis(2, las=2)\r\nbox()\r\n\r\n\r\n# Boxplot simulierte Laufzeit nach Lieferant, Kunde und Komplexität\r\nboxplot(Laufzeit.sim ~ Lieferant+Kunde+Komplexität, pch=8, col=\"lightgrey\", lty=1, \r\n axes=FALSE, xlab=\"\", ylab=\"\", data=MC.daten)\r\nmtext(side=1, line=3.5, at=c(2.5,6.5,10.5), levels(Komplexität))\r\nmtext(side=1, line=3.5, at=-2.0, adj=c(0), \"Komplexität\")\r\nmtext(side=1, line=2, at=seq(1.5,11.5,by=2), levels(Kunde))\r\nmtext(side=1, line=2, at=-2.0, adj=c(0), \"Kunde\")\r\nmtext(side=1, line=0.5, at=1:12, levels(Lieferant))\r\nmtext(side=1, line=0.5, at=-2.0, adj=c(0), \"Lieferant\")\r\naxis(1, at=1:12, labels=NA)\r\naxis(2, las=2)\r\nbox()\r\n\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 208: 95%-Prognosebereich\r\n# Laufzeit.sim wurde berechnet\r\n\r\n# Berechnung und Ausgabe der Quantilwerte\r\nPI = quantile(Laufzeit.sim, prob=c(0.025,0.975))\r\nPI\r\n\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 217: 95%-Streubereich für die Anpassungsgüte R²\r\n#lm.2 wurde berechnet\r\n\r\nrequire(car) # Paket car laden\r\n\r\nBTD.Laufzeit=Boot(lm.2, \r\n\tf = function(mod){summary.lm(mod)$r.squared},\r\n\tR=10^4)\r\n# Boot(daten,funktion,anzahl,...) ist Bootstrapping-Funktion\r\n# f gibt die Funktion für die Kennzahl R² an\r\n# hier: R² aus der Zusammenfassung des Modells \r\n# R: Anzahl Simulationsdurchläufe\r\n\r\nconfint(BTD.Laufzeit, level=.95, type=\"perc\")\r\n# berechnet den Streubereich von BTD.Laufzeit\r\n# level Abdeckbereich\r\n# perc: Perzentil- bzw. Quantil-Methode\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 218: Boxplot\r\n# Simulierte Anpassungsgüten BTD.Laufzeit sind berechnet\r\nboxplot(BTD.Laufzeit$t, pch=8, col=\"lightgrey\", lty=1)\r\n# BTD.Laufzeit$t: simulierte BTD.Laufzeit-Anpassungsgüten\r\nabline(h=c(0.8327,0.9262,0.8744), col=\"red\")\r\n# zeichnet drei horizontale Linien ein\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 218: Zeitreihendiagramm\r\n# Simulierte Anpassungsgüten BTD.Laufzeit sind berechnet\r\nplot(BTD.Laufzeit$t)\r\n# BTD.Laufzeit$t: simulierte BTD.Laufzeit-Anpassungsgüten\r\nabline(h=c(0.8327,0.9262,0.8744), col=\"red\")\r\n# zeichnet drei horizontale Linien ein\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 225: 95%-Streubereich für die Anpassungsgüte R²\r\n# lm.2 wurde berechnet\r\n\r\nrequire(car) # Paket car laden\r\n\r\nBTR.Laufzeit=Boot(lm.2, \r\n\tf = function(mod){summary.lm(mod)$r.squared},\r\n\tR=10^4,\r\n\tmethod=\"residual\")\r\n# Boot(daten,funktion,anzahl,...) ist Bootstrapping-Funktion\r\n# f gibt die Funktion für die Kennzahl R² an\r\n# hier: R² aus der Zusammenfassung des Modells \r\n# R: Anzahl Simulationsdurchläufe\r\n# method wählt die Bootstrap-Methode, \"residual\": Residuen wie in (31)\r\n\r\nconfint(BTR.Laufzeit, level=.95, type=\"perc\")\r\n# berechnet den Streubereich von BTR.Laufzeit\r\n# level Abdeckbereich\r\n# perc: Perzentil- bzw. Quantil-Methode\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 225: Boxplot\r\n# Simulierte Anpassungsgüten BTR.Laufzeit sind berechnet\r\nboxplot(BTR.Laufzeit$t, pch=8, col=\"lightgrey\", lty=1)\r\n# BTR.Laufzeit$t: simulierte BTR.Laufzeit-Anpassungsgüten\r\nabline(h=c(0.8430,0.9083,0.8744), col=\"red\")\r\n# zeichnet drei horizontale Linien ein\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 226: Zeitreihendiagramm\r\n# Simulierte Anpassungsgüten BTR.Laufzeit sind berechnet\r\nplot(BTR.Laufzeit$t)\r\n# BTR.Laufzeit$t: simulierte BTR.Laufzeit-Anpassungsgüten\r\nabline(h=c(0.8430,0.9083,0.8744), col=\"red\")\r\n# zeichnet drei horizontale Linien ein\r\n\r\n\r\n\r\n## Kapitel 10 ##\r\n#\r\n# Optimierung und Nachweis\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 236: Zielgrößenoptimierung\r\n# lm.2 wurde berechnet\r\n\r\n# Zielgrößenoptimierung\r\n# Wunschfunktion d*_min als Funktion \"d.min2\" definieren\r\nd.min2=function(y,OG,Ziel,w) {\r\n ifelse(y>OG, 0, ((OG-y)/(OG-Ziel))^w*100 ) }\r\n# ifelse(Bedingung, A, B): prüft Bedingung für jedes Element\r\n# und gibt A zurück wenn wahr und B wenn falsch\r\n\r\n# Rasterwerte erstellen\r\nd.PL=seq(min(PL),max(PL),length=100)\r\n# seq() erstellt eine Sequenz von Werten\r\n# min() liefert das Minimum\r\n# max() liefert das Maximum\r\n# length gibt an aus wie vielen Werte die Sequenz besteht\r\nd.PMa=seq(min(PMa),max(PMa))\r\nd.Abteilungen=seq(min(Abteilungen),max(Abteilungen))\r\nd.Lieferant=levels(Lieferant)\r\n# levels() gibt die Kategorien eines attributiven Merkmals aus\r\nd.Kunde=levels(Kunde)\r\nd.Methode=levels(Methode)\r\nd.Review=levels(Review)\r\nd.Komplexität=levels(Komplexität)\r\n\r\n# Rasterwerte in einer Liste speichern\r\nRaster.Liste = list(PL=d.PL, PMa=d.PMa,\r\n\tAbteilungen=d.Abteilungen, Lieferant=d.Lieferant,\r\n\tKunde=d.Kunde, Methode=d.Methode,\r\n\tReview=d.Review, Komplexität=d.Komplexität)\r\n\r\n# alle Kombinationen erzeugen\r\nRaster = expand.grid(Raster.Liste)\r\n\r\n# Prognosewerte berechnen und in Spalte \"Prognose\" anhängen\r\nRaster[, \"Prognose\"] = c(predict(lm.2, Raster))\r\n# Werte von d*_min berechnen und in Spalte \"dmin2\" anhängen\r\nRaster[, \"dmin2\"] = d.min2(Raster$Prognose, OG=50, Ziel=25, w=1)\r\n\r\n# Auswahl kurzer Projekte\r\nkurze.Projekte = subset(Raster, Raster$dmin2 > 0)\r\n# subset() erstellt Auswahl nach Kriterium \"Raster-Werte d*_min > 0\"\r\ndim(Raster)\r\ndim(kurze.Projekte)\r\n# dim() liefert die Dimensionen, d. h. Anzahl Zeilen (1. Wert) und Spalten (2. Wert)\r\nsummary(kurze.Projekte)\r\n# summary() liefert Kennzahlen\r\nRaster[Raster$dmin2==max(Raster$dmin2),]\r\n# zeigt die Einstellungen an mit denen dmin2 maximal ist\r\n\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 253: Finden der Kombinationen mit obere Prognosegrenze > 200 Tage\r\n# lm.2 wurde berechnet\r\n\r\n# Worst-Case-Wunschfunktion d*_max als Funktion \"d.max\" definieren\r\nd.max=function(y,UG,Ziel,w) {\r\n ifelse(yZiel, 100, ((y-UG)/(Ziel-UG))^w*100 ))\r\n }\r\n# ifelse(Bedingung, A, B): prüft Bedingung für jedes Element\r\n# und gibt A zurück wenn wahr und B wenn falsch\r\n\r\n# Rasterwerte erstellen\r\nd.PL=seq(min(PL),max(PL),length=100)\r\nd.PMa=seq(min(PMa),max(PMa))\r\nd.Abteilungen=seq(min(Abteilungen),max(Abteilungen))\r\nd.Lieferant=levels(Lieferant)\r\nd.Kunde=levels(Kunde)\r\nd.Methode=levels(Methode)\r\nd.Review=levels(Review)\r\nd.Komplexität=levels(Komplexität)\r\n\r\n# Rasterwerte in einer Liste speichern\r\nRaster.Liste = list(PL=d.PL, PMa=d.PMa,\r\n\tAbteilungen=d.Abteilungen, Lieferant=d.Lieferant,\r\n\tKunde=d.Kunde, Methode=d.Methode,\r\n\tReview=d.Review, Komplexität=d.Komplexität)\r\n\r\n# alle Kombinationen erzeugen\r\nRaster = expand.grid(Raster.Liste)\r\n\r\n# Obere 99%-Prognosegrenze \"OPG\" berechnen\r\nOPG = as.data.frame(predict(lm.2, Raster,\r\n\tinterval=\"prediction\", level=0.98))$upr\r\n# predict(., interval=\"'prediction\"') berechnet zweiseitige Prognosegrenzen\r\n# level=0.98 Anteil innerhalb des zweiseitig begrenzten Prognoseintervalls an\r\n# bei 98% innerhalb liegen 2% außerhalb, mit 1% > OPG\r\n# $upr wählt obere Prognosegrenze (upper prediction limit) für OPG aus\r\n\r\n# und in Spalte \"OPG\" anhängen\r\nRaster[, \"Prognosegrenze\"] = OPG\r\n\r\n# Projekte mit OPG>200 in \"lange.projekte\" speichern\r\nlange.projekte=subset(Raster, Raster$Prognosegrenze > 200)\r\nsummary(lange.projekte)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 254 Einschränkung der Einstellkombinationen auf obere Prognosegrenze < 200 Tage\r\n# Der Datensatz \"lange.projekte\" wurde erstellt\r\n\r\n# Matrixplot für die variablen Merkmale PL, PMa und Abteilungen\r\nrequire(lattice) # Paket lattice laden\r\n\r\nlange.projekte.variabel = subset(lange.projekte, select=c(PL,PMa, Abteilungen))\r\n# Auswahl der variablen Merkmale aus \"lange.projekte\"\r\nsplom (~lange.projekte.variabel) \r\n# Matrixplot für alle variablen Einflüsse\r\n# kaum sicherer Bereich identifizierbar\r\n\r\n# Kombinationen der attributiven Merkmale erstellen\r\nwerte.attributiv = list(Lieferant=d.Lieferant, Kunde=d.Kunde,\r\n Methode=d.Methode, Review=d.Review, \r\n Komplexität=d.Komplexität)\r\nkombi = expand.grid(werte.attributiv)\r\n\r\n# Platzhalter \"Anzahl\" definieren, der so viele Stellen hat wie es Kombinationsmöglichkeiten der attributiven Merkmale gibt (hier: 48)\r\nAnzahl=numeric(nrow(kombi))\r\n# nrow() zählt Anzahl Zeilen (number of rows)\r\n\r\n# Für alle Kombinationsmöglichkeiten in \"kombi\" zählen, wie oft (Anzahl) OPG > 200~Tage im Datensatz \"lange.projekte\" ist\r\nfor(i in 1:nrow(kombi)){\r\n\tAnzahl[i] = nrow(subset(lange.projekte, \r\n\t\t(Lieferant==kombi$Lieferant[i])\r\n\t\t& (Kunde==kombi$Kunde[i]) \r\n\t\t& (Methode==kombi$Methode[i])\r\n\t\t& (Review==kombi$Review[i])\r\n\t\t& (Komplexität==kombi$Komplexität[i])))\r\n\t}\r\n\r\n# Anfügen der Spalte \"Anzahl\" an \"kombi\" und speichern als Datensatz \"Häufigkeit\"\r\nHäufigkeit=data.frame(cbind(kombi,Anzahl))\r\n# data.frame() erstellt Datenstruktur\r\n# cbind() verbindet Spalten (column bind)\r\n\r\n# Auswahl der Kombinationen, in denen nie OPG > 200 bzw. Anzahl(OPG>200)=0 ist\r\nSicher=subset(Häufigkeit, Anzahl==0)\r\n\r\nsummary(Sicher)\r\n# Kennzahlen für Datensatz \"Sicher\"\r\n# kein attributives Merkmal hat bei Projekten mit OPG <= 200 immer dieselbe Einstellung\r\n# tendenziell Merkmal Komplexität am eindeutigsten\r\n\r\n# Kombinationen der attributiven Merkmale mit OPG kleiner 200~Tage je Komplexitäts-Level\r\nsubset(Sicher, Komplexität==\"hoch\")\r\nsubset(Sicher, Komplexität==\"mittel\")\r\nsubset(Sicher, Komplexität==\"niedrig\")\t\r\n\r\n\r\n\r\n## Kapitel 11 ##\r\n#\r\n# Toleranzen\r\n\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 265: Test auf Normalverteilung und Wahrscheinlichkeitsnetz\r\n# Daten PLZ_1.CSV sind eingelesen und mit attach(PLZ_1.CSV) verfügbar gemacht\r\n\r\n# Test auf Normalverteilung (Anderson-Darling)\r\nrequire(nortest) # Paket nortest laden\r\nad.test(Laufzeit)\r\n# berechnet den Anderson-Darling-Test auf Normalverteilung (ad.test) und gibt die Werte aus\r\n\r\n# Wahrscheinlichkeitsnetz\r\nqqnorm(Laufzeit, xlab=\"Normalverteilungs-Quantile\",\r\n\tylab=\"Laufzeit\", main=\"Wahrscheinlichkeitsnetz\", pch=20)\r\nqqline(Laufzeit, col=\"blue\")\r\n# qqnorm zeichnet Wahrscheinlichkeitsnetz für Normalverteilung\r\n# xlab, ylab, main: Beschriftung x-Achse, y-Achse und Titel\r\n# qqline zeichnet Ideallinie (hier: blau)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 271: Toleranzgrenze ohne Normalverteilung\r\n# Daten PLZ_1.CSV sind eingelesen\r\n\r\nrequire(tolerance) # für die Toleranzrechnung muss das Package \"tolerance\" installiert sein\r\n\r\nnptol.int(Laufzeit, alpha=0.05, P=0.99, side=1)\r\n# nptol.int() berechnet verteilungsfreie Toleranzintervalle\r\n# alpha=0.05 entspricht Konfidenz- oder Vertrauensniveau von 95%\r\n# P=0.99 Abdeckung der Laufzeit (99%)\r\n# side=1 berechnet einseitige Toleranzgrenzen\r\n\r\ndistfree.est(n = length(Laufzeit), P = 0.99, side = 1)\r\n# berechnet die erreichte Konfidenz für 99% Abdeckung bei n=132 Werten\r\n\r\ndistfree.est(n = length(Laufzeit), alpha=0.05, side = 1)\r\n# berechnet die erreichbare Abdeckung bei n=132 Werten und 95% Konfidenz\r\n\r\ndistfree.est(P=0.99, alpha=0.05, side = 1)\r\n# berechnet die Mindestanzahl Messwerte für 99% Abdeckung und 95% Konfidenz\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 277: Monte Carlo-Toleranz\r\n# Daten PLZ_1.CSV sind eingelesen und mit attach(PLZ_1.CSV) verfügbar gemacht\r\n# lm.2 wurde berechnet \r\n\r\nSimu=10^4 # Anzahl Simulationen in \"Simu\" speichern\r\n\r\n# Zusammenstellung der Daten für die Monte Carlo-Simulation\r\nsim.PL=abs(rnorm(Simu, mean(PL), sd(PL)))\r\nsim.PMa=ceiling(abs(rnorm(Simu, mean(PMa), sd(PMa))))\r\nsim.Abt=ceiling(abs(rnorm(Simu, mean(Abteilungen),sd(Abteilungen))))\r\nsim.daten = data.frame(Lieferant=\"ja\",Kunde=\"ja\",Methode=\"SiSi\",Review=\"ja\",Komplexität=\"hoch\",PL=sim.PL,PMa=sim.PMa,Abteilungen=sim.Abt)\r\n\r\n# Berechnung der simulierten Projektdurchlaufzeiten\r\nLaufzeit.sim=predict(lm.2,sim.daten)\r\nMC.daten=cbind(sim.daten,Laufzeit.sim)\r\n\r\n# Boxplot simulierte Laufzeit\r\nboxplot(Laufzeit.sim, pch=8, col=\"lightgrey\", lty=1, data=MC.daten)\r\n\r\n# Test auf Normalverteilung (Anderson-Darling) und Wahrscheinlichkeitsnetz\r\nrequire(nortest) # Paket nortest laden\r\nad.test(Laufzeit.sim)\r\n\r\nqqnorm(Laufzeit, xlab=\"Normalverteilungs-Quantile\",\r\n\tylab=\"Laufzeit\", main=\"Wahrscheinlichkeitsnetz\", pch=20)\r\nqqline(Laufzeit, col=\"blue\")\r\n\r\n# Berechnen der 99%-Toleranzgrenze\r\nrequire(tolerance) # Paket tolerance laden\r\n \r\nnptol.int(Laufzeit.sim, alpha=0.05, P=0.99, side=1)\r\ndistfree.est(n = length(Laufzeit.sim), P = 0.99, side = 1)\r\n\r\n#++++++++++++++++++++++++++++++++++++++++++++++++#\r\n# S. 291: Bootstrap-Toleranz\r\n# Daten PLZ_1.CSV sind eingelesen und mit attach(PLZ_1.CSV) verfügbar gemacht\r\n# lm.2 wurde berechnet \r\n\r\n# Einstellwerte für Prognose\r\nx.p = data.frame(Lieferant=\"ja\",Kunde=\"ja\",Methode=\"SiSi\",\r\n\tReview=\"ja\",Komplexität=\"hoch\", PL=median(PL),PMa=median(PMa),\r\n\tAbteilungen=median(Abteilungen))\r\n\r\n# Prognosewert ydach_p\r\ny.p = predict(lm.2,x.p)\r\n\r\n# modifizierte Residuen\r\nri = lm.2$res/sqrt(1-lm.influence(lm.2)$hat) - mean(lm.2$res/sqrt(1-lm.influence(lm.2)$hat))\r\n# lm.2$res Residuen von lm.2\r\n# sqrt() Wurzelfunktion\r\n# lm.influence(lm.2)$hat Hebelwirkungen von lm.2\r\n# mean() Mittelwert\r\n\r\n# Einstellwerte für SPM\r\nx.werte = subset(PLZ.1, select=c(PL, PMa, Abteilungen, Lieferant, Kunde, Methode, Review, Komplexität))\r\n# Auswahl der Spalten mit Einstellwerten\r\n\r\nSimu=10^4 # Anzahl Simulationen in \"Simu\" speichern\r\n\r\n# Container für \"delta.stern\" und \"y.sim\" definieren\r\ndelta.stern = NULL\r\ny.sim = NULL\r\n\r\n# Werte \"delta.stern\" und \"y.sim\" berechnen\r\nfor(i in 1:Simu){ \r\n\te.stern = sample(ri, size=nrow(PLZ.1), replace=TRUE)\r\n\ty.stern = lm.2$fit + e.stern\r\n\tlm.2.stern = lm(y.stern ~ PL + PMa + Abteilungen + Lieferant \r\n\t\t+ Kunde + Methode + Review + Komplexität + I(PMa^2) \r\n\t\t+ PL:PMa + Abteilungen:Methode + Lieferant:Kunde \r\n\t\t+ Kunde:Komplexität + Methode:Review,\r\n\t\tdata=cbind(x.werte,y.stern))\r\n\ty.p.stern = predict(lm.2.stern,x.p)\r\n\te.p.stern = sample(ri, size=nrow(x.p), replace=TRUE)\r\n\tdelta.stern[i] = y.p.stern - (y.p+e.p.stern)\r\n\t# i-ter Wert für delta.stern\r\n\ty.sim[i] = y.p + delta.stern[i]\r\n\t# i-ter simulierte Bootstrap-Prognosewert\r\n\t}\r\n\r\n# Boxplot simulierte Laufzeit\r\nboxplot(y.sim, pch=8, col=\"lightgrey\", lty=1)\r\n\r\n# Test auf Normalverteilung (Anderson-Darling) und Wahrscheinlichkeitsnetz\r\nrequire(nortest) # Paket nortest laden\r\nad.test(y.sim)\r\n\r\nqqnorm(y.sim, xlab=\"Normalverteilungs-Quantile\",\r\n\tylab=\"Laufzeit\", main=\"Wahrscheinlichkeitsnetz\", pch=20)\r\nqqline(y.sim, col=\"blue\")\r\n\r\n# Berechnen der 99%-Toleranzgrenze\r\nrequire(tolerance) # Paket tolerance laden\r\nnormtol.int(y.sim, alpha=0.05, P=0.99, side=1)\r\n# normtol.int() berechnet Toleranzgrenzen für normalverteilte Werte\r\n\r\n", "meta": {"hexsha": "05ca06af155f4d17c246f24b0cd37dc11804ce7e", "size": 37220, "ext": "r", "lang": "R", "max_stars_repo_path": "Bsp2-Projektlaufzeit-R-Ed2/RCode Projektlaufzeit Ed2.r", "max_stars_repo_name": "barbara-bredner/NOT-Statistik2", "max_stars_repo_head_hexsha": "c9168312ab52ab27851580d0381fb8c2f2f9e443", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "Bsp2-Projektlaufzeit-R-Ed2/RCode Projektlaufzeit Ed2.r", "max_issues_repo_name": "barbara-bredner/NOT-Statistik2", "max_issues_repo_head_hexsha": "c9168312ab52ab27851580d0381fb8c2f2f9e443", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Bsp2-Projektlaufzeit-R-Ed2/RCode Projektlaufzeit Ed2.r", "max_forks_repo_name": "barbara-bredner/NOT-Statistik2", "max_forks_repo_head_hexsha": "c9168312ab52ab27851580d0381fb8c2f2f9e443", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 38.0572597137, "max_line_length": 138, "alphanum_fraction": 0.6757119828, "num_tokens": 12450, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7248702761768248, "lm_q2_score": 0.6959583313396338, "lm_q1q2_score": 0.5044795078457225}} {"text": "#' paleohydror: A package for implementing published routines in paleohydrologic reconstruction.\n#'\n#' This package offers several functions and routines for deriving estimates of paleohydrology from field observations of rock outcrops. Data such as grain size are important too.\n#'\n#' @section paleohydror functions:\n#'\n#' The functions here make up a mix of tools to calculate and estimate modern suficial processes and back-calculate those processes in ancient rivers.\n#'\n#' \\code{settle} calculates the settling velocity of particles following the formulation by Dietrich (1982).\n#'\n#' \\code{lynds_one} estimates ancient river slope using the first method described by Lynds et al. (2014)\n#'\n#' \\code{lynds_two} estimates ancient river slope using the second method described by Lynds et al. (2014)\n#'\n#' \\code{lynds_three} estimates ancient river slope using the third method described by Lynds et al. (2014)\n#'\n#' \\code{xset2H} estimates ancient river dune heights based on cross-set thicknesses. Methods from Leclair and Paola.\n#'\n#' \\code{H2h} estimates ancient river depths based on dune heights. Methods from Bradley and Venditti.\n#'\n#' \\code{xset2depthSimple} estimates ancient river depths directly from cross-set thicknesses using the simplest assumptions. Methods from Leclair, Paola, and Bradley and Venditti.\n#'\n#' \\code{azStringConvert} Azimuth conversion for 'quadrant' based strike/trend measurements.\n#'\n#' \\code{ancient_1} Estimating sediment flux in ancient rivers following Mahon + McElroy (2018).\n#'\n#' \\code{modern_1} Estimating sediment flux in modern rivers if bedform height is known following Mahon + McElroy (2018).\n#'\n#' \\code{modern_2} Estimating sediment flux in modern rivers if bedform length is known following Mahon + McElroy (2018).\n#'\n#' \\code{bedform_velocity} Estimating bedform migration rate as a function of reach-averaged slope.\n#'\n#' \\code{bedform_bedload} Estimating bedload as a function of bedform height and migration rate. After Simons et al. (1965)\n#'\n#' \\code{height_length} Estimating bedform height if bedform length is known.\n#'\n#' \\code{trampush_slp} estimates paleoslope using an empirical relationship with median bedload grain size and bankfull flow depth from Trampush (2014).\n#'\n#'\n#'\n#'\n#'\n#' @docType package\n#' @name paleohydror\nNULL\n", "meta": {"hexsha": "d3bc4909a3910237d78e257cb2e159de7a40669b", "size": 2299, "ext": "r", "lang": "R", "max_stars_repo_path": "R/paleohydror.r", "max_stars_repo_name": "ericbarefoot/paleohydror", "max_stars_repo_head_hexsha": "1870e634158b476121228f1d8fe82344af25dc9a", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/paleohydror.r", "max_issues_repo_name": "ericbarefoot/paleohydror", "max_issues_repo_head_hexsha": "1870e634158b476121228f1d8fe82344af25dc9a", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2018-04-17T16:32:53.000Z", "max_issues_repo_issues_event_max_datetime": "2018-04-17T16:32:53.000Z", "max_forks_repo_path": "R/paleohydror.r", "max_forks_repo_name": "ericbarefoot/paleohydror", "max_forks_repo_head_hexsha": "1870e634158b476121228f1d8fe82344af25dc9a", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 49.9782608696, "max_line_length": 180, "alphanum_fraction": 0.773814702, "num_tokens": 579, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8056321889812553, "lm_q2_score": 0.6261241632752915, "lm_q1q2_score": 0.50442578023353}} {"text": "########################### BayesKin ###########################\n## Bayesian estimates of rigid body kinematics\n## Andy Pohl\n## UofC - Faculty of Kinesiology\n\n## Written by: Andy Pohl\n## UofC - Faculty of Kinesiology\n## June-Dec 2020\n## Revision 1: June 2021\n################################################################\n\n##################### example.r ###################\n## Performs analysis comparing Bayesian to Least squares approaches for\n## 3 link kinematic chains.\n################################################################\nrm(list = ls())\nWORKING_DIR = \"\" # Replace with the location of the BayesKin/src directory on the local PC.\n\nsetwd(WORKING_DIR)\nINIT_TYPE = 'ls_est' # specify the type of initial values to be used for MCMC sampling - either 'true_vals', 'ls_est', 'random\nsource('library.r') # Retrieve function library\nset.seed(1) # Set seed for reproducibility\n\n\n## 1) Specify parameters\nn.links = 3 # Number of links.\nseg.length = c(0.45, 0.35, 0.25) # Length of segment \nr_true = c(0.07, 0.03) # Origin location in m.\ntheta_true = c(-55, -110, -10) # Rotation angle in deg. \nsigma_true = 1.5/1000 # Measurement noise in m. \ntrue_vals = c(r_true = r_true, theta_true = theta_true, sigma_true = sigma_true)\n\n# Generate posture\nlinks = list(gen_link(seg.length = seg.length[1],\n plate.center = 0.7*seg.length[1]),\n gen_link(seg.length = seg.length[2],\n plate.center = 0.5*seg.length[2]),\n gen_link(seg.length = seg.length[3],\n plate.center = 0.5*seg.length[3]))\nposture = gen_posture(links, r = r_true, theta = theta_true)\n\n# posture is a list with entries $r containing the location of the origin, $J containing the location of each joint\n# where J[[1]] is the origin and J[[4]] is the termination of the kinematic chain. $alpha gives the actual position markers\n\n# Generate observation\ny = gen_obs(posture, sigma_true)\n# y is a list of length nlinks with each list item containing a 2xnmarkers array of observed marker locations in the \n# global frame.\n\nplot_system(posture, y=y) # Provides a visualisation of the pose with observed markers in orange and true positions in blue\n\n## 2) Apply LS or Bayesian models\n#Compute LS solution.\nLS_result = LS_soln(y, links,\n inits = c(r_true, theta_true, sigma_true),\n init_type = 'true_vals')\n\nLS_result\n# The output of LS_result contains estimates of r and theta_i given by r.hat, theta.hat and $sigma.hat\n# along with 95% confidence intervals for each parameter in $intervals. $time denotes the computation time of the LS solution\n# $value provides the value of the cost function at the LS estimate.\n\n# Compute a Bayesian solution\nBayes_result = Bayes_soln(y=y,\n links = links,\n mdlfile = './BayesModel_p3.jags', # choose what model to use by specifying the appropiate .jags file here.\n init_type = INIT_TYPE,\n true_vals = true_vals,\n ls_est = c(LS_result$r.hat, LS_result$theta.hat, LS_result$sigma.hat))\n# Bayes_result provides a list with $fit being a 'coda' object containing the MCMC samples for the fitted model\n# $time provides the computational \nsummary(Bayes_result$fit) # provides posterior means for each pose parameter along with their 95% credible interval (via 2.5 and 97.5 quantiles)\nplot(Bayes_result$fit) # provides traceplots and density estimates for each parameter.\n", "meta": {"hexsha": "1fef8f45cc1adba9ee5932fd264260845358f591", "size": 3590, "ext": "r", "lang": "R", "max_stars_repo_path": "src/example.r", "max_stars_repo_name": "AndyPohlNZ/BayesKin", "max_stars_repo_head_hexsha": "1eac3d30f09491d8b371a433e4621a53a2b5d2ef", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/example.r", "max_issues_repo_name": "AndyPohlNZ/BayesKin", "max_issues_repo_head_hexsha": "1eac3d30f09491d8b371a433e4621a53a2b5d2ef", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/example.r", "max_forks_repo_name": "AndyPohlNZ/BayesKin", "max_forks_repo_head_hexsha": "1eac3d30f09491d8b371a433e4621a53a2b5d2ef", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 48.5135135135, "max_line_length": 144, "alphanum_fraction": 0.6320334262, "num_tokens": 849, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8006919925839875, "lm_q2_score": 0.6297746074044135, "lm_q1q2_score": 0.5042554852814383}} {"text": "\n#### covidWBE.r\n#### Nick Crosbie, May 2020\n#### source(\"./bin/covidWBE.r\")\n\nlibrary(EnvStats)\n\n# WASTEWATER CATCHMENT (calculate dry weather flow in millilitres per day)\ncatchmentPop <- 10^4 # number of people in catchment\ndryWeatherFlowPerPerson <- 2.12799e-4 # dry weather flow contributed by each person, in ** megalitres ** per day (average derived from all participating ColoSSoS WWTP catchments)\ndryWeatherFlow_MLD <- catchmentPop * dryWeatherFlowPerPerson # catchment dry weather flow, in ** megalitres ** per day\ndryWeatherFlow_mLD <- dryWeatherFlow_MLD * 1e9 # catchment dry weather flow, in ** millilitres ** per day\n\n# SARS-CoV-2 GENOMES SHED PER DAY TO CATCHMENT BY DEFECATION (number of genomes)\nmodeShed <- 10^3 # mode number of genomes shed in faecies by an infected person (genomes per millilitre of faecies)\nminShed <- 10^2 # min number of genomes shed in faecies by an infected person (genomes per millilitre of faecies)\nmaxShed <- 10^7 # max number of genomes shed in faecies by an infected person (genomes per millilitre of faecies)\n\ndefVol <- 200 # volume of faecies defacated per day by an individual, in millilitres (no distinction made between volume defecated by infected and non-infected individuals)\nnumberShedding <- 1 # number of people shedding in catchment\ndefVolInfCat <- defVol * numberShedding # combined volume of faecies defacated by shedders per day, in millilitres\n\nmodeShedCat <- modeShed * defVolInfCat # mode number of genomes shed per day to catchment by infected population\nminShedCat <- minShed * defVolInfCat # minimum number of genomes shed per day to catchment by infected population\nmaxShedCat <- maxShed * defVolInfCat # maximum number of genomes shed per day to catchment by infected population\n\n# TRIANGULAR DISTRIBUTION OF SARS-CoV-2 GENOMES SHED PER DAY TO CATCHMENT BY DEFECATION (number of genomes)\ntriagShedCat <- simulateVector(n = 10000, distribution = \"tri\", param.list = list(min = minShedCat, mode = modeShedCat, max = maxShedCat),\nsample.method = \"LHS\", seed = 47)\n\npdfPlot(distribution = \"tri\", param.list = list(min = minShedCat, mode = modeShedCat, max = maxShedCat), n.points = 10000)\n\n# SARS-CoV-2 CONCENTRATION AT POINT OF WASTEWATER SUBSAMPLING (genomes per mL)\ncatchmentDilution <- defVolInfCat / dryWeatherFlow_mLD\nwastewaterConc <- triagShedCat * catchmentDilution # genomes per mL of catchment wastewater, ** assuming fully mixed catchment **\n\n# plot(ecdf(wastewaterConc))\n\n# SARS-CoV-2 CONCENTRATION OF FINAL CONCENTRATED PREPARATION\nconcFactor <- 4 # concentration factor from point of wastewater sampling to final concentrated preparation\nfinalPrepConc <- wastewaterConc * concFactor # virus concentration prior to subsampling for PCR (genomes per millilitre)\n\n#plot(ecdf(finalPrepConc))\n\n# SARS-CoV-2 GENOMES PER PCR REACTION (number of genomes)\nfinalConcVol <- 2 # volume of sample prior to subsampling for PCR reaction, in millilitres\ngenomesFinalPrep <- finalPrepConc * finalConcVol # number of genomes contained in the final prep (step immediately prior to PCR)\n\ntemplateVol <- 0.005 # volume in which the final preparation is resuspended prior to pipetting to PCR reaction\nvolRxn <- 0.05 # volume of one PCR reaction, in millilitres (i.e. 50 microlitres)\ngenomesRxn <- genomesFinalPrep * (templateVol/volRxn) # genomes per reaction\nLOD <- 10 # PCR limit of detection as genomes per reaction\n\nn <- 10000 # number of wastewater samples analysed for SARS-CoV-2\ndetects <- sum(genomesRxn > LOD) # number of SARS-CoV-2 detects\ntrueDetectProportion <- 0.90 # proportion of detects that are true positives\npercentRealDetects <- ((detects * trueDetectProportion) / n) * 100 # percentage of samples where SARS-CoV-2 is detected \n\nprint(percentRealDetects)\n\n# plot(ecdf(genomesRxn))\n\n\n", "meta": {"hexsha": "6c0560d9ce52378a003392202aa43beecbc4cc7e", "size": 3762, "ext": "r", "lang": "R", "max_stars_repo_path": "covidWBE.r", "max_stars_repo_name": "pprevos/sars-cov-2-analysis", "max_stars_repo_head_hexsha": "e87fca4d0c409367a96f6bc543795e9ee2920c7d", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "covidWBE.r", "max_issues_repo_name": "pprevos/sars-cov-2-analysis", "max_issues_repo_head_hexsha": "e87fca4d0c409367a96f6bc543795e9ee2920c7d", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "covidWBE.r", "max_forks_repo_name": "pprevos/sars-cov-2-analysis", "max_forks_repo_head_hexsha": "e87fca4d0c409367a96f6bc543795e9ee2920c7d", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 58.78125, "max_line_length": 178, "alphanum_fraction": 0.7756512493, "num_tokens": 1054, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.849971175657575, "lm_q2_score": 0.5926665999540698, "lm_q1q2_score": 0.5037495267359383}} {"text": "#==============================================================================\n#\tFunctions used for MESS.\n#==============================================================================\n\n#------------------------------------------------------------------------------\n#'\t(Internal) Calculate distance between test and training data using MESS\n#'\n#'\tThis function calculate multivariate environmental similarity surface\n#'\t(MESS, Elith, Kearney and Phillips 2010) for each point in test data.\n#'\n#'\t@param data.train a data.frame containing training data.\n#'\t@param data.test a data.frame containing test data.\n#'\t@param na.rm a logical indicating ignorance of NA.\n#'\n#'\t@section References:\n#'\tElith, J., M. Kearney, and S. Phillips. 2010.\n#'\tThe art of modelling range-shifting species.\n#'\tMethods in Ecology and Evolution 1:330-342.\n#------------------------------------------------------------------------------\ncalculate.mess <- function(data.train, data.test, na.rm = TRUE) {\n\t# Prepare matrix for the result.\n\tnumeric.columns <- colnames(data.test)[sapply(data.test, is.numeric)]\n\tsimilarity <- matrix(\n\t\tNA, nrow = nrow(data.test), ncol = length(numeric.columns)\n\t)\n\tcolnames(similarity) <- numeric.columns\n\t# Iterate over numeric columns.\n\tfor (i in numeric.columns) {\n\t\t# min_i\n\t\tmin.i <- min(data.train[[i]], na.rm = na.rm)\n\t\t# max_i\n\t\tmax.i <- max(data.train[[i]], na.rm = na.rm)\n\t\t# f_i for all p_i\n\t\tsmaller.fraction <- sapply(\n\t\t\tdata.test[[i]],\n\t\t\tfunction(x) mean(data.train[[i]] < x, na.rm = na.rm) * 100\n\t\t)\n\t\t# Similarity: f_i = 0\n\t\tcase_0 <- (data.test[[i]] - min.i) / (max.i - min.i) * 100\n\t\t# Similarity: 0 < f_i <= 50\n\t\tcase_0_50 <- 2 * smaller.fraction\n\t\t# Similarity: 50 <= f_i < 100\n\t\tcase_50_100 <- 2 * (100 - smaller.fraction)\n\t\t# Similarity: f_i = 100\n\t\tcase_100 <- (max.i - data.test[[i]]) / (max.i - min.i) * 100\n\t\tsimilarity[, i] <- ifelse(\n\t\t\tsmaller.fraction == 0, case_0,\n\t\t\tifelse(\n\t\t\t\tsmaller.fraction <= 50, case_0_50,\n\t\t\t\tifelse(smaller.fraction < 100, case_50_100, case_100)\n\t\t\t)\n\t\t)\n\t}\n\treturn(apply(similarity, 1, min))\n}\n\n\n#------------------------------------------------------------------------------\n#'\tCreate a function calculate MESS of a group.\n#'\n#'\tThis function creates a function which calculate aggregated MESS for a\n#'\tgroup of points.\n#'\n#'\t@param columns\n#'\t\tnames of columns or index of columns used for calculation of MESS.\n#'\t@param aggregate.fun\n#'\t\ta function used for aggregation of values of MESS of a group.\n#'\t\tFunctions like \\code{mean}, \\code{median}, \\code{max} and \\code{min}\n#'\t\tas well as other functions which accept vector of values and return\n#'\t\tsingle value can be used.\n#'\t@param na.rm\n#'\t\tlogical indicating whether NA values should be ignored during\n#'\t\tcalculation of MESS.\n#'\t@param ...\n#'\t\tother arguments passed for \\code{aggregate.fun}.\n#'\n#'\t@export\n#------------------------------------------------------------------------------\nmess <- function(columns = NULL, aggregate.fun = mean, na.rm = TRUE, ...) {\n\tf <- function(data.train, data.test) {\n\t\tif (!is.null(columns)) {\n\t\t\tdata.train <- data.train[columns]\n\t\t\tdata.test <- data.test[columns]\n\t\t}\n\t\treturn(\n\t\t\taggregate.fun(calculate.mess(data.train, data.test, na.rm), ...)\n\t\t)\n\t}\n\treturn(f)\n}\n\n\n#==============================================================================\n#\tFunctions and data used for geographic distance.\n#==============================================================================\n\n#------------------------------------------------------------------------------\n#'\t(Internal) Parameters for several earth ellipsoids.\n#------------------------------------------------------------------------------\nELLIPSOIDS = list(\n grs80 = list(\n semi.major.axis.m = 6378137,\n semi.minor.axis.m = 6356752.3141403561,\n inverse.flattening = 298.25722210100002\n\t),\n wgs84 = list(\n semi.major.axis.m = 6378137,\n semi.minor.axis.m = 6356752.3142451793,\n inverse.flattening = 298.25722356300003\n\t),\n bessel1841 = list(\n semi.major.axis.m = 6377397.1550000003,\n semi.minor.axis.m = 6356078.9628181886,\n inverse.flattening = 299.15281279999999\n\t)\n)\n\n\n#------------------------------------------------------------------------------\n#'\t(Internal) Convert degree to radian.\n#'\n#'\t@param x values in degree.\n#------------------------------------------------------------------------------\ndeg2rad <- function(x) {\n\treturn(x / 180 * pi)\n}\n\n\n#------------------------------------------------------------------------------\n#'\t(Internal) Find parameters for the Vincenty's formulae.\n#'\n#'\t@param f\n#'\t\tflattening of the ellipsoid.\n#'\t@param L\n#'\t\tdifference in longitude of two points.\n#'\t@param U1\n#'\t\treduced latitude (latitude on the auxiliary sphere) of point 1.\n#'\t@param U2\n#'\t\treduced latitude (latitude on the auxiliary sphere) of point 2.\n#'\t@param threshold\n#'\t\tthreshold difference to determine convergence.\n#'\t\tThe value of 1e-12 produces precision of 0.06mm.\n#'\t@param maxit\n#'\t\tmaximum number of iteration to test convergence.\n#------------------------------------------------------------------------------\nfind.vincenty.parameters <- function(f, L, U1, U2, threshold, maxit) {\n\tlambda.prev <- lambda <- L\n\tfor (i in 1:maxit) {\n\t\tsin.sigma <- sqrt(\n\t\t\t(cos(U2) * sin(lambda)) ^ 2\n\t\t\t+ (cos(U1) * sin(U2) - sin(U1) * cos(U2) * cos(lambda)) ^ 2\n\t\t)\n\t\tcos.sigma <- sin(U1) * sin(U2) + cos(U1) * cos(U2) * cos(lambda)\n\t\tsigma <- atan2(sin.sigma, cos.sigma)\n\t\tsin.alpha <- cos(U1) * cos(U2) * sin(lambda) / sin.sigma\n\t\tcos2.alpha <- 1 - sin.alpha ^ 2\n\t\tcos.2sigma.m <- cos.sigma - (2 * sin(U1) * sin(U2) / cos2.alpha)\n\t\tC <- f / 16 * cos2.alpha * (4 + f * (4 - 3 * cos2.alpha))\n\t\tlambda <- (\n\t\t\tL + (1 - C) * f * sin.alpha * (\n\t\t\t\tsigma + C * sin.sigma * (\n\t\t\t\t\tcos.2sigma.m + C * cos.sigma * (\n\t\t\t\t\t\t-1 + 2 * (cos.2sigma.m ^ 2)\n\t\t\t\t\t)\n\t\t\t\t)\n\t\t\t)\n\t\t)\n\t\tif (all(abs(lambda - lambda.prev) <= threshold)) {\n\t\t\tbreak\n\t\t}\n\t\tlambda.prev <- lambda\n\t}\n\tif (i == maxit) {\n\t\tstop(\n\t\t\t\"Couldn't determine distance(s).\\n\",\n\t\t\t\"Try increasing the value of 'maxit'.\"\n\t\t)\n\t}\n\tresult <- list(\n\t\tsigma = sigma, sin.sigma = sin.sigma, cos.sigma = cos.sigma,\n\t\tcos2.alpha = cos2.alpha, cos.2sigma.m = cos.2sigma.m\n\t)\n\treturn(result)\n}\n\n\n#------------------------------------------------------------------------------\n#'\t(Internal) Calculate distance using the Vincenty's formulae.\n#'\n#'\t@param a\n#'\t\tsemi major axis length in m.\n#'\t@param b\n#'\t\tsemi minor axis length in m.\n#'\t@param lon1\n#'\t\tvector of longitude (x) of point 1 in radian.\n#'\t@param phi1\n#'\t\tvector latitude (y) of point 1 in radian.\n#'\t@param lon2\n#'\t\tvector of longitude (x) of point 2 in radian.\n#'\t@param phi2\n#'\t\tvector of latitude (y) of point 2 in radian.\n#'\t@param threshold\n#'\t\tthreshold difference to determine convergence.\n#'\t\tThe value of 1e-12 produces precision of 0.06mm.\n#'\t@param maxit\n#'\t\tmaximum number of iteration to try before convergence.\n#'\n#'\t@section References:\n#'\t\tVincenty, T. (1975)\n#'\t\tDirect and inverse solutions of geodesics on the ellipsoid with\n#'\t\tapplication of nested equations. Survey Review 23:88-93.\n#------------------------------------------------------------------------------\nvincenty.distance <- function(\n\ta, b, lon1, phi1, lon2, phi2, threshold = 1e-12, maxit = 1000\n) {\n\t# Prepare parameters.\n\tf <- (a - b) / a\n\tU1 <- atan((1 - f) * tan(phi1))\n\tU2 <- atan((1 - f) * tan(phi2))\n\tL <- lon2 - lon1\n\t# Find the Vincenty's formulae's parameters.\n\tparams <- find.vincenty.parameters(f, L, U1, U2, threshold, maxit)\n\t# Calculate distances.\n\tmu2 <- params$cos2.alpha * (a ^ 2 - b ^ 2) / b ^ 2\n\tA <- 1 + mu2 / 16384 * (4096 + mu2 * (-768 + mu2 * (320 - 175 * mu2)))\n\tB <- mu2 / 1024 * (256 + mu2 * (- 128 + mu2 * (74 - 47 * mu2)))\n\tdelta.sigma <- B * params$sin.sigma * (\n\t\tparams$cos.2sigma.m\n\t\t+ B / 4 * (\n\t\t\tparams$cos.sigma * (-1 + 2 * params$cos.2sigma.m ^ 2)\n\t\t\t- B / 6 * (\n\t\t\t\tparams$cos.2sigma.m * (-3 + 4 * params$sin.sigma ^ 2)\n\t\t\t\t* (-3 + 4 * params$cos.2sigma.m ^ 2)\n\t\t\t)\n\t\t)\n\t)\n\ts <- b * A * (params$sigma - delta.sigma)\n\treturn(s)\n}\n\n\n#------------------------------------------------------------------------------\n#'\tCreate a function calculates geographic distance.\n#'\n#'\t@param x.name\n#'\t\tcolumn name of longitude.\n#'\t\tData should be stored in decimal degree.\n#'\t@param y.name\n#'\t\tcolumn name of latitutde.\n#'\t\tData should be stored in decimal degree.\n#'\t@param ellipsoid\n#'\t\tname of earth ellipsoids, can be \"grs80\", \"wgs84\" or \"bessel1841\".\n#'\t@param method\n#'\t\tthe method used for calculation of distance.\n#'\t\tCurrently, only the method using the Vincenty's formulae is\n#'\t\timplimented.\n#'\t@param threshold\n#'\t\tthreshold difference to determine convergence.\n#'\t\tThe value of 1e-12 produces precision of 0.06mm.\n#'\t@param maxit\n#'\t\tmaximum number of iteration to try before convergence.\n#'\n#'\t@section Details:\n#'\t\t\\url{https://en.wikipedia.org/wiki/Vincenty's_formulae}.\n#'\n#'\t@export\n#------------------------------------------------------------------------------\ngeographic.distance <- function(\n\tx.name, y.name, ellipsoid = c(\"grs80\", \"wgs84\", \"bessel1841\"),\n\tmethod = c(\"vincenty\"), threshold = 1e-12, maxit = 10000\n) {\n\tellipsoid <- match.arg(ellipsoid)\n\ta <- ELLIPSOIDS[[ellipsoid]]$semi.major.axis.m\n\tb <- ELLIPSOIDS[[ellipsoid]]$semi.minor.axis.m\n\tfun <- function(data.train, data.test, na.rm = TRUE) {\n\t\tphi1 <- deg2rad(mean(data.train[y.name], na.rm = na.rm))\n\t\tphi2 <- deg2rad(mean(data.test[y.name], na.rm = na.rm))\n\t\tlon1 <- deg2rad(mean(data.train[x.name], na.rm = na.rm))\n\t\tlon2 <- deg2rad(mean(data.test[x.name], na.rm = na.rm))\n\t\td <- vincenty.distance(a, b, lon1, phi1, lon2, phi2, threshold, maxit)\n\t\treturn(d)\n\t}\n\treturn(fun)\n}\n\n\n#==============================================================================\n#\tFunctions used for forecast horizon.\n#==============================================================================\n\n#------------------------------------------------------------------------------\n#'\t(Internal) Calculate distance between training and test datasets.\n#'\n#'\t@param fold.index\n#'\t\tan integer representing index of the fold for which the distance\n#'\t\tbetween test and training datasets is calculated.\n#'\t@param data\n#'\t\ta data.frame containing explanatory variable(s).\n#'\t@param cv.group\n#'\t\ta vector of integer representing grouping of observation.\n#'\t@param distance.fun\n#'\t\ta function calculating distance between test and training datasets.\n#'\t\tIt should accept two data.frames in the following form and calculate\n#'\t\tdistance between them: \\code{distance.fun(training.data, test.data)}.\n#'\t\tFunctions produced by \\code{\\link{geographic.distance}} and\n#'\t\t\\code{\\link{mess}} can be used.\n#------------------------------------------------------------------------------\ncalculate.distance.between.train.and.test <- function(\n\tfold.index, data, cv.group, distance.fun\n) {\n\tdata.train <- data[cv.group != fold.index, , drop = FALSE]\n\tdata.test <- data[cv.group == fold.index, , drop = FALSE]\n\treturn(distance.fun(data.train, data.test))\n}\n\n\n#------------------------------------------------------------------------------\n#' (Experimental) Draw a forecast horizon graph.\n#'\n#'\tDraw a forecast horizon graph using result of the cross validation.\n#'\tThis function is experimental.\n#'\n#'\t@param object\n#'\t\ta \\code{cv.models} object.\n#'\t@param metric.name\n#'\t\ta character representing name of metrics to be drawn.\n#'\t@param distance.fun\n#'\t\ta function which calculates distance between training and test dataset.\n#'\t\tIt should accept two data.frames in the following form and calculate\n#'\t\tdistance between them: \\code{distance.fun(training.data, test.data)}.\n#'\t\tFunctions produced by \\code{\\link{geographic.distance}} and\n#'\t\t\\code{\\link{mess}} can be used.\n#'\t@param index\n#'\t\tan index of candidate model in the \\code{cv.models} object.\n#'\t@param ylab\n#'\t\tlabel of Y axis.\n#'\t@param xlab\n#'\t\tlabel of X axis.\n#'\t@param draw\n#'\t\tlogical indicating whether the plot is draw.\n#'\t\tIf FALSE, this function only returns calculated results and draw\n#'\t\tnothing.\n#'\t@param ...\n#'\t\tgraphical parameters passed to plot function.\n#'\n#'\t@return\n#'\ta data.frame having distance between training data and corresponding\n#'\tperformance measure.\n#'\n#'\t@export\n#------------------------------------------------------------------------------\nforecast.horizon <- function(\n\tobject,\n\tmetric.name = ifelse(\n\t\tobject$adapter$model.type == \"regression\", \"q.squared\", \"mcc\"\n\t),\n\tdistance.fun = mess(), index = 1, ylab = metric.name,\n\txlab = deparse(substitute(distance.fun)), draw = TRUE, ...\n) {\n\t# Extract explanatory variables.\n\tdata <- object$adapter$data[object$adapter$x.names()]\n\t# Extract index.\n\tcv.group <- object$cv.results[[index]]$cv.group\n\t# Extract metric.\n\tobject$aggregate.method = \"folds\"\n\tmetrics <- cv.metrics(object, list(object$cv.results[[index]]$fits))[[1]]\n\t# Calculate distance between training and test data.\n\tdistance <- sapply(\n\t\t1:object$folds, calculate.distance.between.train.and.test,\n\t\tdata = data, cv.group = cv.group, distance.fun = distance.fun\n\t)\n\tgraphics::plot(\n\t\tdistance, metrics[, metric.name], ylab = ylab, xlab = xlab, ...\n\t)\n\tinvisible(cbind(data.frame(distance = distance), metrics))\n}\n", "meta": {"hexsha": "301694861586acd4277cbd0ac40f5977e647f096", "size": 13173, "ext": "r", "lang": "R", "max_stars_repo_path": "R/prediction.horizon.r", "max_stars_repo_name": "Marchen/cv.models", "max_stars_repo_head_hexsha": "70af64f72933a4172d229413ff43034a53c93163", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2019-07-01T15:45:35.000Z", "max_stars_repo_stars_event_max_datetime": "2019-07-01T15:45:35.000Z", "max_issues_repo_path": "R/prediction.horizon.r", "max_issues_repo_name": "Marchen/cv.models", "max_issues_repo_head_hexsha": "70af64f72933a4172d229413ff43034a53c93163", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 8, "max_issues_repo_issues_event_min_datetime": "2019-03-11T04:21:27.000Z", "max_issues_repo_issues_event_max_datetime": "2019-07-10T12:18:14.000Z", "max_forks_repo_path": "R/prediction.horizon.r", "max_forks_repo_name": "Marchen/cv.models", "max_forks_repo_head_hexsha": "70af64f72933a4172d229413ff43034a53c93163", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2019-11-16T03:30:36.000Z", "max_forks_repo_forks_event_max_datetime": "2019-11-16T03:30:36.000Z", "avg_line_length": 34.9416445623, "max_line_length": 79, "alphanum_fraction": 0.5678281333, "num_tokens": 3464, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7905303285397348, "lm_q2_score": 0.6370307944803832, "lm_q1q2_score": 0.5035921632505057}} {"text": "################################################################################\r\n# Part of the R/EpiILM package\r\n#\r\n# AUTHORS:\r\n# Vineetha Warriyar. K. V. ,\r\n# Waleed Almutiry , and\r\n# Rob Deardon \r\n#\r\n# Algorithm based on:\r\n# Deardon R, Brooks, S. P., Grenfell, B. T., Keeling, M. J., Tildesley,\r\n# M. J., Savill, N. J., Shaw, D. J., Woolhouse, M. E. (2010).\r\n# Inference for individual level models of infectious diseases in large\r\n# populations. Statistica Sinica, 20, 239-261.\r\n#\r\n# Free software under the terms of the GNU General Public License, version 2,\r\n# a copy of which is available at http://www.r-project.org/Licenses/.\r\n################################################################################\r\n\r\nepidata <- function(type, n, tmin = NULL, tmax, sus.par, trans.par = NULL, beta = NULL, spark = NULL,\r\n Sformula = NULL, Tformula = NULL, x = NULL, y = NULL,\r\n inftime = NULL, infperiod = NULL, contact = NULL) {\r\n\r\n # Error checks for input arguments\r\n if (any(is.null(type) | !(type %in% c(\"SI\", \"SIR\"))) == TRUE) {\r\n stop(\"Specify type as \\\"SI\\\" or \\\"SIR\\\" \", call. = FALSE)\r\n }\r\n\r\n ns <- length(sus.par)\r\n if (!is.null(trans.par)) {\r\n\r\n nt <- length(trans.par)\r\n flag.trans <- 1\r\n phi <- trans.par\r\n\r\n } else {\r\n\r\n nt <- 1\r\n flag.trans <- 0\r\n phi <- 1\r\n\r\n }\r\n\r\n if (all(is.null(contact) & (is.null(x) | is.null(y))) == TRUE) {\r\n stop('epidata: Specify contact network or x, y coordinates')\r\n }\r\n\r\n if (is.null(contact)) {\r\n ni <- length(beta)\r\n if (ni != 1) {\r\n stop(\"epidata: The input of beta has more than one value while the considered distance-based ILM needs only one spatial parameter, beta.\", call. = FALSE)\r\n }\r\n if (!is.null(x)) {\r\n if (any((length(y) != n) | (length(x) != n)) == TRUE) {\r\n stop('epidata: Length of x or y is not compatible ')\r\n }\r\n }\r\n } else {\r\n if (length(contact)/(n * n) == 1) {\r\n ni <- 1\r\n if (is.null(beta)) {\r\n beta <- 1\r\n } else {\r\n stop(\"epidata: As the model has only one contact network, The model does not need a network parameter beta, beta must be assigned to its default values = NULL\", call. = FALSE)\r\n }\r\n } else if (length(contact)/(n * n) > 1) {\r\n ni <- length(beta)\r\n if (length(contact)/(n * n) != ni) {\r\n stop('epidata: Dimension of beta and the number of contact networks are not matching')\r\n }\r\n }\r\n network <- array(contact, c(n, n, ni))\r\n }\r\n\r\n if (is.null(tmin)){\r\n tmin <- 1\r\n }\r\n\r\n if (is.null(spark)) {\r\n spark <- 0\r\n }\r\n\r\n if (!is.null(inftime)) {\r\n if ((length(inftime) != n)) {\r\n stop('epidata: Length of inftime is not compatible ')\r\n }\r\n } else {\r\n inftime <- rep(0, n)\r\n }\r\n\r\n if (all(is.null(infperiod) & type == \"SIR\") == TRUE) {\r\n stop(' epidata: Specify removal, infperiod')\r\n }\r\n\r\n if (!is.null(infperiod)) {\r\n if (length(infperiod) != n) {\r\n stop('epidata: Length of infperiod is not compatible')\r\n }\r\n if (type == \"SI\") {\r\n stop('epidata: Type must be \"SIR\"')\r\n }\r\n remt <- rep(0, n)\r\n }\r\n\r\n # formula for susceptibility (covariate) function\r\n if (!is.null(Sformula)) {\r\n covmat.sus <- model.matrix(Sformula)\r\n\r\n if (all((ncol(covmat.sus) == length(all.vars(Sformula))) & (ns != length(all.vars(Sformula)))) == TRUE) {\r\n stop('epidata: Check Sformula (no intercept term) and the dimension of sus.par')\r\n }\r\n\r\n if (all((ncol(covmat.sus) > length(all.vars(Sformula))) & (ns != ncol(covmat.sus))) == TRUE) {\r\n stop('epidata: Check Sformula (intercept term) and the dimension of sus.par')\r\n }\r\n } else {\r\n if (ns == 1) {\r\n covmat.sus <- matrix(1.0, nrow = n, ncol = ns)\r\n }\r\n if (ns > 1) {\r\n stop('epidata: Please specify the susceptibility covariate')\r\n }\r\n }\r\n\r\n # formula for transmissibility (covariate) function\r\n if (flag.trans == 1) {\r\n if (is.null(Tformula)) {\r\n stop(\"epidata: Tformula is missing. It has to be specified with no intercept term and number of columns equal to the length of trans.par\", call. = FALSE)\r\n } else if (!is.null(Tformula)) {\r\n \t\tcovmat.trans <- model.matrix(Tformula)\r\n\r\n \t\tif (all((ncol(covmat.trans) == length(all.vars(Tformula))) & (nt != length(all.vars(Tformula)))) == TRUE) {\r\n \t\t\tstop(\"epidata: Check Tformula. It has to be with no intercept term and number of columns equal to the length of trans.par\", call. = FALSE)\r\n \t\t}\r\n }\r\n } else if (flag.trans == 0) {\r\n \tcovmat.trans <- matrix(1.0, nrow = n, ncol = nt)\r\n }\r\n\r\n # Calling fortran subroutine - Purely Spatial: Susceptible-Infectious(SI)\r\n if (all((type == \"SI\") & is.null(contact)) == TRUE) {\r\n tmp <- .Fortran(\"dataxy\",\r\n x = as.vector(x, mode = \"double\"),\r\n y = as.vector(y, mode = \"double\"),\r\n n = as.integer(n),\r\n tmin = as.integer(tmin),\r\n tmax = as.integer(tmax),\r\n ns = as.integer(ns),\r\n nt = as.integer(nt),\r\n ni = as.integer(ni),\r\n alpha = as.vector(sus.par, mode = \"double\"),\r\n phi = as.vector(phi, mode = \"double\"),\r\n beta =as.vector(beta, mode = \"double\"),\r\n spark = as.numeric(spark),\r\n covmatsus = matrix(as.double(covmat.sus), ncol = ncol(covmat.sus), nrow = n),\r\n covmattrans = matrix(as.double(covmat.trans), ncol = ncol(covmat.trans), nrow = n),\r\n tau = as.vector(inftime, mode = \"integer\")\r\n )\r\n\r\n result1 <- list(type = type, XYcoordinates = cbind(x, y), contact = NULL, inftime = tmp$tau)\r\n }\r\n\r\n # Calling fortran subroutine - Purely Spatial: Susceptible-Infectious-Removed (SIR)\r\n if (all((type == \"SIR\") & is.null(contact)) == TRUE) {\r\n tmp <- .Fortran(\"dataxysir\",\r\n n = as.integer(n),\r\n tmin = as.integer(tmin),\r\n tmax = as.integer(tmax),\r\n ns = as.integer(ns),\r\n nt = as.integer(nt),\r\n ni = as.integer(ni),\r\n alpha = as.vector(sus.par, mode = \"double\"),\r\n phi = as.vector(phi, mode = \"double\"),\r\n beta =as.vector(beta, mode = \"double\"),\r\n spark = as.numeric(spark),\r\n covmatsus = matrix(as.double(covmat.sus), ncol = ncol(covmat.sus), nrow = n),\r\n covmattrans = matrix(as.double(covmat.trans), ncol = ncol(covmat.trans), nrow = n),\r\n lambda = as.vector(infperiod, mode = \"integer\"),\r\n x = as.vector(x, mode = \"double\"),\r\n y = as.vector(y, mode = \"double\"),\r\n tau = as.vector(inftime, mode = \"integer\"),\r\n remt = as.vector(remt, mode = \"integer\")\r\n )\r\n\r\n result1 <- list(type = type, XYcoordinates = cbind(x, y), contact = NULL, inftime = tmp$tau, remtime = tmp$remt)\r\n }\r\n\r\n # Calling fortran subroutine - Contact networks: Susceptible-Infectious (SI)\r\n if (all((type == \"SI\") & !is.null(contact)) == TRUE) {\r\n tmp <- .Fortran(\"datacon\",\r\n n = as.integer(n),\r\n tmin = as.integer(tmin),\r\n tmax = as.integer(tmax),\r\n ns = as.integer(ns),\r\n nt = as.integer(nt),\r\n ni = as.integer(ni),\r\n alpha = as.numeric(sus.par),\r\n phi = as.numeric(phi),\r\n beta = as.numeric(beta),\r\n spark = as.numeric(spark),\r\n covmatsus = matrix(as.double(covmat.sus), ncol = ncol(covmat.sus), nrow = n),\r\n covmattrans = matrix(as.double(covmat.trans), ncol = ncol(covmat.trans), nrow = n),\r\n network = as.vector(network),\r\n tau = as.integer(inftime)\r\n )\r\n\r\n result1 <- list(type = type, XYcoordinates = cbind(x, y), contact = contact, inftime = tmp$tau)\r\n } else if (all((type == \"SIR\") & !is.null(contact)) == TRUE) {\r\n # Calling fortran subroutine - Contact networks: Susceptible-Infectious-Removed (SIR)\r\n tmp <- .Fortran(\"dataconsir\",\r\n n = as.integer(n),\r\n tmin = as.integer(tmin),\r\n tmax = as.integer(tmax),\r\n ns = as.integer(ns),\r\n nt = as.integer(nt),\r\n ni = as.integer(ni),\r\n lambda = as.integer(infperiod),\r\n alpha = as.numeric(sus.par),\r\n phi = as.numeric(phi),\r\n beta = as.numeric(beta),\r\n spark = as.numeric(spark),\r\n covmatsus = matrix(as.double(covmat.sus), ncol = ncol(covmat.sus), nrow = n),\r\n covmattrans = matrix(as.double(covmat.trans), ncol = ncol(covmat.trans), nrow = n),\r\n network = as.vector(network),\r\n tau = as.integer(inftime),\r\n remt = as.integer(remt)\r\n )\r\n\r\n result1 <- list(type = type, XYcoordinates = cbind(x, y), contact = contact, inftime = tmp$tau, remtime = tmp$remt)\r\n }\r\n class(result1) <- \"epidata\"\r\n\r\n result1\r\n # End of function\r\n}\r\n", "meta": {"hexsha": "2a57634398dae366a5828c127a7374a916b86326", "size": 9640, "ext": "r", "lang": "R", "max_stars_repo_path": "R/epidata.r", "max_stars_repo_name": "waleedalmutiry/EpiILM", "max_stars_repo_head_hexsha": "a22caac0e2f5af69fe202b2c750e237cbe6e295b", "max_stars_repo_licenses": ["Intel"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2020-06-30T03:50:37.000Z", "max_stars_repo_stars_event_max_datetime": "2022-02-21T21:15:25.000Z", "max_issues_repo_path": "R/epidata.r", "max_issues_repo_name": "waleedalmutiry/EpiILM", "max_issues_repo_head_hexsha": "a22caac0e2f5af69fe202b2c750e237cbe6e295b", "max_issues_repo_licenses": ["Intel"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/epidata.r", "max_forks_repo_name": "waleedalmutiry/EpiILM", "max_forks_repo_head_hexsha": "a22caac0e2f5af69fe202b2c750e237cbe6e295b", "max_forks_repo_licenses": ["Intel"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2021-02-01T20:50:30.000Z", "max_forks_repo_forks_event_max_datetime": "2021-02-18T17:44:20.000Z", "avg_line_length": 41.0212765957, "max_line_length": 184, "alphanum_fraction": 0.4984439834, "num_tokens": 2437, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7905303186696748, "lm_q2_score": 0.6370307806984445, "lm_q1q2_score": 0.5035921460679331}} {"text": "n <- 0\n\ncheck <- FALSE\nwhile (!check) {\n n <- n + 380\n check <- TRUE\n for (m in 19:1) {\n if (n %% m != 0) {\n check <- FALSE\n break\n }\n }\n}\n\ncat(n, \"\\n\")\n", "meta": {"hexsha": "0c5a6a74e7b395274660a9b125665b5c2c56f5fe", "size": 201, "ext": "r", "lang": "R", "max_stars_repo_path": "src/r/pe_0005.r", "max_stars_repo_name": "shanedrabing/project-euler", "max_stars_repo_head_hexsha": "65add91195fdde3c99c843743205be6d0b1fe072", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "src/r/pe_0005.r", "max_issues_repo_name": "shanedrabing/project-euler", "max_issues_repo_head_hexsha": "65add91195fdde3c99c843743205be6d0b1fe072", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "src/r/pe_0005.r", "max_forks_repo_name": "shanedrabing/project-euler", "max_forks_repo_head_hexsha": "65add91195fdde3c99c843743205be6d0b1fe072", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 12.5625, "max_line_length": 26, "alphanum_fraction": 0.3482587065, "num_tokens": 68, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7853085808877581, "lm_q2_score": 0.6406358411176238, "lm_q1q2_score": 0.5030968232539165}} {"text": "\"calibration\" <-\n function(obs, preds, family = \"binomial\")\n {\n #\n # j elith/j leathwick 17th March 2005\n # calculates calibration statistics for either binomial or count data\n # but the family argument must be specified for the latter\n # a conditional test for the latter will catch most failures to specify\n # the family\n #\n\n if (family == \"bernoulli\") family <- \"binomial\"\n pred.range <- max(preds) - min(preds)\n if(pred.range > 1.2 & family == \"binomial\") {\n print(paste(\"range of response variable is \", round(pred.range, 2)), sep = \"\", quote = F)\n print(\"check family specification\", quote = F)\n return()\n }\n if(family == \"binomial\") {\n pred <- preds + 1e-005\n pred[pred >= 1] <- 0.99999\n mod <- glm(obs ~ log((pred)/(1 - (pred))), family = binomial)\n lp <- log((pred)/(1 - (pred)))\n a0b1 <- glm(obs ~ offset(lp) - 1, family = binomial)\n miller1 <- 1 - pchisq(a0b1$deviance - mod$deviance, 2)\n ab1 <- glm(obs ~ offset(lp), family = binomial)\n miller2 <- 1 - pchisq(a0b1$deviance - ab1$deviance, 1)\n miller3 <- 1 - pchisq(ab1$deviance - mod$deviance, 1)\n }\n if(family == \"poisson\") {\n mod <- glm(obs ~ log(preds), family = poisson)\n lp <- log(preds)\n a0b1 <- glm(obs ~ offset(lp) - 1, family = poisson)\n miller1 <- 1 - pchisq(a0b1$deviance - mod$deviance, 2)\n ab1 <- glm(obs ~ offset(lp), family = poisson)\n miller2 <- 1 - pchisq(a0b1$deviance - ab1$deviance, 1)\n miller3 <- 1 - pchisq(ab1$deviance - mod$deviance, 1)\n }\n calibration.result <- c(mod$coef, miller1, miller2, miller3)\n names(calibration.result) <- c(\"intercept\", \"slope\", \"testa0b1\", \"testa0|b1\", \"testb1|a\")\n return(calibration.result)\n}\n\n\"calc.dev\" <-\n function(obs.values, fitted.values, weights = rep(1,length(obs.values)), family=\"binomial\", calc.mean = TRUE)\n {\n # j. leathwick/j. elith\n #\n # version 2.1 - 5th Sept 2005\n #\n # function to calculate deviance given two vectors of raw and fitted values\n # requires a family argument which is set to binomial by default\n #\n #\n\n if (length(obs.values) != length(fitted.values))\n stop(\"observations and predictions must be of equal length\")\n\n y_i <- obs.values\n\n u_i <- fitted.values\n\n if (family == \"binomial\" | family == \"bernoulli\") {\n\n deviance.contribs <- (y_i * log(u_i)) + ((1-y_i) * log(1 - u_i))\n deviance <- -2 * sum(deviance.contribs * weights)\n\n }\n\n if (family == \"poisson\" | family == \"Poisson\") {\n\n deviance.contribs <- ifelse(y_i == 0, 0, (y_i * log(y_i/u_i))) - (y_i - u_i)\n deviance <- 2 * sum(deviance.contribs * weights)\n\n }\n\n if (family == \"laplace\") {\n deviance <- sum(abs(y_i - u_i))\n }\n\n if (family == \"gaussian\") {\n deviance <- sum((y_i - u_i) * (y_i - u_i))\n }\n\n\n\n if (calc.mean) deviance <- deviance/length(obs.values)\n dev=list(deviance=deviance,dev.cont=deviance.contribs)\n return(dev)\n\n}\n\nbeachcolours<-function (heightrange, sealevel = 0, monochrome = FALSE, ncolours = if (monochrome) 16 else 64)\n{\n#this function was robbed from the spatstat library internals\n if (monochrome)\n return(grey(seq(0, 1, length = ncolours)))\n stopifnot(is.numeric(heightrange) && length(heightrange) ==\n 2)\n stopifnot(all(is.finite(heightrange)))\n depths <- heightrange[1]\n peaks <- heightrange[2]\n dv <- diff(heightrange)/(ncolours - 1)\n epsilon <- dv/2\n lowtide <- max(sealevel - epsilon, depths)\n hightide <- min(sealevel + epsilon, peaks)\n countbetween <- function(a, b, delta) {\n max(0, round((b - a)/delta))\n }\n nsea <- countbetween(depths, lowtide, dv)\n nbeach <- countbetween(lowtide, hightide, dv)\n nland <- countbetween(hightide, peaks, dv)\n colours <- character(0)\n if (nsea > 0)\n colours <- rev(rainbow(nsea, start = 3/6, end = 4/6))\n if (nbeach > 0)\n colours <- c(colours, rev(rainbow(nbeach, start = 3/12,\n end = 5/12)))\n if (nland > 0)\n colours <- c(colours, rev(rainbow(nland, start = 0, end = 1/6)))\n return(colours)\n}\n", "meta": {"hexsha": "7fb65fdfc92174e345a4742097437e0163f27e44", "size": 4323, "ext": "r", "lang": "R", "max_stars_repo_path": "contrib/sahm/pySAHM/Resources/R_Modules/EvalStatsHelperFcts.r", "max_stars_repo_name": "celiafish/VisTrails", "max_stars_repo_head_hexsha": "d8cb575b8b121941de190fe608003ad1427ef9f6", "max_stars_repo_licenses": ["BSD-3-Clause"], "max_stars_count": 83, "max_stars_repo_stars_event_min_datetime": "2015-01-05T14:50:50.000Z", "max_stars_repo_stars_event_max_datetime": "2021-09-17T19:45:26.000Z", "max_issues_repo_path": "contrib/sahm/pySAHM/Resources/R_Modules/EvalStatsHelperFcts.r", "max_issues_repo_name": "celiafish/VisTrails", "max_issues_repo_head_hexsha": "d8cb575b8b121941de190fe608003ad1427ef9f6", "max_issues_repo_licenses": ["BSD-3-Clause"], "max_issues_count": 254, "max_issues_repo_issues_event_min_datetime": "2015-01-02T20:39:19.000Z", "max_issues_repo_issues_event_max_datetime": "2018-11-28T17:16:44.000Z", "max_forks_repo_path": "contrib/sahm/pySAHM/Resources/R_Modules/EvalStatsHelperFcts.r", "max_forks_repo_name": "celiafish/VisTrails", "max_forks_repo_head_hexsha": "d8cb575b8b121941de190fe608003ad1427ef9f6", "max_forks_repo_licenses": ["BSD-3-Clause"], "max_forks_count": 40, "max_forks_repo_forks_event_min_datetime": "2015-04-17T16:46:36.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-28T22:43:24.000Z", "avg_line_length": 35.1463414634, "max_line_length": 117, "alphanum_fraction": 0.5750636132, "num_tokens": 1330, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7853085808877581, "lm_q2_score": 0.6406358411176238, "lm_q1q2_score": 0.5030968232539165}} {"text": "## 5장 소스코드\n\n# ab.test.imp 데이터에 ab.test.goal 데이터를 결합시키기\n# 데이터를 읽어들이기\nab.test.imp <- read.csv(\"section5-ab_test_imp.csv\",header=T, stringsAsFactors=F) # 배너광고의 표시횟수정보\nab.test.goal <- read.csv(\"section5-ab_test_goal.csv\",header=T, stringsAsFactors=F) # 배너광고의 클릭횟수정보\n# ab.test.imp에 ab.test.goal를 결합시키기\nab.test.imp <- merge(ab.test.imp, ab.test.goal, by=\"transaction_id\", all.x=T, suffixes=c(\"\",\".g\"))\nhead(ab.test.imp)\n\n# \n# 클릭 플래그를 추가\nab.test.imp$is.goal <- ifelse(is.na(ab.test.imp$user_id.g),0,1)\nhead(ab.test.imp)\n\n# \n# 클릭율을 계산하기\nlibrary(plyr)\nddply(ab.test.imp, .(test_case), summarize,\n cvr=sum(is.goal)/length(user_id))\n\n# χ2 검정을 실행하기\nchisq.test(ab.test.imp$test_case, ab.test.imp$is.goal)\n\n# \n# 날짜별, 테스트 케이스별로 클릭율을 산출하기\nab.test.imp.summary <-\n ddply(ab.test.imp, .(log_date, test_case), summarize,\n imp=length(user_id),\n cv=sum(is.goal),\n cvr=sum(is.goal)/length(user_id))\n# 테스트 케이스별로 클릭율을 산출하기\nab.test.imp.summary <-\n ddply(ab.test.imp.summary, .(test_case), transform,\n cvr.avg=sum(cv)/sum(imp))\nhead(ab.test.imp.summary)\n\n# 테스트 케이스별 클릭율의 시계열추이 그래프\nlibrary(ggplot2)\nlibrary(scales)\nab.test.imp.summary$log_date <- as.Date(ab.test.imp.summary$log_date)\nlimits <- c(0, max(ab.test.imp.summary$cvr))\nggplot(ab.test.imp.summary,aes(x=log_date,y=cvr, col=test_case,lty=test_case, shape=test_case)) +\n geom_line(lwd=1) +\n geom_point(size=4) +\n geom_line(aes(y=cvr.avg,col=test_case)) +\n scale_y_continuous(label=percent, limits=limits)", "meta": {"hexsha": "a3cfa6d698a471dc48610b5883035a9c62e3d7f4", "size": 1476, "ext": "r", "lang": "R", "max_stars_repo_path": "ch5.r", "max_stars_repo_name": "yedam-Lee/myfirstrepo", "max_stars_repo_head_hexsha": "39a96f0e6702b4b821cb56ab3ec49a83abe5fb19", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2019-03-13T02:26:29.000Z", "max_stars_repo_stars_event_max_datetime": "2019-03-13T02:26:29.000Z", "max_issues_repo_path": "ch5.r", "max_issues_repo_name": "yedam-Lee/myfirstrepo", "max_issues_repo_head_hexsha": "39a96f0e6702b4b821cb56ab3ec49a83abe5fb19", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ch5.r", "max_forks_repo_name": "yedam-Lee/myfirstrepo", "max_forks_repo_head_hexsha": "39a96f0e6702b4b821cb56ab3ec49a83abe5fb19", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 31.4042553191, "max_line_length": 98, "alphanum_fraction": 0.6991869919, "num_tokens": 567, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8128673087708699, "lm_q2_score": 0.6187804337438501, "lm_q1q2_score": 0.5029863858974349}} {"text": "##################################################\n#### RNAseq data SIMULATION ####\n##################################################\n\n\n# By Sonia Tarazona\n# Created: 4-oct-2012\n\n\n\n### Function to simulate RNA-seq data\n\nsimula.counts = function (counts0 = NULL, ngenes = 20000, nrepl = 5, \n depth = 30, propdeg = 0.05, noise = 0.1, beta = 6) {\n \n # counts0: Counts per gene used to derive counts in condition 1. \n # If NULL, random numbers are generated from a power-law distribution.\n # ngenes: Number of genes in the dataset when counts0 is NULL.\n # nrepl: Number of replicates per condition.\n # depth: Minimum estimation for the total number of counts (in millions) for each sample\n # propdeg: Proportion of genes expected to be differentially expressed.\n # noise: Variability for mu_0 between conditions.\n # beta: Beta distribution parameter (shape2)\n \n \n depth = depth*10^6\n ncond = 2\n\n \n if (length(nrepl) == 1) {\n nrepl = rep(nrepl, ncond)\n }\n \n \n if (!is.null(counts0)) { # counts0 are given\n ngenes = length(counts0)\n \n if (is.null(names(counts0))) {\n mygenes = paste(\"g\", formatC(1:ngenes, width = nchar(ngenes), \n format = \"d\", flag = \"0\"), sep =\"\")\n } else {\n mygenes = names(counts0)\n }\n \n } else { # counts0 are randomly generated\n potencia = 0.5\n x = 1:round(depth/1000,0)\n myprob = x^(-potencia) \n counts0 = sample(x, size = ngenes, replace = TRUE, prob = myprob)\n \n mygenes = paste(\"g\", formatC(1:ngenes, width = nchar(ngenes), \n format = \"d\", flag = \"0\"), sep =\"\")\n }\n \n names(counts0) = mygenes\n \n # DEG\n ndeg = round(propdeg*ngenes, 0) # number of DEG\n \n deg = sample(mygenes, ndeg) # names of DEG\n \n p.up = runif(1, 0.25, 0.75) # percentage of DEG up-regulated\n \n deg1 = sample(deg, round(p.up*ndeg,0)) # names of DEG up-regulated\n \n deg2 = setdiff(deg, deg1) # names of DEG down-regulated\n \n k1 = k2 = 5 # to avoid multiply foldchange by 0 or very low counts\n\n\n forma1 = 1.5\n forma2 = beta\n \n \n # DEG-up \n cond1.change = rbeta(length(deg1), shape1 = forma1, shape2 = forma2)*100 + 1.5\n \n counts1 = counts0\n counts1[deg1] = (counts1[deg1] + k1) * cond1.change\n \n \n # DEG-down\n cond2.change = rbeta(length(deg2), shape1 = forma1, shape2 = forma2)*100 + 1.5\n \n counts2 = counts0\n counts2[deg2] = (counts2[deg2] + k2) * cond2.change\n \n \n \n # Adjusting counts for the desired depth \n counts1 = round(depth * counts1/sum(counts1),0)\n names(counts1) = mygenes\n \n counts2 = round(depth * counts2/sum(counts2),0)\n names(counts2) = mygenes\n \n\n \n # Condition 1 \n cond1 = gener.cond(counts = counts1, noise = noise, nrepl = nrepl[1])\n colnames(cond1) = paste(\"cond1\", 1:nrepl[1], sep = \".\")\n rownames(cond1) = mygenes\n \n \n # Condition 2 \n cond2 = gener.cond(counts = counts2, noise = noise, nrepl = nrepl[2])\n colnames(cond2) = paste(\"cond\", 2, \".\", 1:nrepl[2], sep = \"\")\n rownames(cond2) = mygenes\n\n \n \n # Results\n \n if (propdeg > 0) {\n \n changes = data.frame(\"gene\" = deg1, \"regu\" = 1, \"foldchange\" = cond1.change)\n changes = rbind(changes,\n data.frame(\"gene\" = deg2, \"regu\" = 2, \"foldchange\" = cond2.change))\n changes$gene = as.character(changes$gene)\n \n } else { changes = NULL } \n \n \n list(\"simucounts\" = cbind(cond1,cond2), \"change\" = changes, \n \"changedepth\" = sum(counts1)-sum(counts2))\n \n \n}\n\n\n\n\n\n##-------------------------------------------------------------------##\n##-------------------------------------------------------------------##\n##-------------------------------------------------------------------##\n##-------------------------------------------------------------------##\n\n\n\n### Function phi(mu) obtained from real datasets\n\n# (mu,phi) taken from real data sets + interpolation function\nlasmus = c(0.388888888888889,0.708333333333333,0.881038647342995,1.03381642512077,1.19444444444444,\n 1.3780193236715,1.58913043478261,1.85,2.15217391304348,2.4993961352657,2.90277777777778,\n 3.36111111111111,3.90485185185186,4.55581803542673,5.31763285024155,6.16111111111111,\n 7.12608695652174,8.26358695652174,9.64583333333333,11.2518115942029,13.0646376811594,\n 15.1733333333333,17.6209239130435,20.4375,23.6354166666667,27.3089814814815,\n 31.3814452495974,35.8328804347826,40.8017451690821,46.3384299516908,52.6304347826087,\n 59.5,66.6265217391304,74.3904106280193,82.9713768115942,92.4259118357488,\n 102.620053542673,113.546009661836,125.34107568438,137.798917874396,151.36309178744,\n 165.45229468599,180.591425120773,196.93825,213.954101449275,232.221954508857,\n 251.616769726248,272.179304347826,293.808748792271,317.25170531401,342.512927536232,\n 369.294992753623,397.492036231884,428.298376811594,462.24537037037,498.544742351047,\n 538.499621980676,580.50025,625.7,675.786956521739,732.746376811594,798.775333333334,\n 874.794357487924,959.35825925926,1053.70944444444,1163.35739774557,1294.77174879227,\n 1456.64353623188,1660.2097294686,1919.08176811594,2264.31477938808,2813.65251851852,\n 3755.676,5910.27933333333,369699.430057971)\n\nlasphi = c(1.4921875,2.1625,2.65,2.4703125,2.4375,2.1375,1.94375,1.7828125,1.5890625,1.45078125,\n 1.2921875,1.203515625,1.1140625,1.0265625,0.923046875000001,0.7978515625,0.7607421875,\n 0.718164062500001,0.625,0.5740234375,0.5375,0.5091796875,0.486328125,0.458203125,\n 0.457421875,0.38359375,0.384375,0.3875,0.36318359375,0.33466796875,0.333203125,\n 0.323828125,0.3205078125,0.2845703125,0.27119140625,0.2830078125,0.2716796875,\n 0.265234375,0.24609375,0.2521484375,0.2572265625,0.244921875,0.22734375,0.22275390625,\n 0.240625,0.210546875,0.20947265625,0.2203125,0.228125,0.21796875,0.1970703125,0.1859375,\n 0.18828125,0.1763671875,0.178173828125,0.175390625,0.1888671875,0.178125,0.1591796875,\n 0.169140625,0.1595703125,0.158203125,0.149609375,0.1580078125,0.15556640625,0.14453125,\n 0.1390625,0.14453125,0.13046875,0.139453125,0.1353515625,0.140625,0.15078125,\n 0.1775390625,0.2140625)\n\n\nphimean = approxfun(lasmus, lasphi, rule = 2)\n\n\n\n\n##-------------------------------------------------------------------##\n\n\n\n\n### Function to generate the replicates for a given condition from Binomial distribution + noise\n\ngener.cond = function (counts, noise, nrepl) {\n \n # Allowing for some noise in the NB mean\n counts.noise = t(sapply(counts, function(x) { x + c(-1,1)*noise*x }))\n counts.noise[which(counts.noise < 0)] = 0\n \n mu.noise = apply(counts.noise, 1, function (x) { runif(1, x[1], x[2]) })\n #mu.noise = counts + runif(length(counts), -noise, noise)\n #mu.noise = sapply(mu.noise, function(x) { max(x,0) })\n \n # Computing phi\n phi.mean = phimean(mu.noise)\n phi.sd = 1 / (1 + mu.noise^0.25)\n \n phi = apply(cbind(phi.mean, phi.sd), 1, function (x) { max(0.000001,rnorm(1, x[1], x[2])) } )\n \n # Results\n t(apply(cbind(mu.noise, phi), 1, function (x) { rnbinom(n = nrepl, size = 1/x[2], \n mu = max(x[1], 0.1)) })) \n} \n\n\n\n\n\n##-------------------------------------------------------------------##\n\n\n\n", "meta": {"hexsha": "b149c65f81efe4e14706684f77aa575194dfa326", "size": 7417, "ext": "r", "lang": "R", "max_stars_repo_path": "functions/simulation_high.r", "max_stars_repo_name": "mdozmorov/deconvolution", "max_stars_repo_head_hexsha": "224ab3c47b5b2e59984471ca098614f644849cba", "max_stars_repo_licenses": ["Artistic-2.0"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2018-10-08T11:37:33.000Z", "max_stars_repo_stars_event_max_datetime": "2018-10-08T11:37:33.000Z", "max_issues_repo_path": "functions/simulation_high.r", "max_issues_repo_name": "mdozmorov/deconvolution", "max_issues_repo_head_hexsha": "224ab3c47b5b2e59984471ca098614f644849cba", "max_issues_repo_licenses": ["Artistic-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "functions/simulation_high.r", "max_forks_repo_name": "mdozmorov/deconvolution", "max_forks_repo_head_hexsha": "224ab3c47b5b2e59984471ca098614f644849cba", "max_forks_repo_licenses": ["Artistic-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 34.6588785047, "max_line_length": 99, "alphanum_fraction": 0.5975461777, "num_tokens": 2565, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7401743505760728, "lm_q2_score": 0.679178699175393, "lm_q1q2_score": 0.5027106525872485}} {"text": "s <- sapply (1:10,\n function (x) {\n x # force evaluation of promise x\n\tfunction () {\n R <- x*x\n # evaluate the language expression \"x <- x + 1\" in the persistent parent environment\n evalq (x <- x + 1, parent.env(environment()))\n R # return squared value\n }})\n\ns[[5]]()\n[1] 25 # 5^2\ns[[5]]()\n[1] 36 # now 6^2\ns[[1]]()\n[1] 1 # 1^2\ns[[1]]()\n[1] 4 # now 2^2\n", "meta": {"hexsha": "82a9f56efd06a383bbfbdc9101ab238db82afe5b", "size": 431, "ext": "r", "lang": "R", "max_stars_repo_path": "Task/Closures-Value-capture/R/closures-value-capture-3.r", "max_stars_repo_name": "LaudateCorpus1/RosettaCodeData", "max_stars_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_stars_repo_licenses": ["Info-ZIP"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2018-11-09T22:08:38.000Z", "max_stars_repo_stars_event_max_datetime": "2018-11-09T22:08:38.000Z", "max_issues_repo_path": "Task/Closures-Value-capture/R/closures-value-capture-3.r", "max_issues_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_issues_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_issues_repo_licenses": ["Info-ZIP"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "Task/Closures-Value-capture/R/closures-value-capture-3.r", "max_forks_repo_name": "seanwallawalla-forks/RosettaCodeData", "max_forks_repo_head_hexsha": "9ad63ea473a958506c041077f1d810c0c7c8c18d", "max_forks_repo_licenses": ["Info-ZIP"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2018-11-09T22:08:40.000Z", "max_forks_repo_forks_event_max_datetime": "2018-11-09T22:08:40.000Z", "avg_line_length": 22.6842105263, "max_line_length": 96, "alphanum_fraction": 0.4617169374, "num_tokens": 150, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7122321720225278, "lm_q2_score": 0.7057850216484838, "lm_q1q2_score": 0.5026827989496664}} {"text": "f <- 100\nxs <- seq(0, f-1) \namin <- -128\namax <- 127\numax <- amax - amin\n\ngen_square <- function(t, f) {\n if (t < f/2) {\n amax\n } else {\n amin\n }\n}\n\ngen_saw <- function(t, f) {\n if (f == 0) {\n 0\n } else {\n t * umax / f + amin\n }\n}\n\ngen_tri <- function(t, f) {\n if (f==0) {\n 0\n } else {\n f2 <- f/2\n if (t < f2) {\n t * umax / f2 + amin\n } else {\n amax - ((t-f2)*umax/f2)\n \n \n }\n }\n}\n\ngen_sin <- function(t, f) {\n if (f==0) {\n 0\n } else {\n amax * sin(2*pi * t / f)\n }\n}\n\nsquare <- sapply(xs, FUN=gen_square, f)\n\nsaw <- sapply(xs, FUN=gen_saw, f)\n\ntri <- sapply(xs, FUN=gen_tri, f)\n\ns <- sapply(xs, FUN=gen_sin, f)\n\nplot(xs, square, type=\"o\")\nplot(xs, saw, type=\"o\")\nplot(xs, tri, type=\"o\")\nplot(xs, s, type=\"o\")\n", "meta": {"hexsha": "a03f3528f5b44fba4ea6978857e3b2b81dad903c", "size": 774, "ext": "r", "lang": "R", "max_stars_repo_path": "wave_generators.r", "max_stars_repo_name": "Raffaello/sdl2-sonic-drivers", "max_stars_repo_head_hexsha": "20584f100ddd7c61f584deaee0b46c5228d8509d", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2021-10-31T14:24:00.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-16T08:15:31.000Z", "max_issues_repo_path": "wave_generators.r", "max_issues_repo_name": "Raffaello/sdl2-sonic-drivers", "max_issues_repo_head_hexsha": "20584f100ddd7c61f584deaee0b46c5228d8509d", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": 48, "max_issues_repo_issues_event_min_datetime": "2020-06-05T11:11:29.000Z", "max_issues_repo_issues_event_max_datetime": "2022-02-27T23:58:44.000Z", "max_forks_repo_path": "wave_generators.r", "max_forks_repo_name": "Raffaello/sdl2-sonic-drivers", "max_forks_repo_head_hexsha": "20584f100ddd7c61f584deaee0b46c5228d8509d", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 13.3448275862, "max_line_length": 39, "alphanum_fraction": 0.4664082687, "num_tokens": 319, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8221891305219504, "lm_q2_score": 0.6113819732941511, "lm_q1q2_score": 0.5026716130395125}} {"text": "#' Y-structure collider\n#'\n#' X and Y are continuous, they influence continuous S which predicts binary inclusion variable Z\n#'\n#' @param n Population sample size\n#' @param prop Proportion of population in sample\n#' @param or_sz Odds ratio for the effect of collider variable S on being included\n#' @param b_xy Causal effect of X on Y\n#' @param rsq_xs Variance explained by X on collider variable S\n#' @param rsq_ys Variance explained by Y on collider variable S\n#' @param vx Variance of X. Default = `1`.\n#' @param vy Variance of Y. Default = `1`.\n#' @param vs Variance of S. Default = `1`.\n#' @param sig.level Alpha threshold for power calculation. Default = `5e-8`\n#'\n#' @export\n#' @importFrom stats pnorm power.anova.test\n#' @return List of simulation results\nsimulate_y_structure <- function(n, prop, or_sz, b_xy, rsq_xs, rsq_ys, vx=1, vy=1, vs=1, sig.level=5e-8)\n{\n # infer\n logor_sz <- log(or_sz)\n vz <- prop * (1-prop)\n b_xs <- sqrt(abs(rsq_xs) * vs / (vx)) * sign(rsq_xs)\n b_ys <- sqrt(abs(rsq_ys) * vs / vy) * sign(rsq_ys)\n ve <- vs - b_xs^2 * vx - b_ys^2 * vy\n alpha <- log((1 - prop)/prop)\n bh_xy <- (b_xy * vz * vx - b_xs * logor_sz * vz * b_ys * logor_sz * vz * vx * vy) / ((b_ys * logor_sz * vz)^2 * vx * vy + vx*vz)\n cor_xy <- sqrt(bh_xy^2 * vx / vy) * sign(bh_xy)\n if(is.na(n))\n {\n return(list(bh_xy=bh_xy, cor_xy=cor_xy, pow=NA, se=NA, pval=NA, ns=NA))\n }\n ns <- n * prop\n if(ns == 0)\n {\n pow <- 0\n } else {\n pow <- power.anova.test(groups=2, n=ns, between.var=cor_xy^2, within.var = 1 - cor_xy^2, sig.level= sig.level)$power\n }\n se <- sqrt((1 - cor_xy^2) * vs / (ns * vx))\n pval <- pnorm(abs(bh_xy / se), lower.tail=FALSE)\n return(list(bh_xy=bh_xy, cor_xy=cor_xy, pow=pow, se=se, pval=pval, ns=ns))\n}\n\n#' Plot Y structure bias across range of parameters\n#'\n#' Takes a range of OR_sz values, and ranges of rsq_ys and rsq_xs values. Then, for a given proportion of the population being sampled, provides the expected X-Y association for each of the parameter combinations. Enumerates over all combinations so can be slow \n#'\n#' @param prop Proportion of population included in sample\n#' @param b_xy_thresh Target b_xy association - e.g. what value of b_xy are you suspicious could be due to ascertainment on a collider\n#' @param or_sz_range Range of OR_sz values to enumerate over. Default = `c(1,100)`\n#' @param b_xy Suspected true X-Y effect. Default = `0`\n#' @param rsq_xs_range Range of rsq_xs values to enumerate over. Default = `c(0,1)`\n#' @param rsq_ys_range Range of rsq_ys values to enumerate over. Default = `c(0,1)`\n#' @param gran Granularity of ranges. Default = `30`\n#'\n#' @export\n#' @return ggplot of simulations\nplot_simulate_y_structure <- function(prop, b_xy_thresh, or_sz_range=c(1,100), b_xy=0, rsq_xs_range=c(0,1), rsq_ys_range=c(0,1), gran=30)\n{\n message(\"Calculating surface\")\n param <- expand.grid(\n or_sz = exp(seq(log(or_sz_range[1]), log(or_sz_range[2]), length.out=gran)),\n rsq_xs=seq(rsq_xs_range[1], rsq_xs_range[2], length.out=gran),\n rsq_ys=seq(rsq_ys_range[1], rsq_ys_range[2], length.out=gran),\n bh_xy=NA\n ) %>% dplyr::filter(rsq_xs + rsq_ys <=1)\n for(i in 1:nrow(param))\n {\n o <- simulate_y_structure(NA, prop, param$or_sz[i], b_xy, param$rsq_xs[i], param$rsq_ys[i])\n param$bh_xy[i] <- o$bh_xy\n }\n param <- dplyr::mutate(param, thresh = abs(bh_xy) >= abs(b_xy_thresh) & sign(bh_xy) == sign(b_xy_thresh))\n param2 <- param %>% dplyr::filter(thresh) %>% \n dplyr::group_by(rsq_xs, rsq_ys) %>%\n dplyr::summarise(or_sz = min(or_sz))\n\n message(\"Plotting\")\n g <- ggplot2::ggplot(param, ggplot2::aes(x=rsq_xs, y=rsq_ys)) +\n ggplot2::geom_point(colour=\"grey\", size=0.2) +\n ggplot2::geom_point(data=param2, ggplot2::aes(colour=or_sz), size=2) +\n ggplot2::scale_colour_distiller(palette = \"Spectral\") +\n ggplot2::labs(x=\"Variance in S explained by X\", y=\"Variance in S explained by Y\", colour=\"OR (S->Z)\")\n g\n}\n\n#' Plot the minimum Y-structure required to explain some association\n#'\n#' For a given Y-structure, can an association of b_xy can be obtained through ascertainment on a collider? Use optimisation to determine the minimum OR_sz for a range of rsq_xs and rsq_ys values\n#'\n#' @param prop Proportion of population included in sample\n#' @param b_xy_thresh Target b_xy association - e.g. what value of b_xy are you suspicious could be due to ascertainment on a collider\n#' @param b_xy Suspected true X-Y effect. Default = `0`\n#' @param rsq_xs_range Range of rsq_xs values to enumerate over. Default = `c(0,1)`\n#' @param rsq_ys_range Range of rsq_ys values to enumerate over. Default = `c(0,1)`\n#' @param gran Granularity of ranges. Default = `101`\n#' @param max_or_sz Maximum OR_sz to allow in optimisation.\n#'\n#' @export\n#' @importFrom stats optimize\n#' @return ggplot of simulations\nplot_simulate_y_structure_optim <- function(prop, b_xy_thresh, b_xy=0, rsq_xs_range=c(0,1), rsq_ys_range=c(0,1), gran=101, max_or_sz=20)\n{\n message(\"Calculating surface\")\n param <- expand.grid(\n rsq_xs=seq(rsq_xs_range[1], rsq_xs_range[2], length.out=gran),\n rsq_ys=seq(rsq_ys_range[1], rsq_ys_range[2], length.out=gran),\n bh_xy=NA,\n or_sz=NA\n ) %>% dplyr::filter(rsq_xs + rsq_ys <=1)\n param$id <- 1:nrow(param)\n fn <- function(x, prop, b_xy, rsq_xs, rsq_ys, b_xy_thresh) {\n o <- simulate_y_structure(NA, prop, x, b_xy, rsq_xs, rsq_ys)\n return((o$bh_xy - b_xy_thresh)^2)\n }\n for(i in 1:nrow(param))\n {\n o <- optimize(fn, c(1,max_or_sz+1), prop=prop, b_xy=b_xy, rsq_xs=param$rsq_xs[i], rsq_ys=param$rsq_ys[i], b_xy_thresh=b_xy_thresh)\n param$bh_xy[i] <- b_xy_thresh\n param$or_sz[i] <- o$minimum\n }\n param$thresh <- param$or_sz <= max_or_sz\n \n message(\"Plotting\")\n g <- ggplot2::ggplot(param %>% subset(., !thresh), ggplot2::aes(x=rsq_xs, y=rsq_ys)) +\n ggplot2::geom_point(colour=\"grey\", size=0.2) +\n ggplot2::geom_point(data=param %>% subset(., thresh), ggplot2::aes(colour=or_sz), size=2) +\n ggplot2::scale_colour_distiller(palette = \"Spectral\") +\n ggplot2::labs(x=\"Variance in S explained by X\", y=\"Variance in S explained by Y\", colour=\"OR of S\\non inclusion\")\n g\n}\n", "meta": {"hexsha": "1a0ac1fc7ea4ee30b9e417b711666a0c6aceb010", "size": 6160, "ext": "r", "lang": "R", "max_stars_repo_path": "R/y-structure.r", "max_stars_repo_name": "explodecomputer/collidR", "max_stars_repo_head_hexsha": "d7f7a6dbd2f72e51d17fbed7dceb148b809acf93", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "R/y-structure.r", "max_issues_repo_name": "explodecomputer/collidR", "max_issues_repo_head_hexsha": "d7f7a6dbd2f72e51d17fbed7dceb148b809acf93", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 1, "max_issues_repo_issues_event_min_datetime": "2020-04-21T06:40:54.000Z", "max_issues_repo_issues_event_max_datetime": "2020-04-21T06:40:54.000Z", "max_forks_repo_path": "R/y-structure.r", "max_forks_repo_name": "explodecomputer/collidR", "max_forks_repo_head_hexsha": "d7f7a6dbd2f72e51d17fbed7dceb148b809acf93", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2020-04-15T13:20:56.000Z", "max_forks_repo_forks_event_max_datetime": "2020-04-20T20:10:23.000Z", "avg_line_length": 46.3157894737, "max_line_length": 262, "alphanum_fraction": 0.6827922078, "num_tokens": 2013, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.779992900254107, "lm_q2_score": 0.6442251064863697, "lm_q1q2_score": 0.5024910092248144}} {"text": "model_cropheatflux <- function (netRadiationEquivalentEvaporation = 638.142,\n soilHeatFlux = 188.817,\n potentialTranspiration = 1.413){\n #'- Name: CropHeatFlux -Version: 1.0, -Time step: 1\n #'- Description:\n #' * Title: CropHeatFlux Model\n #' * Author: Pierre Martre\n #' * Reference: abModelling energy balance in the wheat crop model SiriusQuality2:\n #' Evapotranspiration and canopy and soil temperature calculations\n #' * Institution: INRA/LEPSE Montpellier\n #' * Abstract: It is calculated from net Radiation, soil heat flux and potential transpiration \n #'- inputs:\n #' * name: netRadiationEquivalentEvaporation\n #' ** variablecategory : auxiliary\n #' ** description : net Radiation Equivalent Evaporation\n #' ** datatype : DOUBLE\n #' ** default : 638.142\n #' ** min : 0\n #' ** max : 10000\n #' ** unit : g m-2 d-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' * name: soilHeatFlux\n #' ** description : soil Heat Flux\n #' ** variablecategory : rate\n #' ** datatype : DOUBLE\n #' ** default : 188.817\n #' ** min : 0\n #' ** max : 1000\n #' ** unit : g m-2 d-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #' * name: potentialTranspiration\n #' ** description : potential Transpiration\n #' ** variablecategory : rate\n #' ** datatype : DOUBLE\n #' ** default : 1.413\n #' ** min : 0\n #' ** max : 1000\n #' ** unit : g m-2 d-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n #' ** inputtype : variable\n #'- outputs:\n #' * name: cropHeatFlux\n #' ** description : crop Heat Flux\n #' ** variablecategory : rate\n #' ** datatype : DOUBLE\n #' ** min : 0\n #' ** max : 10000\n #' ** unit : g m-2 d-1\n #' ** uri : http://www1.clermont.inra.fr/siriusquality/?page_id=547\n cropHeatFlux <- netRadiationEquivalentEvaporation - soilHeatFlux - potentialTranspiration\n return (list('cropHeatFlux' = cropHeatFlux))\n}", "meta": {"hexsha": "c385140a34f6b20a90addfec0d95cde0f2c28ba1", "size": 3049, "ext": "r", "lang": "R", "max_stars_repo_path": "test/Models/energybalance_pkg/src/r/Cropheatflux.r", "max_stars_repo_name": "brichet/PyCrop2ML", "max_stars_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2020-06-21T18:58:04.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-29T21:32:28.000Z", "max_issues_repo_path": "test/Models/energybalance_pkg/src/r/Cropheatflux.r", "max_issues_repo_name": "brichet/PyCrop2ML", "max_issues_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 27, "max_issues_repo_issues_event_min_datetime": "2018-12-04T15:35:44.000Z", "max_issues_repo_issues_event_max_datetime": "2022-03-11T08:25:03.000Z", "max_forks_repo_path": "test/Models/energybalance_pkg/src/r/Cropheatflux.r", "max_forks_repo_name": "brichet/PyCrop2ML", "max_forks_repo_head_hexsha": "7177996f72a8d95fdbabb772a16f1fd87b1d033e", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 7, "max_forks_repo_forks_event_min_datetime": "2019-01-15T04:33:23.000Z", "max_forks_repo_forks_event_max_datetime": "2021-12-09T07:29:46.000Z", "avg_line_length": 56.462962963, "max_line_length": 110, "alphanum_fraction": 0.4165300098, "num_tokens": 663, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929002541068, "lm_q2_score": 0.6442251064863697, "lm_q1q2_score": 0.5024910092248143}} {"text": "#Calculate SNP-index\n\n#' @title Calculate SNP-index\n#' @description This function calculates, for both wild-type and mutant bulks, the SNP-index value of each variant stored inside the data frame (returned by the readBSA_vcf() function), \n#' applying the following formula:\n#' \\deqn{SNPindex=AD_alt/(AD_ref + AD_alt)} \n#' Bulks get joined together in a single data frame.\n#' \n#' Note that the user can select specific variants to consider, \n#' by setting the \"variants\" parameter to \"SNP\" (default) or \"all\" (InDels+SNPs).\n#'\n#' @param vcf.df Data frame of the vcf file\n#' @param wtBulk Wild-Type pool\n#' @param mBulk Mutant pool \n#' @param variants variants to be considered. Default is \"SNP\" (allowed: \"SNP\" or \"all\")\n#'\n#' @return Data frame containing variand and SNP-index information for each bulk.\n#'\n#' @details The data frame returned by readBSA_vcf() is filtered by bulk ID to create two separate data frames: \n#' one specific to the wild-type bulk variants information and other one specific to the mutant bulk variants information.\n#' For every variant in each data frame, the SNP-index value gets calculated and added as a new column.\n#' \n#' In the instance that the type of variants to be considered is set to \"SNP\",the variants corresponding to InDel sare discarded. \n#' The rows containing more than three characters in the GT_alleles column (e.g.,\"T/AT\"corresponds to an insertion and contains 4 characters) \n#' or containing three characters but one of them being \"*\", meaning a deletion, are removed from the data frame. \n#' \n#' Both dataframes are then joined by same chromosome and position to ensure that all the merged rows contain information on the same genomic position.\n#' This data frame is returned by the function\n#'\n#' @export calc_SNPindex\n#' @examples\n#' ## Calculate SNP-index for both bulks (only SNPs will be considered)\n#' vcf_df_SNPindex <- calc_SNPindex(vcf.df=vcf_list$df, \n#' wtBulk=\"pool_S3781_minus\", \n#' mBulk=\"pool_S3781_plus\") \n#'\n#' ## Calculate SNP-index considering both InDels and SNPs\n#' vcf_df_SNPindex <- calc_SNPindex(vcf.df=vcf_list$df, \n#' wtBulk=\"pool_S3781_minus\", \n#' mBulk=\"pool_S3781_plus\",\n#' variants=\"all\") \n\n\ncalc_SNPindex <- function(vcf.df, wtBulk, mBulk, variants=\"SNP\") {\n \n #Create data frame for each bulk AND include a new column with SNP index\n vcf.df.wtBulk <- vcf.df %>% dplyr::filter(Indiv==wtBulk) %>% dplyr::mutate(\"SNPindex\"= as.numeric(AD_alt)/(as.numeric(AD_ref) + as.numeric(AD_alt)))\n vcf.df.mBulk <- vcf.df %>% dplyr::filter(Indiv==mBulk) %>% dplyr::mutate(\"SNPindex\"= as.numeric(AD_alt)/(as.numeric(AD_ref) + as.numeric(AD_alt)))\n \n #Stop the program and show message if user selects a non-allowed 'variants' value\n if (variants!=\"SNP\" & variants!=\"all\") {\n stop(\"The allowed values for the 'variants' argument are: 'SNP' or 'all'. The latter will consider both InDels and SNPs.\")\n }\n \n #Remove from data frames those rows corresponding to InDel variants (if variants==\"SNP\")\n if (variants==\"SNP\") {\n vcf.df.wtBulk <- vcf.df.wtBulk %>% dplyr::filter(nchar(GT_alleles)==3 &\n grepl(\"[^*]{3}\", GT_alleles))\n vcf.df.mBulk <- vcf.df.mBulk %>% dplyr::filter(nchar(GT_alleles)==3 &\n grepl(\"[^*]{3}\", GT_alleles))\n }\n \n #Join data frames of each bulk by same chromosome (ChromKey) and position\n vcf.df.SNPindex <- dplyr::inner_join(vcf.df.wtBulk, \n vcf.df.mBulk, \n by = c(\"ChromKey\",\"POS\"), \n copy = F, \n suffix = c(\".WT\", \".M\"))\n \n return(vcf.df.SNPindex)\n}\n\n", "meta": {"hexsha": "245dfc11d28f5b37f6115bd79a61379b6859f052", "size": 3883, "ext": "r", "lang": "R", "max_stars_repo_path": "BSAvis/R/calc_SNPIndex.r", "max_stars_repo_name": "FadyMohareb/BSAvis_GP_2020", "max_stars_repo_head_hexsha": "6ce28be7250c0cc117b5d8ccb23ac02bb34a7d64", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2021-04-28T10:50:00.000Z", "max_stars_repo_stars_event_max_datetime": "2022-03-26T13:43:06.000Z", "max_issues_repo_path": "R/calc_SNPIndex.r", "max_issues_repo_name": "EG-lisy/BSAvis", "max_issues_repo_head_hexsha": "bfc3ea0612bab439f8ab956e179e4bcd88678bee", "max_issues_repo_licenses": ["Unlicense"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/calc_SNPIndex.r", "max_forks_repo_name": "EG-lisy/BSAvis", "max_forks_repo_head_hexsha": "bfc3ea0612bab439f8ab956e179e4bcd88678bee", "max_forks_repo_licenses": ["Unlicense"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2021-05-03T23:24:03.000Z", "max_forks_repo_forks_event_max_datetime": "2022-03-31T15:47:15.000Z", "avg_line_length": 53.1917808219, "max_line_length": 186, "alphanum_fraction": 0.6381663662, "num_tokens": 1001, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7799929002541068, "lm_q2_score": 0.6442250928250375, "lm_q1q2_score": 0.5024909985690722}} {"text": "#' @title calcPropDia\n#'\n#' @description Calculate propeller diameter (\\code{propDiam}) (m).\n#'\n#'@param shipType Ship type (vector of strings, see \\code{\\link{calcShipType}}).\n#' Must align with \\code{tankerBulkCarrierGCargoShipTypes} and\n#' \\code{containerShipTypes} groupings\n#' @param maxDraft Maximum summer load line draft (vector of numericals, m)\n#' @param tankerBulkCarrierGCargoShipTypes Ship types specified in input\n#' \\code{shipTypes} to be modeled as tankers, bulk carriers and general cargo\n#' vessels (vector of strings)\n#' @param containerShipTypes Ship types specified in input \\code{shipTypes} to\n#' be modeled as container ships (vector of strings)\n#'\n#'@details\n#'This method this requires ship types to be grouped. Use the\n#' \\code{tankerBulkCarrierGCargoShipTypes} and \\code{containerShipTypes} grouping\n#' parameters to provide these ship type groupings. Any ship types not included\n#' in these groupings will be considered as miscellaneous vessels.\n#'\n#' @return \\code{propDiam} (vector of numericals, m)\n#'\n#' @references\n#'Kristensen, H. O. and Lutzen, M. 2013. \"Prediction of Resistance and Propulsion\n#'Power of Ships.\"\n#'\n#'\\href{https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}{Kristensen, H. O.\n#'\"Ship-Desmo-Tool.\" https://gitlab.gbar.dtu.dk/oceanwave3d/Ship-Desmo}\n#'\n#' @examples\n#' calcPropDia(c(\"bulk.carrier\",\"container.ship\"), c(13.6,15.6))\n#' calcPropDia(c(\"other.tanker\",\"container.ship\"), c(13.6,15.6),\n#' tankerBulkCarrierGCargoShipTypes=c(\"other.tanker\",\"bulk.carrier\"))\n#'\n#' @export\n\ncalcPropDia <-function(shipType, maxDraft,\n tankerBulkCarrierGCargoShipTypes=c(\"general.cargo\",\"tanker\",\"chemical.tanker\",\"liquified.gas.tanker\",\"oil.tanker\",\"other.tanker\",\"bulk.carrier\"),\n containerShipTypes=c(\"container.ship\")\n ){\n\n\n propDiam<-ifelse( #case 1\n shipType %in% tankerBulkCarrierGCargoShipTypes,\n #if true return:\n 0.395*maxDraft+1.3,\n #otherwise:\n ifelse( #case 2\n shipType%in%containerShipTypes,\n #if true return:\n 0.623*maxDraft-0.16,\n #otherwise: (RoRo etc...)\n 0.713*maxDraft-0.08\n )#end case 2\n )#end case 1\n\n return(propDiam)\n}\n", "meta": {"hexsha": "2ab2ed428fff5e1b60fbf764bed592f30e11fb48", "size": 2207, "ext": "r", "lang": "R", "max_stars_repo_path": "ShipPowerModel/R/calcPropDia.r", "max_stars_repo_name": "USEPA/Marine_Emissions_Tools", "max_stars_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 7, "max_stars_repo_stars_event_min_datetime": "2020-05-13T17:14:11.000Z", "max_stars_repo_stars_event_max_datetime": "2022-01-26T18:47:39.000Z", "max_issues_repo_path": "ShipPowerModel/R/calcPropDia.r", "max_issues_repo_name": "USEPA/Marine_Emissions_Tools", "max_issues_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "ShipPowerModel/R/calcPropDia.r", "max_forks_repo_name": "USEPA/Marine_Emissions_Tools", "max_forks_repo_head_hexsha": "28e12dc51acb5baafc460b1a9de35d355f3cc64f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2021-09-08T15:55:06.000Z", "max_forks_repo_forks_event_max_datetime": "2021-09-08T15:55:06.000Z", "avg_line_length": 37.406779661, "max_line_length": 168, "alphanum_fraction": 0.6991391029, "num_tokens": 635, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8267118026095991, "lm_q2_score": 0.6076631698328916, "lm_q1q2_score": 0.5023623145120127}} {"text": "\ncode 'sce_contrast_duration'\nname '市值与久期对照'\ndimensions {\n\taccbook null\n\tscene null\n\ttarget {\n\t\theadModels {\n\t\t\tmodel ([id:'market',name:'市值'])\n\t\t\tmodel ([id:'macaulayDuration',name:'麦考利久期'])\n\t\t\tmodel ([id:'modifiedDuration',name:'修正久期'])\n\t\t}\n\t}\n}\nlayout {\n\trows 'accbook'\n\tcolumns (['scene','target'])\n}\ndataRequest {\n\tname 'scene1'\n\ttable 'sceDuration'\n\tfields 'accbook,marketValue,macaulayDuration,modifiedDuration,cashflowPresentValue'\n\tcontextParams (['scene1'])\n}\ndataRequest {\n\tname 'scene2'\n\ttable 'sceDuration'\n\tfields 'accbook,marketValue,macaulayDuration,modifiedDuration,cashflowPresentValue'\n\tcontextParams (['scene2'])\n}\ndataGrids {\n\tapply selectRows({children!=null},{params['target']=='market'}),{\n\t\tformula {sumChildren()}\n\t}\n\tapply selectColumns({head.value=='macaulayDuration'||head.value=='modifiedDuration'}),{\n\t\tdataType 'numeric'\n\t\tformula {\n\t\t\tif(children==null){\n\t\t\t\tdef pv=v([target:'cfpv'])\n\t\t\t\t// v([target:'macaulayDuration']) or v([target:'modifiedDuration'])\n\t\t\t\tdef dur=v()\n\t\t\t\t[pv*dur,pv]\n\t\t\t}else{\n\t\t\t\tdef durs=[0.0,0.0]\n\t\t\t\tchildren.each{\n\t\t\t\t\tdef childDur=it.result\n\t\t\t\t\tdurs[0]+=childDur[0]\n\t\t\t\t\tdurs[1]+=childDur[1]\n\t\t\t\t}\n\t\t\t\tdurs\n\t\t\t}\n\t\t}\n\t\tevaluatedCallback {\n\t\t\tdef duration=value[0]\n\t\t\tdef cfpv=value[1]\n\t\t\tif(cfpv > 0.0){\n\t\t\t\tvalue=duration/cfpv\n\t\t\t}else{\n\t\t\t\tvalue=0.0\n\t\t\t}\n\t\t}\n\t}\n\tapply sel([scene:'scene2']),{\n\t\tparams {\n\t\t\tdataRequest 'scene2'\n\t\t}\n\t}\n}\n", "meta": {"hexsha": "162ff1a17462ac4b1e8c492b924e1c9c40804984", "size": 1400, "ext": "rd", "lang": "R", "max_stars_repo_path": "demo-untidy/sce_contrast_duration.rd", "max_stars_repo_name": "wushexu/jyreport", "max_stars_repo_head_hexsha": "7a4e2beec321aa3244e4ba0636066cd517a1c347", "max_stars_repo_licenses": ["Apache-2.0"], "max_stars_count": 2, "max_stars_repo_stars_event_min_datetime": "2018-04-13T01:51:58.000Z", "max_stars_repo_stars_event_max_datetime": "2021-04-20T03:19:05.000Z", "max_issues_repo_path": "demo-untidy/sce_contrast_duration.rd", "max_issues_repo_name": "wushexu/jyreport", "max_issues_repo_head_hexsha": "7a4e2beec321aa3244e4ba0636066cd517a1c347", "max_issues_repo_licenses": ["Apache-2.0"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "demo-untidy/sce_contrast_duration.rd", "max_forks_repo_name": "wushexu/jyreport", "max_forks_repo_head_hexsha": "7a4e2beec321aa3244e4ba0636066cd517a1c347", "max_forks_repo_licenses": ["Apache-2.0"], "max_forks_count": 2, "max_forks_repo_forks_event_min_datetime": "2015-06-27T04:06:48.000Z", "max_forks_repo_forks_event_max_datetime": "2016-08-05T03:04:09.000Z", "avg_line_length": 20.2898550725, "max_line_length": 88, "alphanum_fraction": 0.6571428571, "num_tokens": 453, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7662936430859597, "lm_q2_score": 0.6548947155710234, "lm_q1q2_score": 0.501841657432663}} {"text": "subroutine dldins(a,b,slope,rwu,ai,bi,rw,intfnd,bpt,nedge)\n\n# Get a point ***inside*** the rectangular window on the ray from\n# one circumcentre to the next one. I.e. if the `next one' is\n# inside, then that's it; else find the intersection of this ray with\n# the boundary of the rectangle.\n# Called by dirseg, dirout.\n\nimplicit double precision(a-h,o-z)\ndimension rw(4)\nlogical intfnd, bpt, rwu\n\n# Note that (a,b) is the circumcentre of a Delaunay triangle,\n# and that slope is the slope of the ray joining (a,b) to the\n# corresponding circumcentre on the opposite side of an edge of that\n# triangle. When `dldins' is called by `dirout' it is possible\n# for the ray not to intersect the window at all. (The Delaunay\n# edge between the two circumcentres might be connected to a `fake\n# outer corner', added to facilitate constructing a tiling that\n# completely covers the actual window.) The variable `intfnd' acts\n# as an indicator as to whether such an intersection has been found.\n\n# The variable `bpt' acts as an indicator as to whether the returned\n# point (ai,bi) is a true circumcentre, inside the window (bpt == .false.),\n# or is the intersection of a ray with the boundary of the window\n# (bpt = .true.).\n\n\nintfnd = .true.\nbpt = .true.\n\n# Dig out the corners of the rectangular window.\nxmin = rw(1)\nxmax = rw(2)\nymin = rw(3)\nymax = rw(4)\n\n# Check if (a,b) is inside the rectangle.\nif(xmin<=a&a<=xmax&ymin<=b&b<=ymax) {\n ai = a\n bi = b\n\tbpt = .false.\n nedge = 0\n return\n}\n\n# Look for appropriate intersections with the four lines forming\n# the sides of the rectangular window.\n\n# If not \"the right way up\" then the line joining the two\n# circumcentres is vertical.\n\nif(!rwu) {\n if(b < ymin) {\n ai = a\n bi = ymin\n nedge = 1\n if(xmin<=ai&ai<=xmax) return\n }\n if(b > ymax) {\n ai = a\n bi = ymax\n nedge = 3\n if(xmin<=ai&ai<=xmax) return\n }\n intfnd = .false.\n return\n}\n\n# Line 1: x = xmin.\nif(axmax) {\n ai = xmax\n bi = b + slope*(ai-a)\n nedge = 4\n if(ymin<=bi&bi<=ymax) return\n}\n\n# Line 4: y = ymax.\nif(b>ymax) {\n bi = ymax\n ai = a + (bi-b)/slope\n nedge = 3\n if(xmin<=ai&ai<=xmax) return\n}\n\nintfnd = .false.\nreturn\nend\n", "meta": {"hexsha": "925ee8eb3046c3423d584c58b2a67fa4e95832c4", "size": 2594, "ext": "r", "lang": "R", "max_stars_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/dldins.r", "max_stars_repo_name": "hyeongmokoo/SAAR_beta1", "max_stars_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 5, "max_stars_repo_stars_event_min_datetime": "2018-08-23T15:35:47.000Z", "max_stars_repo_stars_event_max_datetime": "2020-11-24T12:20:59.000Z", "max_issues_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/dldins.r", "max_issues_repo_name": "hyeongmokoo/SAAR_beta1", "max_issues_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_issues_repo_licenses": ["MIT"], "max_issues_count": 2, "max_issues_repo_issues_event_min_datetime": "2020-08-17T15:14:11.000Z", "max_issues_repo_issues_event_max_datetime": "2020-09-23T21:55:49.000Z", "max_forks_repo_path": "VisUncertainty/bin/Debug/R-3.4.4/library/deldir/ratfor/dldins.r", "max_forks_repo_name": "hyeongmokoo/SAAR_beta1", "max_forks_repo_head_hexsha": "7406f30d7f39a03e8494b2171c2bc09904a36f62", "max_forks_repo_licenses": ["MIT"], "max_forks_count": 1, "max_forks_repo_forks_event_min_datetime": "2020-11-04T05:34:16.000Z", "max_forks_repo_forks_event_max_datetime": "2020-11-04T05:34:16.000Z", "avg_line_length": 24.7047619048, "max_line_length": 75, "alphanum_fraction": 0.6087124133, "num_tokens": 794, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7772998714925403, "lm_q2_score": 0.6442250928250375, "lm_q1q2_score": 0.5007560818651715}} {"text": "###############################################################################\n# This program is free software: you can redistribute it and/or modify\n# it under the terms of the GNU General Public License as published by\n# the Free Software Foundation, either version 3 of the License, or\n# (at your option) any later version.\n#\n# This program is distributed in the hope that it will be useful,\n# but WITHOUT ANY WARRANTY; without even the implied warranty of\n# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\n# GNU General Public License for more details.\n#\n# You should have received a copy of the GNU General Public License\n# along with this program. If not, see .\n###############################################################################\n# Binary Branch and Bound and it's adaption for QP problem\n# Copyright (C) 1998-2000 Alberto Bemporad, Domenico Mignone - author's of the original Matlab version\t\t\t\t\t\n# Copyright (C) 2011 Michael Kapler - adapted code to R\n#\n# For more information please visit my blog at www.SystematicInvestor.wordpress.com\n# or drop me a line at TheSystematicInvestor at gmail\n###############################################################################\n\n\n###############################################################################\n# Binary Branch and Bound algorithm adpated from\n# miqp.m, a Matlab function for solving Mixed Integer Quadratic Programs\n# by Alberto Bemporad, Domenico Mignone\n# The routine was modified to work with large set of optimization problems.\n#\n# http://www.aut.ee.ethz.ch/~hybrid/miqp/ \n#' @export \n###############################################################################\nbinary_branch_bound <- function\n(\n\tindex_binvar,\t\t# index of binary[0/1] variables\n\tbbb_data, \t\t\t# data used for solving problems\n\tbbb_solve, \t\t\t# bbb_solve(bbb_data, binvar_lb, binvar_ub) - function to solve problems\n\tcontrol = bbb_control()\t# control the behavior of binary_branch_bound\n)\n{\n# Output: \n# xmin: minimizer of the cost function\n# fmin: minimum value of the cost function\n# counter: number of executions\n# flag: integer flag characterizing the result, where:\n# if flag = 1 there exists a feasible solution\n# if flag = 5 the solution is not integer feasible\n# if flag = 7 no feasible solution exists\n\t\n\tfbest = Inf\n\txbest = 0 * bbb_data$x0\n\tcounter = 0\n\tnbinvar = length(index_binvar)\n\tflag = 7 \t# by default it is infeasible\n\t\t\t\t\n\t# The Variable STACK will contain the subproblems\n\tstack = new.env()\n\t\tstack$data = list()\n\t\tstack$cost = c()\n\t\tstack$pointer = c()\n\tstack$data[[1]] = list(lb = bbb_data$lb, \n\t\t\t\t\t\t\tub = bbb_data$ub, \n\t\t\t\t\t\t\tvar = 1:nbinvar, \n\t\t\t\t\t\t\tpath = rep(0,nbinvar), \n\t\t\t\t\t\t\tlevel = 0, \n\t\t\t\t\t\t\tfval = Inf)\t\n\tstack$cost = 0\t\t# Array storing the cost of the problems, ordered in decreasing fashion (cost(1)=largest value)\n\tstack$pointer = 1\t# pointer stores the order of the list\t\n\t\n\tcontrol$proborder.selected = control$proborder\n\t\n\tif(F) {\n\t\tlb = bbb_data$lb \n\t\tub = bbb_data$ub\n\t\t \n\t\t# presolve two default cases\n\t\tfor( i in 0:1 ) {\n\t\t\tlb[] = i\n\t\t\tub[] = i\n\t\t\tsol = match.fun(bbb_solve)(bbb_data, lb, ub)\n\t\t\n\t\t\tif( sol$ok ) { \n\t\t\t\tx = sol$x\n\t\t\t\tfval = sol$fval\n\t xi = x[index_binvar]\t# binary variables\n\t\t\t\t \n\t\t\t\t# found solution\n\t\t\t\tif ( max(abs( round(xi,0) - xi )) < control$bineps ) {\n\t\t\t\t\tfbest = fval\n\t\t\t\t xbest = x \n\t\t\t\t flag = 1\n\t\t\t\t if( !control$silent ) cat('FOUND SOLUTION =', fbest, '\\n');\n\t\t\t\t} \n\t\t\t}\n\t\t}\n\t}\n\t\n\t# Main Loop\n\twhile ( length(stack$data) > 0 ) {\t\n\t # Get the next subproblem from the STACK\n\t\tsubprob = bbb_pop(stack)\n\t\t\n\t\tif( !control$silent ) {\n\t\t\tcat('-----------------------------------------------------', '\\n')\n\t\t\tif( max(subprob$path) > 0 ) {\n\t\t\t\ttemp.index = order(-subprob$path)[1 : sum(subprob$path > 0)]\n\t\t\t\tcat('\\t', \n\t\t\t\t\tpaste('b', temp.index, ' = ', subprob$lb[temp.index],sep='') \n\t\t\t\t\t, '\\n')\t\t\t\t\n\t\t\t} else {\n\t\t\t\tcat(counter, '\\t', 'FIRST NODE', '\\n')\n\t\t\t}\n\t\t\t\n\t\t\tcat(counter, '\\t', subprob$lb, '\\t', subprob$var, '\\t', subprob$fval, '\\t', fbest, '\\n')\n\t\t\tcat('\\t', subprob$ub, '\\n')\n\t\t\tcat('stack size =', len(stack$pointer), '\\n')\n\t\t}\n\t \n\t\tif( is.finite( subprob$fval ) & is.finite( fbest ) & fbest <= subprob$fval ) {\n\t\t\t# skip this problem because fbest is alredy smaller\n\t\t\tif( !control$silent ) cat('SKIP this problem because a solution with lower FVAL already found\\n')\n\t\t} else {\n\t\t\t\n\t\t # Solve the qp\n\t\t counter = counter + 1\n\t\t\tsol = match.fun(bbb_solve)(bbb_data, subprob$lb, subprob$ub)\n\n\t\t\t \n\t\t\tif( !sol$ok ) { \n\t\t\t\tif( !control$silent ) cat('NO SOLUTION EXISTS\\n\\n');\t\t\t\t\t\t\t\t\t\t\t\t \t\t\t\n\t\t\t} else {\n\t\t\t\tx = sol$x\n\t\t\t\tfval = sol$fval\n\t\t\t\t\n\t\t\t\tif( !control$silent ) {\n\t\t\t\t\tcat('SOLUTION OK', '\\t', sol$fval, '\\n')\t\t\t\t\t\t\t\t \t\t\t\t\t\t\t\t\n\t\t\t\t\tcat('\\t', round(x[index_binvar[subprob$var]],3), '\\n\\n')\n\t\t\t\t}\n\n\t\t\t\t\n\t\t if ( flag !=1 ) flag=5\n\t\t\n\t\t # Check if value function is better than the value so far\n\t\t if ( fval <= fbest ) {\n\t\t\t if ( length(subprob$var ) == 0 ) {\n\t\t\t \t# found solution\n\t\t\t\t\t\tfbest = fval \n\t\t\t xbest = x \n\t\t\t flag = 1 \n\t\t\t\t\t\tif( !control$silent ) cat('FOUND SOLUTION =', fbest, '\\n');\n\t\t\t\t\t} else {\n\t\t\t xi = x[index_binvar[subprob$var]]\t# binary variables\n\t\t\t \n\t\t\t # found solution\n\t\t\t if ( max(abs( round(xi,0) - xi )) < control$bineps ) {\n\t\t\t fbest = fval\n\t\t\t xbest = x \n\t\t\t flag = 1\n\t\t\t if( !control$silent ) cat('FOUND SOLUTION =', fbest, '\\n');\n\t\t\t } else {\n\t\t\t # split problem in 0/1 subproblems\t \n\t\t\t branchvar = bbb_decision(xi,control)\n\t\t\t probs = bbb_separate(subprob, branchvar, fval)\n\t\t\t p0 = probs$p0\n\t\t\t p1 = probs$p1\n\t\t\t \n\t\t\t if( !control$silent ) cat('Branch on =', subprob$var[branchvar], '\\n');\n\t\t\t \n\t\t\t \n\t\t\t \n\t\t\t if( control$searchdir == 0 ) { \n\t\t\t\t\t\t\t\tcost=1/(subprob$level+1) \t# Depth first\n\t\t\t\t\t\t\t} else if( control$searchdir == 1 ) { \n\t\t\t\t\t\t\t\tcost=subprob$level+1\t\t# Breadth first\n\t\t\t\t\t\t\t} else if( control$searchdir == 2 ) { \n\t\t\t\t\t\t\t\tcost=fval\t\t\t\t\t# Best-first. This tends to go breadth-first\n\t\t\t\t\t\t\t} else if( control$searchdir == 3 ) { \n\t\t\t\t\t\t\t\tcost=fval/(subprob$level+1)\t# This privilegiates deep nodes\n\t\t\t\t\t\t\t}\t\t\t\t\t\n\t\t\t\t\t\t\t\n\t\t\t\t\t\t\tif( control$proborder == 2 ) {\n\t\t\t\t\t\t\t\tcontrol$proborder.selected = round(xi[branchvar],0)\n\t\t\t\t\t\t\t}\n\t\t\t\t\t\t\t\n\t\t\t if( control$proborder.selected == 0 ) {\n\t\t\t \tbbb_push(stack, p1, p0, cost)\n\t\t\t } else {\n\t\t\t \tbbb_push(stack, p0, p1, cost) \n\t\t\t\t\t\t\t}\n\t\t\t\t\t\t}\n\t\t\t\t\t}\n\t\t\t\t}\n\t\t\t}\n\t\t\t\t\t\t\n\t\t # verbose\n\t\t if( F ) {\n\t\t cat('counter =', counter, '\\n')\n\t\t cat('fbest =', fbest, '\\n')\n\t\t cat('stack$pointer =', stack$pointer, '\\n')\n\t\t cat('\\n')\n\t\t }\n\t\t}\n\t} #end while\n\trm(list=ls(stack,all=TRUE), envir=stack)\n\n\t#xbest[index_binvar] = round(xbest[index_binvar],0) # ROUNDOFF binary solution\t\n\treturn(list(xmin = xbest, fmin = fbest, counter = counter, flag = flag))\t\n}\n\n###############################################################################\n# Decision: find next branching variable\n###############################################################################\nbbb_decision <- function\n(\n\txi,\t\t\t# x for binary variables\n\tcontrol\t\t# control the behavior of binary_branch_bound\n)\n{\n\tif( control$branchvar == 0 ) {\n\t\t# first free variable is chosen as branching variable\n branchvar = 1\n\t} else if( control$branchvar == 1 ) {\n # variable with max frac part is chosen as branching variable\n branchvar = which.max( abs(xi-round(xi,0)) )\t#pick up the first of with max value\t\n\t} else if( control$branchvar == 2 ) {\n # variable with min frac part is chosen as branching variable\n branchvar = which.min( abs(xi-round(xi,0)) )\t#pick up the first of with min value\t \n\t} else {\n\t\tbranchvar = 1\n\t}\n\treturn(branchvar)\n}\n\n###############################################################################\n# Pop: returns top element of the STACK and eliminate the element from the stack\n###############################################################################\nbbb_pop <- function(stack)\n{\n\ti = stack$pointer[ length(stack$data) ]\n\tsubprob = stack$data[[i]]\n\t\n\tstack$pointer[ stack$pointer > i ] = stack$pointer[ stack$pointer > i ] - 1\n\t\n\t# remove last\n\tstack$data[[i]] = NULL\n\tlength(stack$cost) = length(stack$data)\n\tlength(stack$pointer) = length(stack$data)\n\t\n\treturn(subprob)\n}\n\n###############################################################################\n# Push: puts a subproblem onto the STACK\n###############################################################################\nbbb_push <- function\n(\n\tstack, \t\t# stack structure\n\telement1, \t# element to push on stack\n\telement2,\t# element to push on stack\n\tcost\t\t# cost\n)\n{\n\tn = length(stack$data)\n\t\n\t# Determine position in STACK where problem is inserted, according to a best first strategy\n\ti = match(TRUE, stack$cost <= cost)\t\t# EX: STACKCOST=[100 80 33 22 ^ 5 3 2], cost=10\n\tif( is.na(i) ) i = n else i = i - 1\n\n\tstack$data[[ (n+1) ]] = element1\n\tstack$data[[ (n+2) ]] = element2\n\n\tif(i == 0) {\n\t\tstack$pointer=c((n+1),(n+2), stack$pointer)\n\t\tstack$cost=c(cost,cost, stack$cost)\n\t} else {\t\n\t\tstack$pointer=c(stack$pointer[1:i], (n+1),(n+2), stack$pointer[-c(1:i)])\n\t\tstack$cost=c(stack$cost[1:i], cost, cost, stack$cost[-c(1:i)])\n\t}\n}\n\n###############################################################################\n# Separate: generates 2 new suproblems from a given problem\n###############################################################################\nbbb_separate <- function\n(\n\tprob,\t\t# QP parent problem\n\tbranchvar,\t# branching variable\n\tfval\t\t# fval for parent problem\n)\n{\n\tif(length(prob$var) >= 1) {\n\t\tp0 = prob\n\t\t\tp0$fval = fval\n\t\t p0$level = prob$level + 1\n\t\t p0$var = prob$var[-branchvar]\n\t\t p0$path[ prob$var[branchvar] ] = 1 + max(p0$path)\n\t\tp1 = p0\n \t\t\n\t\tp0$lb[ prob$var[branchvar] ] = 0\n\t\tp0$ub[ prob$var[branchvar] ] = 0\n\t\t\n\t\tp1$lb[ prob$var[branchvar] ] = 1\n\t\tp1$ub[ prob$var[branchvar] ] = 1\n\t} else {\n\t\tstop('no more integer variables to branch on')\n\t}\n\treturn( list(p0 = p0, p1 = p1) )\n}\n\n", "meta": {"hexsha": "5809e9d30e482bdd5037feec6d53726526fa9f19", "size": 10521, "ext": "r", "lang": "R", "max_stars_repo_path": "patterns.matching/SIT/branchbound.r", "max_stars_repo_name": "wisonhang/Shiny_report", "max_stars_repo_head_hexsha": "bed828a4c3d88f37ba1cd16f31354541693f64fc", "max_stars_repo_licenses": ["MIT"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "patterns.matching/SIT/branchbound.r", "max_issues_repo_name": "wisonhang/Shiny_report", "max_issues_repo_head_hexsha": "bed828a4c3d88f37ba1cd16f31354541693f64fc", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "patterns.matching/SIT/branchbound.r", "max_forks_repo_name": "wisonhang/Shiny_report", "max_forks_repo_head_hexsha": "bed828a4c3d88f37ba1cd16f31354541693f64fc", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 33.9387096774, "max_line_length": 112, "alphanum_fraction": 0.5224788518, "num_tokens": 2778, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7371581626286834, "lm_q2_score": 0.679178699175393, "lm_q1q2_score": 0.500662121980672}} {"text": "#' title: Model definitions\n#'\n#' @usage\n#'\n#' death(rho, parms)\n#' colonization(rho, parms)\n#' mortality(rho, parms)\n#' growth(rho, parms)\n#'\n#' @details defining chances of death and colonization per location, depending on the global and local vegetation cover.\n#'\n#' @export\n#'\n\ndeath <- function(rho, parms = livestock$parms) {\n\n # set clustering zero if cover is zero\n cc = q_11(rho)/rho[[1]]\n cc[is.nan(cc)] <- 0\n\n # substitutions\n a = parms$a + parms$v * cc #* (1 - parms$p) * q_11(rho)\n a = a * rho[[1]]^(parms$q) # density dependent search efficiency: fr type III\n h = parms$h #^(1 + parms$p * q_11(rho))\n L = parms$L * (1 - parms$p * q_11(rho) )\n\n # individual death rate\n out <- parms$m + a * L / (1 + a * h * rho[[1]])\n\n out[out < 0] <- 0\n return(as.vector(out))\n\n}\n\n#' @export\nmortality <- function(rho, parms = livestock$parms) {\n death(rho, parms) * rho[[1]]\n}\n\n#' @export\n\ncolonization <- function(rho, parms = livestock$parms) {\n\n # substitutions\n b = parms$b + (1- parms$b) * parms$f * q_01(rho) # facilitation\n r = parms$r * rho[[1]]^parms$alpha # water runoff (not discussed)\n K = parms$K * (1 - parms$c * q_01(rho))\n\n # individual colonization rate\n out <- r * rho[[1]] * b * (1 - rho[[1]]/K ) / (1-rho[[1]])\n\n out[out < 0] <- 0\n return(as.vector(out))\n}\n\n#' @export\ngrowth <- function(rho, parms = livestock$parms) {\n colonization(rho, parms) * (1-rho[[1]])\n}\n", "meta": {"hexsha": "554dd712e70080519bb65aea99c034d15820a24f", "size": 1416, "ext": "r", "lang": "R", "max_stars_repo_path": "R/definitions.r", "max_stars_repo_name": "fdschneider/livestock", "max_stars_repo_head_hexsha": "d7f8767f1bfd447f6f885bcce527ca2fcc723be4", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 1, "max_stars_repo_stars_event_min_datetime": "2021-11-01T02:59:02.000Z", "max_stars_repo_stars_event_max_datetime": "2021-11-01T02:59:02.000Z", "max_issues_repo_path": "R/definitions.r", "max_issues_repo_name": "fdschneider/livestock", "max_issues_repo_head_hexsha": "d7f8767f1bfd447f6f885bcce527ca2fcc723be4", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "R/definitions.r", "max_forks_repo_name": "fdschneider/livestock", "max_forks_repo_head_hexsha": "d7f8767f1bfd447f6f885bcce527ca2fcc723be4", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 23.6, "max_line_length": 120, "alphanum_fraction": 0.5974576271, "num_tokens": 470, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.8289388167733099, "lm_q2_score": 0.6039318337259584, "lm_q1q2_score": 0.5006225396605313}} {"text": "#------------------------------------------------------------------------------------------------------------\r\n#\tQuasi-Likelihood modeling\r\n#\tLi et al. (2010) Comparitive shotgun proteomics using spectral count data and quasi-likelihood modeling.\r\n#\tJPR. 9(8):4295-305. PMID: 20586475\r\n#\r\n#\tPulled from http://forge.fenchurch.mc.vanderbilt.edu/scm/viewvc.php/branches/IDPicker-3/QuasiTel%20V2.R?root=idpicker&view=log\r\n#\tAdapted by Joseph N. Brown, Ph.D.\r\n#------------------------------------------------------------------------------------------------------------\r\n\r\nquasitel <- function(data, group1, group2, weight=NULL, rm.SID=TRUE, rm.zero=FALSE, minavgcount=NULL)\r\n{\r\n # should pass these as arguments instead\r\n option.contrasts <- getOption('contrasts')\r\n options(contrasts=c('contr.SAS', 'contr.treatment'))\r\n\r\n # make a grouping factor that will be applicable to the subsetted data\r\n group <- character()\r\n if (is.character(group1)) {\r\n group1 <- which(colnames(data) %in% group1)\r\n }\r\n if (is.character(group2)) {\r\n group2 <- which(colnames(data) %in% group2)\r\n }\r\n grp <- union(group1, group2)\r\n group[group2] <- \"group2\"\r\n group[group1] <- \"group1\"\r\n group <- factor(group[grp])\r\n gpl <- split(1:length(group), group)\r\n\r\n # subset the data and groups\r\n data <- subset(data, select=grp)\r\n #group1 <- gpl$group1\r\n #group2 <- gpl$group2\r\n\r\n # filter based on minimum count\r\n if (!is.null(minavgcount)) {\r\n grp12count <- apply(data, 1, mean)\r\n data <- subset(data, grp12count >= minavgcount)\r\n }\r\n\r\n #grp1zero <- apply(subset(data, select=group1), 1, sum) == 0\r\n #grp2zero <- apply(subset(data, select=group2), 1, sum) == 0\r\n #if (rm.zero) {\r\n # # only keep features that are not both zero\r\n # data <- subset(data, !(grp1zero | grp2zero))\r\n #} else {\r\n # # or add a single count\r\n # data[grp1zero, group1[1]] <- 1\r\n # data[grp2zero, group2[1]] <- 1\r\n #}\r\n\r\n # prepare the weight\r\n if (is.null(weight)) {\r\n offset <- NULL\r\n wei <- sapply(gpl, length)\r\n } else {\r\n weight <- weight[colnames(data)]\r\n offset <- log(weight)\r\n wei <- sapply(gpl, function(x) { sum(weight[x]) })\r\n }\r\n\r\n Nprotein <- nrow(data)\r\n result <- matrix(numeric(), nrow=Nprotein, ncol=11)\r\n\r\n for (i in 1:Nprotein) {\r\n count <- as.numeric(data[i,])\r\n\r\n # poisson p-value\r\n g1a <- glm(count ~ group, offset=offset, family=poisson)\r\n g1 <- glm(count ~ 1, offset=offset, family=poisson)\r\n anovaP <- data.frame(anova(g1, g1a, test=\"Chisq\"))\r\n Pvalues <- ifelse(anovaP[2,4] < 0.1e-15, 1, anovaP[2,5])\r\n\r\n # quasi p-value\r\n gquasi1a <- glm(count ~ group, offset=offset, family=quasi(link=log, variance=mu))\r\n gquasi1 <- glm(count ~ 1, offset=offset, family=quasi(link=log, variance=mu))\r\n anovaPq <- data.frame(anova(gquasi1, gquasi1a, test=\"F\"))\r\n Pvaluesq <- ifelse(anovaPq[2,4] < 0.1e-15, 1, anovaPq[2,6])\r\n\r\n lambda <- exp(rev(cumsum(as.numeric(g1a$coef))))\r\n totcot <- round(wei * lambda, 0)\r\n rateratio <- log2(lambda[1] / lambda[2])\r\n\r\n sdl <- sapply(gpl, function(x) { sd(count[x]) })\r\n meanl <- sapply(gpl, function(x) { mean(count[x]) })\r\n cvl <- mapply(\"/\", sdl, meanl)\r\n\r\n result[i, ] <- c( totcot, # 2 items\r\n lambda, # 2 items\r\n rateratio,\r\n Pvalues, NA,\r\n Pvaluesq, NA,\r\n cvl) # 2 items\r\n }\r\n # fdr adjustment\r\n for (j in c(6, 8)) {\r\n result[, j+1] <- p.adjust(result[, j], method=\"fdr\")\r\n }\r\n rownames(result) <- rownames(data)\r\n colnames(result) <- c( \"count1\", \"count2\",\r\n \"rates1\", \"rates2\",\r\n \"2log(rate1/rate2)\",\r\n \"poisson.p\", \"poisson.fdr\",\r\n \"quasi.p\", \"quasi.fdr\",\r\n \"cv1\", \"cv2\")\r\n options(contrasts=option.contrasts)\r\n result\r\n}", "meta": {"hexsha": "cde03f4002f8cf6e30c129f99ea9004374ef2ca2", "size": 4179, "ext": "r", "lang": "R", "max_stars_repo_path": "TestHarness/Docs/R_Scripts/QuasiTelV2.r", "max_stars_repo_name": "PNNL-Comp-Mass-Spec/Cyclops", "max_stars_repo_head_hexsha": "8e635dd41369247b869714eeff7380e013331ff1", "max_stars_repo_licenses": ["BSD-2-Clause"], "max_stars_count": null, "max_stars_repo_stars_event_min_datetime": null, "max_stars_repo_stars_event_max_datetime": null, "max_issues_repo_path": "TestHarness/Docs/R_Scripts/QuasiTelV2.r", "max_issues_repo_name": "PNNL-Comp-Mass-Spec/Cyclops", "max_issues_repo_head_hexsha": "8e635dd41369247b869714eeff7380e013331ff1", "max_issues_repo_licenses": ["BSD-2-Clause"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "TestHarness/Docs/R_Scripts/QuasiTelV2.r", "max_forks_repo_name": "PNNL-Comp-Mass-Spec/Cyclops", "max_forks_repo_head_hexsha": "8e635dd41369247b869714eeff7380e013331ff1", "max_forks_repo_licenses": ["BSD-2-Clause"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 38.6944444444, "max_line_length": 129, "alphanum_fraction": 0.5084948552, "num_tokens": 1171, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7853085808877581, "lm_q2_score": 0.6370308082623217, "lm_q1q2_score": 0.5002657600182654}} {"text": "#!/usr/bin/Rscript\n\narg <- commandArgs(trailingOnly=TRUE)\n\nrate_summary <- function(filepath)\n{\n\nrequire(MASS, quietly=TRUE, warn.conflicts=FALSE)\nrequire(TeachingDemos, quietly=TRUE, warn.conflicts=FALSE)\nrequire(hdrcde, quietly=TRUE, warn.conflicts=FALSE)\n\ncat(\" reading log-file...\\n\")\n\ntdat <- read.table(filepath, header=T)\nheadernames <- names(tdat)\n\nif (any(headernames == \"net.diversification\")) {\n\ta <- grep(\"speciation.rate\", headernames)\n\tb <- grep(\"extinction.rate\", headernames)\t\n\tc <- grep(\"net.diversification\", headernames)\n\td <- grep(\"extinction.fraction\", headernames)\n\t} else if (any(headernames == \"sp.rate_1\")) {\n\t\ta <- grep(\"^sp.rate_\", headernames)\n\t\tb <- rep(0, length(a))\n\t\tc <- rep(0, length(a))\n\t\td <- rep(grep(\"ext.frac\", headernames), length(a))\n\t\t} else {\n\t\t\ta <- grep(\"^speciation_\", headernames)\n\t\t\tb <- grep(\"^extinction_\", headernames)\n\t\t\tc <- grep(\"^net_div_\", headernames)\n\t\t\td <- grep(\"^ext_frac_\", headernames)\n\t}\n\nif (any(headernames == \"net.diversification\" | headernames == \"sp.rate_1\")) {\n\tclade.name <- NULL\n\t} else {\n\tclade.name <- headernames[c]\n\tclade.name <- gsub(\"net_div_\", \"\", clade.name)\n\t}\n\ndf <- data.frame(a,b,c,d)\n\nl <- c(\"speciation rate\", \"extinction rate\", \"net diversification\", \"extinction fraction\")\ncat(\" computing HPDs and printing pdf...\\n\")\n\noutput <- gsub (\"\\\\.log$\", \".pdf\", filepath)\n\npdf (file=output)\n\n# check if all models are pb\nif (sum(tdat[,d]) == 0) {\n\tfor (n in 1:length(a)) {\n\t\t\tvar_name <- l[1]\n\t\t\tcred_int <- emp.hpd (tdat[,a[n]], conf=0.95)\n\t\t\tmap <- round(hdr(tdat[,a[n]], all.modes=F)$mode, digits=3)\n\t\t\tminX <- floor(10*min(tdat[,a[n]]))/10\n\t\t\tmaxX <- ceiling(10*max(tdat[,a[n]]))/10\n\t\t\tahist <- hist(tdat[,a[n]], plot=F)\n\t\t\trate <- round(mean(tdat[,a[n]]), digits=3)\t\t\t\n\t\t\tif (length(a) > 1) {\n\t\t\t\thist(tdat[,a[n]], xlab = paste(sprintf(\"%s\", var_name), \"- mean:\", rate[1], \"/ MAP:\", map), sub = paste(\"95% HPD:\", sprintf(\"%.3f-%.3f\", cred_int[1], cred_int[2])), main = sprintf(\"rate %s\", n), ylim = c(0, max(ahist$counts+(ahist$counts/3))), xlim = c(minX, maxX))\n\t\t\t} else {\n\t\t\t\thist(tdat[,a[n]], xlab = paste(sprintf(\"%s\", var_name), \"- mean:\", rate[1], \"/ MAP:\", map), sub = paste(\"95% HPD:\", sprintf(\"%.3f-%.3f\", cred_int[1], cred_int[2])), main = \"speciation rate\", ylim = c(0, max(ahist$counts+(ahist$counts/3))), xlim = c(minX, maxX))\n\t\t\t}\n\t\t}\n\t\t\n\t} else {\n\npar(mfrow=c(2,2))\nfor (n in 1:length(a)) {\n\tif (sum(tdat[,d[n]]) == 0) {\n\t\tvar_name <- l[1]\n\t\tcred_int <- emp.hpd (tdat[,a[n]], conf=0.95)\n\t\tmap <- round(hdr(tdat[,a[n]], all.modes=F)$mode, digits=3)\n\t\tminX <- floor(10*min(tdat[,a[n]]))/10\n\t\tmaxX <- ceiling(10*max(tdat[,a[n]]))/10\n\t\tahist <- hist(tdat[,a[n]], plot=F)\n\t\trate <- round(mean(tdat[,a[n]]), digits=3)\n\t\thist(tdat[,a[n]], xlab = paste(sprintf(\"%s\", var_name), \"- mean:\", rate[1], \"/ MAP:\", map), sub = paste(\"95% HPD:\", sprintf(\"%.3f-%.3f\", cred_int[1], cred_int[2])), main = sprintf(\"clade: %s\", clade.name[n]), ylim = c(0, max(ahist$counts+(ahist$counts/3))), xlim = c(minX, maxX))\n\tfor (m in 1:3){\n\t\tframe()\n\t\t}\n\t\t\n\t} else {\n\t\t\n\tj = 1\n\tcoln <- as.vector(df[n,])\n\t\tfor(i in coln){\n\t\t\tvar_name <- l[j]\n\t\t\tcred_int <- emp.hpd (tdat[,i], conf=0.95)\n\t\t\tmap <- round(hdr(tdat[,i], all.modes=F)$mode, digits=3)\n\t\t\tminX <- floor(10*min(tdat[,i]))/10\n\t\t\tmaxX <- ceiling(10*max(tdat[,i]))/10\n\t\t\t\tif(j==4){\n\t\t\t\t\tminX=0\n\t\t\t\t\tmaxX=1}\n\t\t\tahist <- hist(tdat[,i], plot=F)\n\t\t\trate <- round(mean(tdat[,i]), digits=3)\n\t\t\thist(tdat[,i], xlab = paste(sprintf(\"%s\", var_name), \"- mean:\", rate[1], \"/ MAP:\", map), sub = paste(\"95% HPD:\", sprintf(\"%.3f-%.3f\", cred_int[1], cred_int[2])), main = sprintf(\"clade: %s\", clade.name[n]), ylim = c(0, max(ahist$counts+(ahist$counts/3))), xlim = c(minX, maxX), cex.lab=0.8, cex.sub=0.8, cex.main=0.9, cex.axis=0.9)\n\tj <- j+1\n\t\t\t}\n\t\t}\n\t}\n}\n\ndev.off()\ncat(\" results are saved in: \", sprintf(\"%s\", output), \"\\n\\n\")\n}\n\nrate_summary(as.character(arg[1]))", "meta": {"hexsha": "ad9f88eb0c4f095d91f18e9475f61308aa13e47f", "size": 3884, "ext": "r", "lang": "R", "max_stars_repo_path": "bayesrate/r_functions/rateHPD1.3.44.r", "max_stars_repo_name": "schnitzler-j/BayesRate", "max_stars_repo_head_hexsha": "86fb252df8589c89fce1c42699c69c5e2bbccb6f", "max_stars_repo_licenses": ["MIT"], "max_stars_count": 3, "max_stars_repo_stars_event_min_datetime": "2016-11-10T00:04:47.000Z", "max_stars_repo_stars_event_max_datetime": "2018-02-01T17:39:09.000Z", "max_issues_repo_path": "bayesrate/r_functions/rateHPD1.3.44.r", "max_issues_repo_name": "schnitzler-j/BayesRate", "max_issues_repo_head_hexsha": "86fb252df8589c89fce1c42699c69c5e2bbccb6f", "max_issues_repo_licenses": ["MIT"], "max_issues_count": null, "max_issues_repo_issues_event_min_datetime": null, "max_issues_repo_issues_event_max_datetime": null, "max_forks_repo_path": "bayesrate/r_functions/rateHPD1.3.44.r", "max_forks_repo_name": "schnitzler-j/BayesRate", "max_forks_repo_head_hexsha": "86fb252df8589c89fce1c42699c69c5e2bbccb6f", "max_forks_repo_licenses": ["MIT"], "max_forks_count": null, "max_forks_repo_forks_event_min_datetime": null, "max_forks_repo_forks_event_max_datetime": null, "avg_line_length": 35.3090909091, "max_line_length": 333, "alphanum_fraction": 0.5942327497, "num_tokens": 1385, "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.7634837635542925, "lm_q2_score": 0.6548947425132315, "lm_q1q2_score": 0.5000015027459214}}