diff --git a/5g_energy_consumption/019db4c5-d45f-7c07-80c9-bdd75ea01338/dataset_metadata.dataset-mold-v1.json b/5g_energy_consumption/019db4c5-d45f-7c07-80c9-bdd75ea01338/dataset_metadata.dataset-mold-v1.json index be120af6cd5c5713788c1b14b5a85951e0496d2c..86ece785c4598f6341523bac1ffc9937d30f06d8 100644 --- a/5g_energy_consumption/019db4c5-d45f-7c07-80c9-bdd75ea01338/dataset_metadata.dataset-mold-v1.json +++ b/5g_energy_consumption/019db4c5-d45f-7c07-80c9-bdd75ea01338/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- Note: The dataset was used in a Zindi challenge, but also uploaded to Huggingface under a MIT license by the company (Huawei).\n- The corresponding ITU/Zindi challenge explicitly emphasizes generalization to unseen base station products/configurations. We therefore split by base_station (BS).\n- With this setup, the task is development of predictive models for network optimization where models need to generalize to new unseen base station configurations and estimate their energy consumption under similar conditions (during the same time period).\n- For preprocessing, we orient on a top solution from the Zindi challenge: https://github.com/ITU-AI-ML-in-5G-Challenge/5G-Energy-Consumption-Modelling-Solution-Team-Farzi-Data-Scientists/tree/main.\n- Note that the competition also used mostly samples from the known BS as the test set, but weighted unknown base stations higher in evaluation.\n- The data itself is time-series. However, because we predict entirely unseen base stations in a per-sample fashion, the row-wise dependencies resulting from the temporal components cannot be used to improve performance unless the task is treated as transductive learning.\n- Following the Zindi solution, we merge the three given tables and use the Cell0 information only from the cell level table. In addition, we merge the Cell1 information since it contains information as well which might be useful in a grouped split setting.\n- Following the Zindi solution, we derive calendar features (`day`, `hour`, `weekday`) and drop the absolute timestamp.\n- We drop constant columns.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/acquire_valued_shoppers_challenge/019d7376-a3c5-7df7-88c0-ba4e87c3dfa0/dataset_metadata.dataset-mold-v1.json b/acquire_valued_shoppers_challenge/019d7376-a3c5-7df7-88c0-ba4e87c3dfa0/dataset_metadata.dataset-mold-v1.json index a5f5bbfc65976dbd3bbfa8498f86756b48c5e970..167e834fb1165f7e9c19de1cdd4e72e6d9ebc3e5 100644 --- a/acquire_valued_shoppers_challenge/019d7376-a3c5-7df7-88c0-ba4e87c3dfa0/dataset_metadata.dataset-mold-v1.json +++ b/acquire_valued_shoppers_challenge/019d7376-a3c5-7df7-88c0-ba4e87c3dfa0/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe follow the preprocessing from TabRed https://github.com/yandex-research/tabred/tree/main/preprocessing#ecom-offers-acquire-valued-shoppers-by-dmdave (which follows a top solution https://github.com/MLWave/kaggle_acquire-valued-shoppers-challenge).\n\n- The preprocessing by TabRed collapses customer groups into single entries per customer. So we go from grouped-temporal data to only temporal data. Future work could look into a version without preprocessing.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/airfoil_self_noise/019d7366-4673-7d9a-9821-35604af616f6/dataset_metadata.dataset-mold-v1.json b/airfoil_self_noise/019d7366-4673-7d9a-9821-35604af616f6/dataset_metadata.dataset-mold-v1.json index c314c8ea8d459b46318338c17055635e0d41a6f7..9b004775b2485d0c818bafcbf70a1c463847cda3 100644 --- a/airfoil_self_noise/019d7366-4673-7d9a-9821-35604af616f6/dataset_metadata.dataset-mold-v1.json +++ b/airfoil_self_noise/019d7366-4673-7d9a-9821-35604af616f6/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "N/A", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/allstate_claims_severity/019d736a-2321-76da-a36a-ddc5e1fbf7b6/dataset_metadata.dataset-mold-v1.json b/allstate_claims_severity/019d736a-2321-76da-a36a-ddc5e1fbf7b6/dataset_metadata.dataset-mold-v1.json index 7236d0674c41056dddcc0449cfe0fb757d126aa0..eeba5bd1567d8f0160e6f3152a2543d31dd961ff 100644 --- a/allstate_claims_severity/019d736a-2321-76da-a36a-ddc5e1fbf7b6/dataset_metadata.dataset-mold-v1.json +++ b/allstate_claims_severity/019d736a-2321-76da-a36a-ddc5e1fbf7b6/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized.\n- Top Kaggle solutions did not perform any relevant preprocessing.\n- We drop the ID column, as it does not contain any signal.\n- The data contains 1 duplicate row when ignoring the target. We drop this artifact.\n- We log scale the target.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/amazon_employee_access/019d7366-58b8-7e97-bf34-83c96a915561/dataset_metadata.dataset-mold-v1.json b/amazon_employee_access/019d7366-58b8-7e97-bf34-83c96a915561/dataset_metadata.dataset-mold-v1.json index fddb7c05d114ac92bad79859975db2d29ea7fb75..b693ae920df3248dcbc08ff848038d6dbeabaa73 100644 --- a/amazon_employee_access/019d7366-58b8-7e97-bf34-83c96a915561/dataset_metadata.dataset-mold-v1.json +++ b/amazon_employee_access/019d7366-58b8-7e97-bf34-83c96a915561/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We only use the training data from Kaggle.\n- We renamed the target \"ACTION\" to \"ResourceApproved\" and mapped binary values to \"Yes\"/\"No\".\n- Anomaly: the data might contain sub-groups related to managers and resources.\n- Anomaly: likely, similar to the test data, each sample represents a unique employee.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/amex_non_iid/versions/019d7455-0e4e-7261-9842-93177684d486/dataset_metadata.dataset-mold-v1.json b/amex_non_iid/versions/019d7455-0e4e-7261-9842-93177684d486/dataset_metadata.dataset-mold-v1.json index a9f2a70b596bb48a18d5e1926293d0015f52a237..506f1e3b54096a29838cbcd1cd0f2556674f3433 100644 --- a/amex_non_iid/versions/019d7455-0e4e-7261-9842-93177684d486/dataset_metadata.dataset-mold-v1.json +++ b/amex_non_iid/versions/019d7455-0e4e-7261-9842-93177684d486/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with the raw data from Kaggle and do not apply any further preprocessing to simulate a pipeline that can handle raw non-IID grouped data.\n", "version_from_unique_name": "amex_non_iid", "version_comment": "\nWe randomly sub-sample to 1.25 million rows such that train and test splits are roughly 1M and 250k rows respectively.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/anes_voting_2026/019db4ba-6124-7a56-a0b5-9df8811adc2f/dataset_metadata.dataset-mold-v1.json b/anes_voting_2026/019db4ba-6124-7a56-a0b5-9df8811adc2f/dataset_metadata.dataset-mold-v1.json index 3df29303189b4d92c624a32192d7fead059d3e06..5c26a9e4a7d8b9060a21ce73c3e5510ba6e6f7b1 100644 --- a/anes_voting_2026/019db4ba-6124-7a56-a0b5-9df8811adc2f/dataset_metadata.dataset-mold-v1.json +++ b/anes_voting_2026/019db4ba-6124-7a56-a0b5-9df8811adc2f/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- We remove all samples with missing pre-election data, missing post-election data or a missing target.\n- We drop 327 that are missing from the last 9 years, which will be used for the test splits.\n- We drop all columns with questions that appear only in one year.\n- The question columns partially stem from post-election interviews. However, there is no simple way to tell which is post-election. We therefore drop all questions for which \"no post IW\" is listed as a reason for missing values. Some additional questions are dropped based on common sense.\n- We encode gender as a categorical variable, because Other was introduced in 2016. However there are very few of these samples. This would actually require split-specific preprocessing.\n- For all features, we assign \" \" as missing values. Most features have additional types of missingness with own codes. We keep them as categories. That means NA only is assigned if a question was missing in a survey.\n- We transform only high-cardinality (>10) numeric features to numeric, and leave all other ordinal features as categorical, since they all include at least one value that is out-of-order (e.g., \"Don't know\", \"Refused\", \"Not applicable\", \"Other\", etc.).\n- Although there are several ways to further preprocess features, we leave the preprocessing minimal to enable the development of models that can handle even complex preprocessing on their own.\n- Note: Alternative target could be constructed using who the candidate voted for (VCF0704). This could be a multi-class target with [Democrat, Republican, Other Party, Did not vote, Voted but unknown who]. However, the last category is less useful to predict, and might introduce problems. We therefore stick to predicting whether someone voted or not.\n- Note: While the data was also used in the TableShift benchmark, we define an entirely different task, with a temporal split and more features.\n- Note: Some information is represented across multiple columns, e.g., ['VCF0009x', 'VCF0009y', 'VCF0009z', 'VCF0010x', 'VCF0010y', 'VCF0010z', 'VCF0011x', 'VCF0011y', 'VCF0011z'] all correspond to weight. We keep all of these columns, even though some might be redundant, because we want to keep the data as close to the original as possible leaving the challenge to the model.\n- Note: Ideally, we would select different feature sets per data split, depending on feature availability. However, for simplicity, we just keep all features.\n- Note: We considered keeping the unique respondent identifier in the data, because otherwise models that are able to reconstruct this information would unfairly benefit. However, for now we drop the identifier since it is not a desirable predictive feature for the given task, because the goal is generalization to the population, not memorizing individual behavior.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/aps_failure/019d7366-7373-7942-9438-aa122a6ca492/dataset_metadata.dataset-mold-v1.json b/aps_failure/019d7366-7373-7942-9438-aa122a6ca492/dataset_metadata.dataset-mold-v1.json index f370dec5ff00a78125396e906044c86f23def72e..c72565a3217d506a1ddc4142c10af6233ec5691c 100644 --- a/aps_failure/019d7366-7373-7942-9438-aa122a6ca492/dataset_metadata.dataset-mold-v1.json +++ b/aps_failure/019d7366-7373-7942-9438-aa122a6ca492/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n - We combined the original training and testing data into a single dataset.\n - We renamed the target feature to \"AirPressureSystemFailure\".\n - We converted \"na\" strings to real NaN/missing values, making the data numeric.\n - Anomaly: we cannot determine the data types of the features.\n - Anomaly: some features are bins of histograms (see original data description).\n - Anomaly: the original task used a cost matrix for evaluation.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/asp_potassco_classification/019d73ec-935f-7f88-8a82-af3ad2f44e9f/dataset_metadata.dataset-mold-v1.json b/asp_potassco_classification/019d73ec-935f-7f88-8a82-af3ad2f44e9f/dataset_metadata.dataset-mold-v1.json index 0f8f8d91b352b9200125a49c634ddaa2bf86fb34..c810b14cdbd5fb79e24e3e6c7ddb9f292d9694ad 100644 --- a/asp_potassco_classification/019d73ec-935f-7f88-8a82-af3ad2f44e9f/dataset_metadata.dataset-mold-v1.json +++ b/asp_potassco_classification/019d73ec-935f-7f88-8a82-af3ad2f44e9f/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe get the data from ASlib and merge them into one file.\n\n- We treat it as a multiclass classification task to solve the algorithm selection task as in the OpenML version (https://openml.org/d/41705).\n- We drop all cases where no algorithm was able to finish before the timeout as these are essentially random labels.\n- We resole the instance ID to a task ID by mapping the instance ID back to the source task based on their naming convention. From our understanding, the data contains multiple instance from the same solver task, differing only in seed or task configurations. We want to avoid having samples from the same task in both train and test set, as this would leak what algorithms are best for this task, and since in real-world you might not have samples from your new task. Thus, we treat the data as a grouped task, where we aim to generalize the predicting the best algorithm to new tasks across a set of instances of this task.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/audiology_diagnosis/019d736a-3b3d-73da-8d65-b4b7a39e1590/dataset_metadata.dataset-mold-v1.json b/audiology_diagnosis/019d736a-3b3d-73da-8d65-b4b7a39e1590/dataset_metadata.dataset-mold-v1.json index 15cf347338b4d500524263563ad273823736c7a4..666cc4295b37e0a5cd36cdb307935abe9813a29f 100644 --- a/audiology_diagnosis/019d736a-3b3d-73da-8d65-b4b7a39e1590/dataset_metadata.dataset-mold-v1.json +++ b/audiology_diagnosis/019d736a-3b3d-73da-8d65-b4b7a39e1590/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI and merge train and test data.\n\n- We drop the ID column as it is uninformative here.\n- The target label contains groups of labels with specifications. We merge labels into groups that represent general a diagnosis. We create 3 labels: \"normal\", \"cochlear\", \"other\". There is likely a much better way to partition these labels, but this is most reasonable partitioning for an acutal task, going from names and my limited domain knowledge about the various diagnoses.\n- We drop duplicated rows as they might introduce too much information leakage for such small data and are likely not natural but rather an artifact of the limited number of features.\n- We remove one constant column (history_fullness).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/bad_customer_detection/019d7366-8875-7ee1-8f3b-5ad96bdd2324/dataset_metadata.dataset-mold-v1.json b/bad_customer_detection/019d7366-8875-7ee1-8f3b-5ad96bdd2324/dataset_metadata.dataset-mold-v1.json index dae44a28f07d20736de792dab47573a2089398fe..80c8e4d3afe5a3ab82d6de257821fafd4d1ac60f 100644 --- a/bad_customer_detection/019d7366-8875-7ee1-8f3b-5ad96bdd2324/dataset_metadata.dataset-mold-v1.json +++ b/bad_customer_detection/019d7366-8875-7ee1-8f3b-5ad96bdd2324/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the values of the target variable to be more descriptive.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/bank_customer_churn/019d7366-99cc-77cc-8d41-5c3ccc8eff96/dataset_metadata.dataset-mold-v1.json b/bank_customer_churn/019d7366-99cc-77cc-8d41-5c3ccc8eff96/dataset_metadata.dataset-mold-v1.json index 605f181b9dbb122fe070a8e4aeffc18e5ca688f0..b8134233ddfc05a18b18ed5bf913c164e88111f3 100644 --- a/bank_customer_churn/019d7366-99cc-77cc-8d41-5c3ccc8eff96/dataset_metadata.dataset-mold-v1.json +++ b/bank_customer_churn/019d7366-99cc-77cc-8d41-5c3ccc8eff96/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We remove the customer_id column.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/bank_marketing/019d7366-abe0-7dcc-a9fb-069bfb233c78/dataset_metadata.dataset-mold-v1.json b/bank_marketing/019d7366-abe0-7dcc-a9fb-069bfb233c78/dataset_metadata.dataset-mold-v1.json index a494a3ed6e3179a66c0083c6c21b76437ab55fe3..cf0326f8ef2d57e98bc02e86e9bb8c598b26334b 100644 --- a/bank_marketing/019d7366-abe0-7dcc-a9fb-069bfb233c78/dataset_metadata.dataset-mold-v1.json +++ b/bank_marketing/019d7366-abe0-7dcc-a9fb-069bfb233c78/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We removed the \"duration\" feature following its original description to obtain a \"realistic predictive model\".\n- We further remove the \"month\" and \"day_of_week\" features, as they also relate to the last contact -- which is not available in a real-world scenario.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/biogeographical_ancestry_prediction/019d7374-b899-714c-9372-37a48f580b0f/dataset_metadata.dataset-mold-v1.json b/biogeographical_ancestry_prediction/019d7374-b899-714c-9372-37a48f580b0f/dataset_metadata.dataset-mold-v1.json index decb6f51d219634e4497162a2e9ab5d1d91d6ee3..49671034f1c7a7d419c981cf6ebee3e5b50f4347 100644 --- a/biogeographical_ancestry_prediction/019d7374-b899-714c-9372-37a48f580b0f/dataset_metadata.dataset-mold-v1.json +++ b/biogeographical_ancestry_prediction/019d7374-b899-714c-9372-37a48f580b0f/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe use this dataset as one of the most recent example of a machine learning task based on the Human Genome project.\nWe take the targets from the paper by Heinzel et al. (2025) and only rename the targets to be standardized and more concise.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/biomechanical_orthopaedic_prediction/019d736a-4c2d-74b6-95a3-c63bd4b8fff7/dataset_metadata.dataset-mold-v1.json b/biomechanical_orthopaedic_prediction/019d736a-4c2d-74b6-95a3-c63bd4b8fff7/dataset_metadata.dataset-mold-v1.json index 03ef25a7a93cbf0e5359eb2183fa34190851b8fa..d4fc594b3c7dc1dbd7fe71f4aae7f5a0096d9b7b 100644 --- a/biomechanical_orthopaedic_prediction/019d736a-4c2d-74b6-95a3-c63bd4b8fff7/dataset_metadata.dataset-mold-v1.json +++ b/biomechanical_orthopaedic_prediction/019d736a-4c2d-74b6-95a3-c63bd4b8fff7/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the UCI version, which is the oldest existing version of the dataset we found.\n\n-\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/bioresponse/019d7366-c770-7d3c-8dd9-b9cc456542d6/dataset_metadata.dataset-mold-v1.json b/bioresponse/019d7366-c770-7d3c-8dd9-b9cc456542d6/dataset_metadata.dataset-mold-v1.json index 8c5f420d209bad358343166ed0e7a386ed62c784..408982ba323cffcc89035ea8352ee154a6baa0e8 100644 --- a/bioresponse/019d7366-c770-7d3c-8dd9-b9cc456542d6/dataset_metadata.dataset-mold-v1.json +++ b/bioresponse/019d7366-c770-7d3c-8dd9-b9cc456542d6/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We changed the name of the target and mapped it to yes/no\n- Anomaly: Only train data is used, since test data is empty for target\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/blood_tests_drink_prediction/019d736a-5d45-743a-b7c5-11da6b494fb7/dataset_metadata.dataset-mold-v1.json b/blood_tests_drink_prediction/019d736a-5d45-743a-b7c5-11da6b494fb7/dataset_metadata.dataset-mold-v1.json index f066e8c0abdc6d3e60b08698aba3e970878f59bb..e4696b5d7781efb8d0b663e5d98b4feb166cc100 100644 --- a/blood_tests_drink_prediction/019d736a-5d45-743a-b7c5-11da6b494fb7/dataset_metadata.dataset-mold-v1.json +++ b/blood_tests_drink_prediction/019d736a-5d45-743a-b7c5-11da6b494fb7/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nThe task is framed as a liver disorder prediction task, but we only have data bout the amount of drinks consumed. So instead, we will frame it as a task to predict the number of drinks based on the blood work data. This is a proxy for the original task, where later based on the number of drinks the liver disorder was determined.\n\n- We drop the selector column, as we ignore the original train/test split, and will create our own splits.\n- The data contains natural duplicates, so we keep them.\n- We log1p scale the target.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/blood_transfusion/019d7366-d9d0-777d-b0ae-e7a5175f7f09/dataset_metadata.dataset-mold-v1.json b/blood_transfusion/019d7366-d9d0-777d-b0ae-e7a5175f7f09/dataset_metadata.dataset-mold-v1.json index 629f4eb81097b69793ef7fc6f17986d831127ac0..0b4b4b9856d930b567528603aab035bf5634d5b3 100644 --- a/blood_transfusion/019d7366-d9d0-777d-b0ae-e7a5175f7f09/dataset_metadata.dataset-mold-v1.json +++ b/blood_transfusion/019d7366-d9d0-777d-b0ae-e7a5175f7f09/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We made feature names more descriptive.\n- We renamed the target and mapped binary values to \"Yes\"/\"No\".\n- Anomaly: the data has a lot of duplicates (29%) and several duplicates with different target values.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/body_density_prediction/019d736a-6e2b-77a5-ae25-1222ef4fe4e9/dataset_metadata.dataset-mold-v1.json b/body_density_prediction/019d736a-6e2b-77a5-ae25-1222ef4fe4e9/dataset_metadata.dataset-mold-v1.json index 32b5d36d29778a45e1129d0f5215057b44e03710..8b015f101fa2f420ac0011de54885f429143934e 100644 --- a/body_density_prediction/019d736a-6e2b-77a5-ae25-1222ef4fe4e9/dataset_metadata.dataset-mold-v1.json +++ b/body_density_prediction/019d736a-6e2b-77a5-ae25-1222ef4fe4e9/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from Kaggle.\n\n- Note, the task is not about bodyfat prediction as this is determined by a deterministic formula. Instead, the task is to estimate the density (which can be used to get the body fat via the deterministic formula). However, density requires a test. So we aim to predict from data the density such that we can skip this test. Which can make a lot of sense, if the test is too expensive as it requires underwater weighing.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/california_house_prices_2020/019d7388-c562-72b3-a533-5f588dc737c9/dataset_metadata.dataset-mold-v1.json b/california_house_prices_2020/019d7388-c562-72b3-a533-5f588dc737c9/dataset_metadata.dataset-mold-v1.json index ec9cd2f7a4ca9ea9d8cf2f7563357498100e0f8e..54ba3be7cbb54140e8edee58db48d04320a9430b 100644 --- a/california_house_prices_2020/019d7388-c562-72b3-a533-5f588dc737c9/dataset_metadata.dataset-mold-v1.json +++ b/california_house_prices_2020/019d7388-c562-72b3-a533-5f588dc737c9/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe follow some common preprocessing steps from Kaggle and enhance features as much as possible.\n\nWe found out that the data was web scraped from redfin.com and that the houses are ordered by the time they were sold such that a house with a higher ID was sold later. We confirmed this by looking at the history of various houses on redfin.com that we found in the dataset and checking their order in the ID with the time they were sold. Houses with a higher ID were sold later.\n\n- We log scale the target variable.\n- We do not have the exact dates but we know a higher ID means a later sale. We name the ID column accordingly to \"time_index\" and use it as a time feature.\n- The descriptions did contain the last sold price with a standard phrase such as \"This home last sold for $X in January 2020. The Zestimate for this house is $Y The Rent Zestimate for this home is $Z/mo.\". In these cases $X is identical to the target variable. Given such a drastic data leakage, we remove all rows (5413) that contain this phrase in the description to ensure they are not different in other ways too.\n- Note, sometime descriptions are missing and we intend that models/pipelines need to be able to handle this.\n- When investigating the \"Lot\" column we found many cases where the lot size (given in sq ft) is incorrect compared to the official lot size on the internet. The parser seems to have had an issue because the website often incorrectly showed the sq ft but gave the unit as acres. We found cases that were wrong by checking the acres size of the houses and anything with more than 2000 acres was investigated and subsequently corrected.\n- We parsed the bedrooms with a proxy following Kaggle and added an additional column containing only the semi-free text descriptions of the bedrooms.\n- We fix the parser errors for two cases for total interior livable area that were clear outliers based on the real data on redfin.\n- Garage and total space were in most cases the same values parsed from different fields. We keep Garage space and add an 'Extra Space' column that is the difference between total and garage space as an additional feature. This results in some cases with negative extra space that are most likely parser errors. We set them to nan. Note, we observed that that garage space number is in many cases a parsing error and does not reflect the real garage count.\n- The data has a lot of spatial features that are already semi-resolved to distances. This could be further enhanced using the address information. We leave it to the pipeline to handle this, if at all.\n- The dataset contains many text-like many categorical that are also high-cardinality, so perfect string data.\n- We drop all entries from the State of Arizona (AZ) as they are only a small fraction (n=431 after our preprocessing). The dataset is specified to be for California house prices.\n- We treat ZIP codes as strings. But we add a new categorical features that puts zip code in to approximate relevant buckets of regions.\n- We fixed one entry that had a negative value for garage space, likely due to a typo. This error still exists on the website.\n- After all the above preprocessing, there exist 63 houses that appear multiple times in the dataset (same address, zip code, year built). We only keep the newest entry to avoid group-related target leakage.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/cardiotocography/019d7392-12c2-7a87-991c-13a7246555d4/dataset_metadata.dataset-mold-v1.json b/cardiotocography/019d7392-12c2-7a87-991c-13a7246555d4/dataset_metadata.dataset-mold-v1.json index 6e222941d8bec7de48293a7b4fe50ac3212edd57..e472259f5b51601cf58d7e6b5bc6dd6274b864dc 100644 --- a/cardiotocography/019d7392-12c2-7a87-991c-13a7246555d4/dataset_metadata.dataset-mold-v1.json +++ b/cardiotocography/019d7392-12c2-7a87-991c-13a7246555d4/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe use the data from UCI and the 3 class problem of predicting NSP as it is more medical relevant.\n\n- We transform the file names into patient IDs to indicate the sub-group of samples recorded from the same patient.\n- The raw data has dates and the start and end index of the recording. We drop these as they are not relevant for the predictive task and would not be available for test samples.\n- We drop all other columns that represent the label.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/churn/019d7366-eb2d-72f9-9998-72dfc5b6cc79/dataset_metadata.dataset-mold-v1.json b/churn/019d7366-eb2d-72f9-9998-72dfc5b6cc79/dataset_metadata.dataset-mold-v1.json index 61b7f2b75854482d5096b9d8c27dd61a82cedf00..b7ed3aec6267d5b1bd598a5115b9daf000ed1f6f 100644 --- a/churn/019d7366-eb2d-72f9-9998-72dfc5b6cc79/dataset_metadata.dataset-mold-v1.json +++ b/churn/019d7366-eb2d-72f9-9998-72dfc5b6cc79/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- The original source is lost, so we use https://github.com/EpistasisLab/pmlb/tree/master/datasets/churn (or https://www.openml.org/d/40701)\n- We dropped the \"phone_number\" feature as it seems to be an index in the original data. \n- We renamed the target variable to \"CustomerChurned\" and mapped binary variables to \"Yes\"/\"No\"\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/cirrhosis_patient_survival_prediction/019d736a-9116-7589-9cad-4cee540f2926/dataset_metadata.dataset-mold-v1.json b/cirrhosis_patient_survival_prediction/019d736a-9116-7589-9cad-4cee540f2926/dataset_metadata.dataset-mold-v1.json index 48cd0d2d8cb7e99963a03a0b261273c0ed8115f3..f7b22717851a3e48ddc5dda3bc4383966ebe1030 100644 --- a/cirrhosis_patient_survival_prediction/019d736a-9116-7589-9cad-4cee540f2926/dataset_metadata.dataset-mold-v1.json +++ b/cirrhosis_patient_survival_prediction/019d736a-9116-7589-9cad-4cee540f2926/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the UCI data.\n\nWe note that this is a survival prediction task. But the way we split the data and score the task, it is not treated correctly as a survival prediction task. At the same time, it is randomized control data based on the Drug column. We treat it as a regression task to predict the time-to-death for non-censored patients, including the drug column into the modelling.\n\n- We drop ID and Status column as they have no relevance after filtering to only non-censored patients.\n- We log transform N_days and rename it to the target \"log_days_to_death\".\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/climate_model_weather_forecasting/versions/019d7379-7906-707e-95e9-f479afb2e2c1/dataset_metadata.dataset-mold-v1.json b/climate_model_weather_forecasting/versions/019d7379-7906-707e-95e9-f479afb2e2c1/dataset_metadata.dataset-mold-v1.json index 68caeba3204f32a2d94a995e28bac7ec8ef523d9..8ed4b2c00456ac636e0f8a4fecf36c9a7b89f7cc 100644 --- a/climate_model_weather_forecasting/versions/019d7379-7906-707e-95e9-f479afb2e2c1/dataset_metadata.dataset-mold-v1.json +++ b/climate_model_weather_forecasting/versions/019d7379-7906-707e-95e9-f479afb2e2c1/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with data from TabRed, which already comes preprocessed.\n\n- This data is similar to data from https://arxiv.org/abs/2406.19380. This paper also describes the feature in more detail. The column `apply_time_rl` is new and might relate to a rolling lag/window indicator.\n- We drop duplicated columns `cmc_0_1_66_0` , `cmc_0_1_67_0`, `cmc_0_1_68_0`.\n", "version_from_unique_name": "climate_model_weather_forecasting", "version_comment": "\nWe randomly sub-sample the train to 1 million and test data 250k rows. We follow TabReD and use random sub-sampling. The idea behind this instead of a time-based subsampling is to keep data from various time periods and model the distribution shift across the full time horizon.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/clock_protein_toxicity/019d7375-2841-7bfb-a6da-5adfdcd43d98/dataset_metadata.dataset-mold-v1.json b/clock_protein_toxicity/019d7375-2841-7bfb-a6da-5adfdcd43d98/dataset_metadata.dataset-mold-v1.json index 26069aa418c604b9f13125c2ccf87415a0c57a7f..9bafd93a14fc3ba54186a06e2a6e41e85c8b9823 100644 --- a/clock_protein_toxicity/019d7375-2841-7bfb-a6da-5adfdcd43d98/dataset_metadata.dataset-mold-v1.json +++ b/clock_protein_toxicity/019d7375-2841-7bfb-a6da-5adfdcd43d98/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We remove duplicated columns (same values for all rows).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/coffee_rating_prediction/019d7388-dffd-7bc4-ab3d-40bf6581786e/dataset_metadata.dataset-mold-v1.json b/coffee_rating_prediction/019d7388-dffd-7bc4-ab3d-40bf6581786e/dataset_metadata.dataset-mold-v1.json index 0d4680149a2c6fbed234378d26f8827a3cc50c2e..270f36eaf67598a1f662888f8a4710d2f9366b53 100644 --- a/coffee_rating_prediction/019d7388-dffd-7bc4-ab3d-40bf6581786e/dataset_metadata.dataset-mold-v1.json +++ b/coffee_rating_prediction/019d7388-dffd-7bc4-ab3d-40bf6581786e/dataset_metadata.dataset-mold-v1.json @@ -17,6 +17,5 @@ "curation_comments": "\nWe start with the Kaggle version. Note, we acknowledge that the dataset is not of the highest quality and likely does not represent a real-world prediction task. Nevertheless, it can be used to test models.\n\n- We drop slug as it is a unique identifier that does not have predictive power.\n- We drop desc_2 as it in general includes notes that may or may not be related to the coffee quality. Moreover, it sometimes leaks the rating (e.g. through award notes). We only keep desc_1 and desc_3, which are mostly descriptions of the coffee.\n- We drop all_text as it is a concatenation of desc_1, desc_2, and desc_3. And desc_2 includes data leakage (such as if the coffee won an award or not). Thus, we drop all_test.\n- We drop aroma, acid, body, flavor, aftertaste, and with_milk as they are all features of the rating and thus leaking the target. Any of them could be used as a target, we focus on the overall rating.\n- We parse agtron into a lower and upper value as float.\n- Note, the data contains several spatial features about locations. We do not resolve these.\n- The price feature is extremely messy as a result of web scrapping. We resolve this feature as real tabular data (not from a web scrapper but from the database source), would have these information stored in a more structured way. We drop coffees that are measured in \"capsules\" units as they exist only 8 times and seem to be strong outliers. We normalizes the price to be per grams. Moreover, we convert all currencies to USD (with a fixed rate); we ignore the temporal incorrectness this may introduce as it is most likely minimal.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/coil_2000/019d7366-fdc2-7225-a22c-314c624711f6/dataset_metadata.dataset-mold-v1.json b/coil_2000/019d7366-fdc2-7225-a22c-314c624711f6/dataset_metadata.dataset-mold-v1.json index d9843a69e582e0c9c2be376aa34ee82dd8282c82..21eb5f5d8087be4de1aec7e9f097d44dc24f4c0e 100644 --- a/coil_2000/019d7366-fdc2-7225-a22c-314c624711f6/dataset_metadata.dataset-mold-v1.json +++ b/coil_2000/019d7366-fdc2-7225-a22c-314c624711f6/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We created semantic meaningful names for the features.\n- We combined the original training and validation data into one new dataset.\n- We reversed the ordinal encoding of the original data where possible.\n- Anomaly: the data has 15% duplicates.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/concrete_compressive_strength/019d7367-0f12-7c25-b411-de9c711bd4ec/dataset_metadata.dataset-mold-v1.json b/concrete_compressive_strength/019d7367-0f12-7c25-b411-de9c711bd4ec/dataset_metadata.dataset-mold-v1.json index efb3ba2fa2e54475d5c935b62f27a4fdeb4d6f7d..966fb4a6465d622fe441cb8ebbe86709e710b4ad 100644 --- a/concrete_compressive_strength/019d7367-0f12-7c25-b411-de9c711bd4ec/dataset_metadata.dataset-mold-v1.json +++ b/concrete_compressive_strength/019d7367-0f12-7c25-b411-de9c711bd4ec/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We rename features to be shorter while similar to the original names.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/consumer_complaints/versions/019d738b-4c6e-751f-972a-5f63b1508f70/dataset_metadata.dataset-mold-v1.json b/consumer_complaints/versions/019d738b-4c6e-751f-972a-5f63b1508f70/dataset_metadata.dataset-mold-v1.json index 07cf06aa0af14df3165a2a95f03c90ef9dcb0497..d0b3aaa2a0e141ab2e32ec640e98cad2f02a19aa 100644 --- a/consumer_complaints/versions/019d738b-4c6e-751f-972a-5f63b1508f70/dataset_metadata.dataset-mold-v1.json +++ b/consumer_complaints/versions/019d738b-4c6e-751f-972a-5f63b1508f70/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nThe dataset on Kaggle (https://www.kaggle.com/datasets/selener/consumer-complaint-database) has data from up until 2019. We use the newest version from the government website with data up until 2025.\n\n- Context and descriptions of the features can be found here: https://cfpb.github.io/api/ccdb/fields.html\n- \"Company public response\" is not free text but selected \"from a set list of options\".\n- Following TexTabBench, we focus on predicting the type of closure a complaint got, that is the \"Company response to consumer\" column. We want to predict if a complaint will be closed with just an explanation, with non-monetary relief, or with monetary relief.\n- We filter the data to only include entries after consumer disputations were discontinued as this represent a shift in protocol. This filters all data before April 24th 2017.\n- We filter all cases where the \"Consumer consent provided?\" is in progress or got an untimely response.\n- We filter all rows that do not include consent to share their complaint narrative. This ensures the data contains text sentences.\n- We drop \"Company public response\" as it leaks the target variable.\n- We only allow complaints from US states (no territories or international complaints) and remove cases with missing states.\n- The data has unresolved spatial information in the ZIP code. Some ZIPs are censored (ending in \"XXX\" or full removed \"XXXXXX\").\n- We add a feature for \"Low population area\", which determines that the ZIP is censored.\n- The tags filed contains only three non-nan labels, of which one is a duplicate of the others. We created two categorical features from it instead.\n- We drop duplicates (2% of the data) as the data should not contain naturally occurring duplicates and this likely results from some overlap in data collection or data entry or faulty re-submissions. We investigated some of the duplicates and they appear to be identical complaints. There exist duplicates with different target labels, we also drop these as we have no way to determine what the correct label is.\n- We drop constant columns, the complaint ID, and when the complaint was send to the company (as it does not related to the target task)\n", "version_from_unique_name": "consumer_complaints", "version_comment": "\nWe use the last 3 months as test data randomly sub-sample the train to 1 million and test data 250k rows. We follow TabReD and use random sub-sampling. The idea behind this instead of a time-based subsampling is to keep data from various time periods and model the distribution shift across the full time horizon.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/cooking_time/versions/019d737d-b4b1-7c24-8930-70a2c242b3f9/dataset_metadata.dataset-mold-v1.json b/cooking_time/versions/019d737d-b4b1-7c24-8930-70a2c242b3f9/dataset_metadata.dataset-mold-v1.json index 0f6674f76fde1a7077d150087f98cb43a762a053..f0654cb238b30bd8da11e805b9ed44f35649f84b 100644 --- a/cooking_time/versions/019d737d-b4b1-7c24-8930-70a2c242b3f9/dataset_metadata.dataset-mold-v1.json +++ b/cooking_time/versions/019d737d-b4b1-7c24-8930-70a2c242b3f9/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with data from TabRed, which already comes preprocessed.\n", "version_from_unique_name": "cooking_time", "version_comment": "\nWe randomly sub-sample the train to 1 million and test data 250k rows. We follow TabReD and use random sub-sampling. The idea behind this instead of a time-based subsampling is to keep data from various time periods and model the distribution shift across the full time horizon.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/covertype/019d7391-8e33-757b-b38a-79ec188fa001/dataset_metadata.dataset-mold-v1.json b/covertype/019d7391-8e33-757b-b38a-79ec188fa001/dataset_metadata.dataset-mold-v1.json index 3e34d332211483ec9aad84c06caee484da555e88..7cb470298c91152822cb9b9147073e59ca6b5114 100644 --- a/covertype/019d7391-8e33-757b-b38a-79ec188fa001/dataset_metadata.dataset-mold-v1.json +++ b/covertype/019d7391-8e33-757b-b38a-79ec188fa001/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\nWe create a special version of the dataset to avoid leakage. This version consists only of three 3 classes and 3 (spatial) wilderness areas. The investigation of Covertype shows that the dataset comprises multiple wilderness areas (Rawah, Comanche Peak, Neota, Cache la Poudre) that can be treated as subgroups but are not strictly IID, with wilderness area and soil type encoded as one-hot categorical features. Although TabRed argues for a time split to reflect a real task, the dataset lacks both explicit time and precise spatial (e.g., GNSS) features, leaving only area identifiers, which implicitly encode collection time and location; consequently, IID splits would introduce temporal or spatial leakage. Additional issues include transformed features in the OpenML/TALENT version that leak test distribution, incorrect column order in the UCI release, and evidence from EDA that class distributions differ across areas, confirming the dataset\u2019s grouped nature where cover type characteristics strongly depend on area. The only robust strategy is a spatially motivated grouped split by area; however, because some classes appear only in specific areas, the dataset is restricted to classes present in the three largest areas and evaluated using leave-one-area-out grouped splits.\n\nOther steps:\n- We add the column names in the correct way.\n- We reverse the one-hot encoding of the Wilderness_Area and Soil_Type features, and add human-readable descriptions for these features. Moreover, we add the climatic and geologic zones categorical features for each soil type based on the codes.\n- We reverse the ordinal encoding of the class name.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/credit_approval/019d736a-a27c-7c85-90c5-ca1d5dc849bd/dataset_metadata.dataset-mold-v1.json b/credit_approval/019d736a-a27c-7c85-90c5-ca1d5dc849bd/dataset_metadata.dataset-mold-v1.json index 3cc56c60e81c178d1b0dad7cbfc0c8314cdbbcb9..d31820187ba8365e1dec14a01f1b9e0d319d0594 100644 --- a/credit_approval/019d736a-a27c-7c85-90c5-ca1d5dc849bd/dataset_metadata.dataset-mold-v1.json +++ b/credit_approval/019d736a-a27c-7c85-90c5-ca1d5dc849bd/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe get the data from UCI.\n\n- We encode A14 as numeric, following the description of it being a continuous feature.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/credit_card_clients_default/019d7367-23b0-7125-9a7b-9ad1a90dc75d/dataset_metadata.dataset-mold-v1.json b/credit_card_clients_default/019d7367-23b0-7125-9a7b-9ad1a90dc75d/dataset_metadata.dataset-mold-v1.json index 556ab935daddaf8d2cbc551447c40e6b73379ac6..d88d2b9c21b85e038884c730ceec2fe66c6d0eb3 100644 --- a/credit_card_clients_default/019d7367-23b0-7125-9a7b-9ad1a90dc75d/dataset_metadata.dataset-mold-v1.json +++ b/credit_card_clients_default/019d7367-23b0-7125-9a7b-9ad1a90dc75d/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n - We rename the target variable and restore the original class names.\n - We drop the \"ID\" column.\n - Anomaly: the data has temporal features but the task is time-invariant.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/credit_g/019d7367-39a7-7335-8448-2dc9b2872361/dataset_metadata.dataset-mold-v1.json b/credit_g/019d7367-39a7-7335-8448-2dc9b2872361/dataset_metadata.dataset-mold-v1.json index c8ed289139781d013f5d6d4d9e6c62ccd30b540c..fec230f9d3e78c5e946ffb06aa51da0625b2cfcb 100644 --- a/credit_g/019d7367-39a7-7335-8448-2dc9b2872361/dataset_metadata.dataset-mold-v1.json +++ b/credit_g/019d7367-39a7-7335-8448-2dc9b2872361/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We reversed the original ordinal encoding.\n- Anomaly: the original task used a cost matrix for evaluation.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/customer_satisfaction_in_airline/019d7367-4d56-7274-a470-1609f3a3d6fe/dataset_metadata.dataset-mold-v1.json b/customer_satisfaction_in_airline/019d7367-4d56-7274-a470-1609f3a3d6fe/dataset_metadata.dataset-mold-v1.json index 32c301b62e478f849d79e643fdea1870b19c0d98..c19ad34201e31c8022a604ccc3c3003d4a42053d 100644 --- a/customer_satisfaction_in_airline/019d7367-4d56-7274-a470-1609f3a3d6fe/dataset_metadata.dataset-mold-v1.json +++ b/customer_satisfaction_in_airline/019d7367-4d56-7274-a470-1609f3a3d6fe/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the target column from \"satisfaction\" to \"satisfied\" for clarity and mapped the values to \"Yes\" and \"No\"\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/delivery_eta/versions/019d7382-df54-7dd3-8198-0567d7499858/dataset_metadata.dataset-mold-v1.json b/delivery_eta/versions/019d7382-df54-7dd3-8198-0567d7499858/dataset_metadata.dataset-mold-v1.json index 2988748e73f7e0a5dd1b93767f7556439dd0d94e..f88e79e5c0b6a9b8af2f683c86b1b5237275f777 100644 --- a/delivery_eta/versions/019d7382-df54-7dd3-8198-0567d7499858/dataset_metadata.dataset-mold-v1.json +++ b/delivery_eta/versions/019d7382-df54-7dd3-8198-0567d7499858/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with data from TabRed, which already comes preprocessed.\n\n- We drop a duplicated column `cat_2` in the preprocessing.\n", "version_from_unique_name": "delivery_eta", "version_comment": "\nWe randomly sub-sample the train to 1 million and test data 250k rows. We follow TabReD and use random sub-sampling. The idea behind this instead of a time-based subsampling is to keep data from various time periods and model the distribution shift across the full time horizon.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/dementia_prediction/019d7391-61e5-7cca-9a66-e85718233be9/dataset_metadata.dataset-mold-v1.json b/dementia_prediction/019d7391-61e5-7cca-9a66-e85718233be9/dataset_metadata.dataset-mold-v1.json index 98403f13b8ce1164446add06947a406b5f0e6982..70ecb4db7190ae9ac630c18ad93d447befc313ff 100644 --- a/dementia_prediction/019d7391-61e5-7cca-9a66-e85718233be9/dataset_metadata.dataset-mold-v1.json +++ b/dementia_prediction/019d7391-61e5-7cca-9a66-e85718233be9/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with the data from Mendeley.\n\n- CDR and Group are both the same target variable. CDR is the Clinical Dementia Rating (0 = no dementia, 0.5 = very mild AD, 1 = mild AD, 2 = moderate AD), which determines the group variable. The cases for converted / changed their dementia rating over time. We make this a task to predict for a patient (at any given time point of a scan) their dementia rating. The rating is ordinal but discrete, so we treat it as a classification problem (not regression). We drop cases with moderate_AD (n=3), as we do not have enough data on this class to include them in our prediction task.\n- We drop the group variable and any information about the time of the scan as our goal to predict the rating from eTIV, nWBV, and ASF which are all derived from the MRI scan and independent of time.\n- We also drop the constant hand column.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/diabetes_130_us/019d7367-64e4-7883-956d-9ba2fc8db41b/dataset_metadata.dataset-mold-v1.json b/diabetes_130_us/019d7367-64e4-7883-956d-9ba2fc8db41b/dataset_metadata.dataset-mold-v1.json index e7a7f6e4498dfc89b94081303a0228992fa5c857..05c1832de9193951a345b8de875e04f82b696d32 100644 --- a/diabetes_130_us/019d7367-64e4-7883-956d-9ba2fc8db41b/dataset_metadata.dataset-mold-v1.json +++ b/diabetes_130_us/019d7367-64e4-7883-956d-9ba2fc8db41b/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We drop duplicated patients based on the \"patient_nbr\" feature to avoid target leakage.\n- We reversed the original ordinal encoding for three ID-based features (admission_type_id, discharge_disposition_id, admission_source_id).\n- We created the target from the \"readmitted\" column following the original task description: \"<30\" becomes \"Yes\", everything else \"No\".\n- We dropped \"encounter_id\" and \"patient_nbr\", which are both unique identifiers for each row.\n- We keep original \"?\", NULL-codes, and NaN values because they exist in different ways across the columns.\n- Anomaly: There is a distribution shift based on the original order. The reason for this might be that the encounters are ordered in some way such that later parts of the data contain different sub-groups than earlier parts. This is also indicated by the fact that the \"payer_code\" feature is responsible for the shift. This distribution shift vanishes after randomly shuffling the data (as done by default for this and all other datasets used in TabArena).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/diamonds/019d7367-779c-707e-b777-849180166b74/dataset_metadata.dataset-mold-v1.json b/diamonds/019d7367-779c-707e-b777-849180166b74/dataset_metadata.dataset-mold-v1.json index 4a7b2233f1699927d70dc2aed39708219846e5fa..a0824acc2925552357d16d42eb29e91e785ee5cd 100644 --- a/diamonds/019d7367-779c-707e-b777-849180166b74/dataset_metadata.dataset-mold-v1.json +++ b/diamonds/019d7367-779c-707e-b777-849180166b74/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- Unlike TabArena, we log scale the target as it is a price and hence log scaling is generally recommended.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/drug_induced_autoimmunity_prediction/019d736a-b4b1-7f51-9563-dde2b6ecd948/dataset_metadata.dataset-mold-v1.json b/drug_induced_autoimmunity_prediction/019d736a-b4b1-7f51-9563-dde2b6ecd948/dataset_metadata.dataset-mold-v1.json index a37190ac94341b0cb976a82f77f31b86a793ff6b..3f6476ed170cbaaafd0c12cefa05120c0960bd03 100644 --- a/drug_induced_autoimmunity_prediction/019d736a-b4b1-7f51-9563-dde2b6ecd948/dataset_metadata.dataset-mold-v1.json +++ b/drug_induced_autoimmunity_prediction/019d736a-b4b1-7f51-9563-dde2b6ecd948/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the dataset from UCI.\n\n- We drop the constant columns.\n- We keep the SMILES code as string for later pipelines to handle.\n\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/early_learning_predictors/019db4ff-b5e9-7b27-80b1-8cf2ca025d20/dataset_metadata.dataset-mold-v1.json b/early_learning_predictors/019db4ff-b5e9-7b27-80b1-8cf2ca025d20/dataset_metadata.dataset-mold-v1.json index 53149d51e22c2f4b55736786841d671314b6ddbc..2071de1d14c971c867dabe0a5528c0dbeb262aca 100644 --- a/early_learning_predictors/019db4ff-b5e9-7b27-80b1-8cf2ca025d20/dataset_metadata.dataset-mold-v1.json +++ b/early_learning_predictors/019db4ff-b5e9-7b27-80b1-8cf2ca025d20/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- The data is collected for predictive analytics to find major drivers of early learning outcomes. Nevertheless, predictive performance matters and the data could also be used for identifying at risk children.\n- The data consists of different subsets that were collected at different times. Sometimes the target was even collected before the predictors. Although the data is collected over time, the task is cross-sectional. Therefore, we drop most date features, after checking whether relevant information was already extracted.\n- We use a grouped split based on id_facility to reflect a potential deployment scenario where we want to predict for a new facility. Note that a date (year) is available, but the data collection is conducted per facility so we don't use a time-based split.\n- The target score ELOM is continuous, but decisions are made based on cutoffs to identify at-risk children. We keep the task as regression, to maintain the most information.\n- Note: The data consists of different age groups, for which the ELOM score is computed differently. While under these circumstances a multi-class target using the given cutoffs might be a valid choice, the age is given as a feature and therefore, we assume that a model can be expected to learn patterns for subgroups.\n- The data is collected from several sources, including direct assessments, teacher questionnaires, parent surveys and facility data. The data is collected by different enumerators across different facilities. The target is from a separate tool (see further information below).\n- We drop all columns which are components or alternative versions of the target, to avoid leakage.\n- We drop all samples with a missing target.\n- Due to most samples having missing values, we drop all \"HOME LEARNING ENVIRONMENT DATA\" features. This mostly aligns with the Zindi challenge, except that the features hle_ind, quintile_used, and ses_proxy were kept. We keep the latter two, because they are not missing for all samples and are derived from other; not-HLE features.\n- We drop all features with 100% missing values and one unique value.\n- We drop all indicator columns.\n- We drop some repeated columns, such as age_group_label.\n- Because all features could generally be available at prediction time, even if would not be ideal effort-wise, we keep all features that don't leak the target or are otherwise problematic.\n- We transform all date columns to datetime.\n- Additional description of the target: The ELOM 4 & 5 Years Assessment Tool measures whether preschool children are on track for their age in key areas of development. It is a standardised tool that measures performance across five key developmental domains for children aged 50 to 59 months and 60 to 69 months. The scoring assesses a child in 23 items, across five domains, namely: gross motor development, fine motor development, emergent numeracy and mathematics, cognition and executive functioning, and emergent literacy and language. These five domains form part of the direct assessment.\n- Note: We keep some features that were not in the zindi challenge, such as the teacher and parent questionnaires, because they are potentially predictive and could be available at prediction time. Moreover, some columns present in the zindi challenge (like latitude and longitude) are not available.\n- Note: The data consists of a few subsets with partially distinct features collected. This might not represent a real task, but we don't know for sure, so we keep all samples.\n- Note: The enumerator_id is a control variable of the data collection rather than a factor that should be used for prediction. However, it is predictive and explains noise in the target, so we keep it as a feature.\n- Note: There are strong patterns of missingness within the data, due to different uses of various tools and measurements across subgroups\n- Anomaly: Many missing values\n- Anomaly: This is a highly underspecified task, leaving a lot of data cleaning to the user/model.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/early_stage_diabetes_risk_prediction/019d736a-c5ee-7069-9b14-686535f9caa9/dataset_metadata.dataset-mold-v1.json b/early_stage_diabetes_risk_prediction/019d736a-c5ee-7069-9b14-686535f9caa9/dataset_metadata.dataset-mold-v1.json index 8a0473cd4a0e2eeb9c958c42df626bfcc98ede0f..9045dadd4c4fdd06b882829b15dbbd20bb26872e 100644 --- a/early_stage_diabetes_risk_prediction/019d736a-c5ee-7069-9b14-686535f9caa9/dataset_metadata.dataset-mold-v1.json +++ b/early_stage_diabetes_risk_prediction/019d736a-c5ee-7069-9b14-686535f9caa9/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe use the data as is from UCI.\n\n- Note, the dataset contains a lot of naturally occurring duplicates (50%). We drop them to avoid too extreme duplicated-based data leakage biasing the evaluation.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/ecoli_proteins/019d736a-d6d2-7e28-aeff-7da6b7126b9d/dataset_metadata.dataset-mold-v1.json b/ecoli_proteins/019d736a-d6d2-7e28-aeff-7da6b7126b9d/dataset_metadata.dataset-mold-v1.json index b14edba2f937fc967946e5055a892f922fc45e76..663ba6b9e576a49d25e75a30246dc5de06725a77 100644 --- a/ecoli_proteins/019d736a-d6d2-7e28-aeff-7da6b7126b9d/dataset_metadata.dataset-mold-v1.json +++ b/ecoli_proteins/019d736a-d6d2-7e28-aeff-7da6b7126b9d/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the dataset from UCI.\n\n- We drop the sequence name column, which is a unique identifier.\n- We drop three classes with less than 10 samples each: omL (5), imL (2), and imS (3).\n- We drop \"chg\", which becomes constant after our preprocessing steps.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/ecommerce_shipping/019d7367-8a04-7bf2-8ab2-68d1726cf31e/dataset_metadata.dataset-mold-v1.json b/ecommerce_shipping/019d7367-8a04-7bf2-8ab2-68d1726cf31e/dataset_metadata.dataset-mold-v1.json index 401b32764244286cbd85836b5f101a848b61f43d..c9fb246b4a4782dbf865f0a410d64ff491001485 100644 --- a/ecommerce_shipping/019d7367-8a04-7bf2-8ab2-68d1726cf31e/dataset_metadata.dataset-mold-v1.json +++ b/ecommerce_shipping/019d7367-8a04-7bf2-8ab2-68d1726cf31e/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We dropped the ID column.\n- We renamed the target feature \"Reached.on.Time_Y.N\" to \"ArrivedLate\" and mapped binary values to \"Yes\"/\"No\".\n- Anomaly: the target and task seems somewhat disconnected from the features. Moreover, some source information on the data is missing and there might be some translation issues.\n- Anomaly: there might be some data issues related to \"Warehouse_block\" and the value \"F\" consisting of two block \"E\" and \"F\".\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/electric_motor_temperature_prediction/019d7392-e21f-7e60-9efb-2141c3513fd9/dataset_metadata.dataset-mold-v1.json b/electric_motor_temperature_prediction/019d7392-e21f-7e60-9efb-2141c3513fd9/dataset_metadata.dataset-mold-v1.json index b5ce865c71a1aee72cc7bf51bb515eaf263c46cd..166f4645ae9e709bba9aabf6fedb18a458c5177b 100644 --- a/electric_motor_temperature_prediction/019d7392-e21f-7e60-9efb-2141c3513fd9/dataset_metadata.dataset-mold-v1.json +++ b/electric_motor_temperature_prediction/019d7392-e21f-7e60-9efb-2141c3513fd9/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe select the task to predict permanent magnet temperature from the input features. We create grouped splits on the profile_id, which corresponds to predicting the temperature for an unseen motor run session, thus simulating how the model would be used in reality.\n\n- The original study also incorporates temporal connection within the session. We re-create a time-index (https://www.kaggle.com/datasets/wkirgsn/electric-motor-temperature/discussion/147446).\n- We also add the derived inputs used in the original study.\n- We also add the EWMA/WEMS features with window sizes of [500, 2000, 4000, 8000]. To avoid the problem of having an incorrect EWMA/WEMS state and since not all recording are warmed up (https://www.kaggle.com/datasets/wkirgsn/electric-motor-temperature/discussion/117319), we drop the first 500 samples (first span) of each profile to simulate such a warm-up.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/emscad/019d736a-ed4a-7aaa-89df-d52558565db9/dataset_metadata.dataset-mold-v1.json b/emscad/019d736a-ed4a-7aaa-89df-d52558565db9/dataset_metadata.dataset-mold-v1.json index a64b5e90f77deef13124c5fc73e3896e5e32d229..ecd3ff9ea07df496095eaa59a7c71c27c523f47f 100644 --- a/emscad/019d736a-ed4a-7aaa-89df-d52558565db9/dataset_metadata.dataset-mold-v1.json +++ b/emscad/019d736a-ed4a-7aaa-89df-d52558565db9/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe have no access to the original state and the data already contains some preprocessed features.\nThe paper describes them in more detail as well in Table 2.\n\nIn the original paper, the authors subsampled the data to have 450 fraudulent and 450 non-fraudulent samples.\nNo details are given on how the subsampling was done. Moreover, duplicates were skipped, but no details are given on how duplicates were identified.\n\n- The dataset was collected from 2012 to 2014. But the data contains no feature to identify the time of a sample. So we can expect that some bias from temporal leakage. Moreover, the data might contain multiple job postings from the same fraudster, which would need to be grouped, but we cannot identify them from the data.\n- Following the original paper, we drop all duplicates (when ignoring the job_id column) to avoid data leakage.\n- The data has a location description, we do not resolve it lat/longitude but leave it to the pipelines.\n- The salary range has several weird quirks. It contains data either in the thousands, or is missing the \"k\" to denote thousands. Moreover, it contains empty or 0-0 entries. The column might contain yearly salary, hourly salary, or one-time payment. Finally, for some jobs, the column contains a date (e.g. \"Oct-20\"). We keep the column as complex as it is. We add one column that contains the maximum salary parsed from the text and we only keep salary values above 50k as valid values to create a numerical feature for differences in the larger ranges.\n- We removed job listings with non english texts (138 rows).\n- We do not subsample the data, as a result, the data is heavily imbalanced.\n- We found no relation of job_id with a timestamp and even found cases where a lower job_id has a description claiming to be a job from 2014 (so the end of the collection period). Thus, we do dropped job_id as it does not contain any signal.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/eryhemato_squamous_disease/019d736b-020a-7fbc-9785-6438802781f2/dataset_metadata.dataset-mold-v1.json b/eryhemato_squamous_disease/019d736b-020a-7fbc-9785-6438802781f2/dataset_metadata.dataset-mold-v1.json index 582e2460067f10fe8645d3287286147637386136..015ac114cc00aafeb348040877f065c9fd0b6ec2 100644 --- a/eryhemato_squamous_disease/019d736b-020a-7fbc-9785-6438802781f2/dataset_metadata.dataset-mold-v1.json +++ b/eryhemato_squamous_disease/019d736b-020a-7fbc-9785-6438802781f2/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- We encode all features but age as categorical, since they are ordinal features in nature.\n- We ensure missing values in age are encoded as NaN.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/fiat_500/019d7367-9b82-78a9-a53a-570038a28d0a/dataset_metadata.dataset-mold-v1.json b/fiat_500/019d7367-9b82-78a9-a53a-570038a28d0a/dataset_metadata.dataset-mold-v1.json index 6c479b241d825d0ed6ce7fbc398e8f8f329afac1..7f1ca7fe7b257cb72c83eb149d4c0af47978c3e0 100644 --- a/fiat_500/019d7367-9b82-78a9-a53a-570038a28d0a/dataset_metadata.dataset-mold-v1.json +++ b/fiat_500/019d7367-9b82-78a9-a53a-570038a28d0a/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "N/A", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/fitness_club/019d7367-aced-7960-94c9-b19b393cb0ad/dataset_metadata.dataset-mold-v1.json b/fitness_club/019d7367-aced-7960-94c9-b19b393cb0ad/dataset_metadata.dataset-mold-v1.json index a6978db6bdb1f234c502d4acecf6e45dfcd1eec5..c4ad63558c7567ad13f03a74701774e050cc1891 100644 --- a/fitness_club/019d7367-aced-7960-94c9-b19b393cb0ad/dataset_metadata.dataset-mold-v1.json +++ b/fitness_club/019d7367-aced-7960-94c9-b19b393cb0ad/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We dropped the booking_id column.\n- We renamed the values of the target variable to be more meaningful (\"Yes\"/\"No\").\n- We treat \"-0\" values as \"0\" values for the target, following the description.\n- We removed trailing words (like \"days\") from \"days_before\".\n- We aligned the naming of \"day_of_week\" to be consistent per day.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/food_delivery_time/019d7367-beff-7ca7-9033-8f346962f7a1/dataset_metadata.dataset-mold-v1.json b/food_delivery_time/019d7367-beff-7ca7-9033-8f346962f7a1/dataset_metadata.dataset-mold-v1.json index 7947c9827ce0626900a1ef11b6bf096bcc484721..2a6aeba11102aa120dfa7f065a26d5de8220fbc5 100644 --- a/food_delivery_time/019d7367-beff-7ca7-9033-8f346962f7a1/dataset_metadata.dataset-mold-v1.json +++ b/food_delivery_time/019d7367-beff-7ca7-9033-8f346962f7a1/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We dropped entries with a duplicated ID, keeping only the first one.\n- We dropped the ID column.\n- Anomaly: The ID of the delivery person is given as a feature. In some contexts, this feature might not be allowed to use. Moreover, using this information might require considering a temporal split where a cold-start problem is covered.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/forensic_glass_identification/019d736b-12da-73ae-a0d2-65e6294a7f01/dataset_metadata.dataset-mold-v1.json b/forensic_glass_identification/019d736b-12da-73ae-a0d2-65e6294a7f01/dataset_metadata.dataset-mold-v1.json index c62b3057feea8223de806917ab0dd3a6dbf381f2..7b6482f9bb071f0ae4b88a3e4f65d498b2bb8304 100644 --- a/forensic_glass_identification/019d736b-12da-73ae-a0d2-65e6294a7f01/dataset_metadata.dataset-mold-v1.json +++ b/forensic_glass_identification/019d736b-12da-73ae-a0d2-65e6294a7f01/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- We drop the ID column, as it is a just a row identifier and does not contain any useful information for modeling.\n- The data contains one naturally occurring duplicate.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/forest_fires/019d736b-2610-7f4f-ae46-3f6a4c98962e/dataset_metadata.dataset-mold-v1.json b/forest_fires/019d736b-2610-7f4f-ae46-3f6a4c98962e/dataset_metadata.dataset-mold-v1.json index c1788845e7fc039cc291f4b6872c4521cbcf7dd0..ccec848893d6462cafad8271d3078e0b548d02e0 100644 --- a/forest_fires/019d736b-2610-7f4f-ae46-3f6a4c98962e/dataset_metadata.dataset-mold-v1.json +++ b/forest_fires/019d736b-2610-7f4f-ae46-3f6a4c98962e/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- The original paper used a IID split. The tasks sounds like it requires a temporal split. However, the data does not contain the original time indices (or order of samples) and just month/day over 3 years. Furthermore, the authors made sure to only use time-invariant features as it seems they wanted to develop a time-invaraint model for deployment and analysis of the data. In other words, this task can be handled by IID splits. Furthermore, the task in the paper was about using algorithms also for feature selection (so scientific discovery).\n- We log1p scale the target as suggest by the authors.\n- The data contains naturally occurring duplicates.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/gallstone_disease/019d736b-3795-7523-ac59-9fbe676ebc70/dataset_metadata.dataset-mold-v1.json b/gallstone_disease/019d736b-3795-7523-ac59-9fbe676ebc70/dataset_metadata.dataset-mold-v1.json index 4dd3489a4dad762b91a5c0e0fb307e9475ebb56b..40fe31a34be5c455162f233e2ece984a94e8041a 100644 --- a/gallstone_disease/019d736b-3795-7523-ac59-9fbe676ebc70/dataset_metadata.dataset-mold-v1.json +++ b/gallstone_disease/019d736b-3795-7523-ac59-9fbe676ebc70/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe use the data without any further curation.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/garments_worker_productivity/019db501-f4c4-7de2-b626-fcb769c7d82e/dataset_metadata.dataset-mold-v1.json b/garments_worker_productivity/019db501-f4c4-7de2-b626-fcb769c7d82e/dataset_metadata.dataset-mold-v1.json index 267f425a0bd76d415cf9d24c54d31d74fd7b22af..03795a5eeaf82c3745f65bee27d1291f75a517a7 100644 --- a/garments_worker_productivity/019db501-f4c4-7de2-b626-fcb769c7d82e/dataset_metadata.dataset-mold-v1.json +++ b/garments_worker_productivity/019db501-f4c4-7de2-b626-fcb769c7d82e/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- We fix typos in data entries (e.g., \"finishing \" becomes \"finishing\").\n- We transform the date to datetime.\n- The associated paper conceptualizes the task as an interpretable ML task without consideration of realistic predictive scenarios. Therefore, we define a new predictive ML task.\n- We define the decision point in time as the start of the day.\n- We set the target variable to \"actual_productivity\", and use the targeted_productivity as an input feature.\n- We do not include all features directly for prediction, because some can't be expected to be available at the decision point in time: \"smv\", \"wip\", \"over_time\", \"idle_time\", \"idle_men\". To still keep as much information as possible, we impute the value of the previous recorded working day (by department and team). \n- Because the timestamps are irregularly spaced, we additionally include the days passed sind the last recording per department and team.\n- Note that we try to keep feature engineering minimalistic and only with the sole purpose to prevent leaks with minimal loss of information.\n- It can be expected that feature engineering exposing the non-iid properties of the dataset will be crucial for good predictive performance.\n- We cannot fully exclude the possibility of leaks, since not enough information about the date of collection for some features is given.\n- We define 30 splits with one day for testing each, resulting in small test sizes of 17-24 samples per split.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/ghanas_indigenous_intel/019dd375-4259-77c1-a177-5acc1f4d0134/dataset_metadata.dataset-mold-v1.json b/ghanas_indigenous_intel/019dd375-4259-77c1-a177-5acc1f4d0134/dataset_metadata.dataset-mold-v1.json index c541cfc002c4cc7406c191a2fcdcdc4b9db36537..a0f212bc8e0fd517bee1bcd2233b31cb89460f40 100644 --- a/ghanas_indigenous_intel/019dd375-4259-77c1-a177-5acc1f4d0134/dataset_metadata.dataset-mold-v1.json +++ b/ghanas_indigenous_intel/019dd375-4259-77c1-a177-5acc1f4d0134/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\n- Information about the data source: The data was collected by local farmers in the Pra River Basin of Ghana using the Smart Indigenous Weather App, a custom-built mobile app for data collection. Farmers made weather forecasts based on traditional signs such as cloud formations, sun position, wind, moon, heat, and specific tree or animal behaviors. The task is to predict the farmers' forecasts based on these traditional signs, which are represented as features in the dataset. The target variable is the actual amount of rain in four categories, therefore the actual task is to make a prediction correcting the farmers' forecasts.\n- The data was initially used in a Zindi competition with a temporal split, we do the same.\n- We create 6 splits using a sliding window approach with a window size of 3 days and a step size of 2 days. This means that one day overlaps among consecutive test splits.\n- We ensure that each split uses at least half of the available samples for training.\n- We drop the ID column.\n- We convert prediction_time to datetime and sort by it.\n- We rename the target column to \"rainfall\" for better clarity.\n- We transform object columns and user_id to categorical.\n\n- Note: The split that would be ideal for the task would be grouped+temporal. However, the available data is not enough for this scenario. In general the available data size reduces the task quality a lot.\n- Anomaly: For this dataset it is possible that test folds don't contain some classes.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/give_me_some_credit/019d7367-d331-7295-991e-7eab46ea49ba/dataset_metadata.dataset-mold-v1.json b/give_me_some_credit/019d7367-d331-7295-991e-7eab46ea49ba/dataset_metadata.dataset-mold-v1.json index 1bc65e6c15e7368d9acbe50fca7f09fc5a5848dc..0820def042d51acacbdeec939950f1b2cb0dc588 100644 --- a/give_me_some_credit/019d7367-d331-7295-991e-7eab46ea49ba/dataset_metadata.dataset-mold-v1.json +++ b/give_me_some_credit/019d7367-d331-7295-991e-7eab46ea49ba/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the target feature and its value to be more descriptive.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/hazelnut_spread_contaminant_detection/019d7367-e731-79fc-9677-4c5d1bab73ff/dataset_metadata.dataset-mold-v1.json b/hazelnut_spread_contaminant_detection/019d7367-e731-79fc-9677-4c5d1bab73ff/dataset_metadata.dataset-mold-v1.json index 542f4a6a007517dfda51e031ec85c2ea32e49133..544e1fb3b480e7d43cb36761485f1371aa9909d0 100644 --- a/hazelnut_spread_contaminant_detection/019d7367-e731-79fc-9677-4c5d1bab73ff/dataset_metadata.dataset-mold-v1.json +++ b/hazelnut_spread_contaminant_detection/019d7367-e731-79fc-9677-4c5d1bab73ff/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We select a features from a single frequency (10 GHz) as the authors also only considered this frequency for the final experiments.\n- Anomaly: we use the publicly available dataset state, which is without preprocessing. Moreover, we were not able to inverse the original ordinal encoding of the label.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/healthcare_insurance_expenses/019d7367-f8cf-77a8-86e8-cabed951a9f6/dataset_metadata.dataset-mold-v1.json b/healthcare_insurance_expenses/019d7367-f8cf-77a8-86e8-cabed951a9f6/dataset_metadata.dataset-mold-v1.json index 6fc3d46a9b69a9a310458d9152787a875ec919bc..ac836a8dc5f2a30a8bbced32e5fedcec922d0498 100644 --- a/healthcare_insurance_expenses/019d7367-f8cf-77a8-86e8-cabed951a9f6/dataset_metadata.dataset-mold-v1.json +++ b/healthcare_insurance_expenses/019d7367-f8cf-77a8-86e8-cabed951a9f6/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- Unlike in TabArena, we log scale the target as the distribution is very skewed. This is a common practice for price-related targets.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/heart_disease_cleveland/019d736b-4900-714b-8dfc-be588ce802c3/dataset_metadata.dataset-mold-v1.json b/heart_disease_cleveland/019d736b-4900-714b-8dfc-be588ce802c3/dataset_metadata.dataset-mold-v1.json index 27d6a650f6f3c0c1463531ada0042713bfd77a73..19e5d21da017d5ed6661651095a5966297989e08 100644 --- a/heart_disease_cleveland/019d736b-4900-714b-8dfc-be588ce802c3/dataset_metadata.dataset-mold-v1.json +++ b/heart_disease_cleveland/019d736b-4900-714b-8dfc-be588ce802c3/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the processed version and the subset of 14 attributes used in the study and clinical practice.\n\n- We encode missing values as np.nan instead of \"?\".\n- We make the target binary (0=no heart disease, 1=heart disease). This follows the original study in attempting to distinguish presence (values 1,2,3,4) from absence (value 0).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/heart_disease_hungary/019d736b-5a1a-7c95-92f0-8c2fe5334ad2/dataset_metadata.dataset-mold-v1.json b/heart_disease_hungary/019d736b-5a1a-7c95-92f0-8c2fe5334ad2/dataset_metadata.dataset-mold-v1.json index c777499325ee2d336e330ce6326b74bdff2e49c8..b93393bb41b708d5bdcefae181534a23dfe1c7de 100644 --- a/heart_disease_hungary/019d736b-5a1a-7c95-92f0-8c2fe5334ad2/dataset_metadata.dataset-mold-v1.json +++ b/heart_disease_hungary/019d736b-5a1a-7c95-92f0-8c2fe5334ad2/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the processed version and the subset of 14 attributes used in the study and clinical practice.\n\n- We encode missing values as np.nan instead of \"?\".\n- We make the target binary (0=no heart disease, 1=heart disease). This follows the original study in attempting to distinguish presence (values 1,2,3,4) from absence (value 0).\n- The data has one naturally occurring duplicate, which we do not drop.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/heart_disease_va_long_beach/019d736b-7c92-71e1-a6d6-27a91ed15d3d/dataset_metadata.dataset-mold-v1.json b/heart_disease_va_long_beach/019d736b-7c92-71e1-a6d6-27a91ed15d3d/dataset_metadata.dataset-mold-v1.json index 1fd5674f6b6002e9786f26b00695fa8d824e4505..19539a72fed7cf786fce43206ad2e12e86c7fb17 100644 --- a/heart_disease_va_long_beach/019d736b-7c92-71e1-a6d6-27a91ed15d3d/dataset_metadata.dataset-mold-v1.json +++ b/heart_disease_va_long_beach/019d736b-7c92-71e1-a6d6-27a91ed15d3d/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the processed version and the subset of 14 attributes used in the study and clinical practice.\n\n- We encode missing values as np.nan instead of \"?\".\n- We make the target binary (0=no heart disease, 1=heart disease). This follows the original study in attempting to distinguish presence (values 1,2,3,4) from absence (value 0).\n- The dataset has one naturally occurring duplicate, which we do not drop.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/heart_failure_followup_survival/019d736b-8d9f-7f20-8f5c-45207fb0251d/dataset_metadata.dataset-mold-v1.json b/heart_failure_followup_survival/019d736b-8d9f-7f20-8f5c-45207fb0251d/dataset_metadata.dataset-mold-v1.json index fc46442dfe3172f02088c5978152d6cf34ed3152..222d1758bb454c96d1ab5f14f8c7f30d5c50f364 100644 --- a/heart_failure_followup_survival/019d736b-8d9f-7f20-8f5c-45207fb0251d/dataset_metadata.dataset-mold-v1.json +++ b/heart_failure_followup_survival/019d736b-8d9f-7f20-8f5c-45207fb0251d/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe use the data as is from UCI.\n\n- We keep all features of the dataset. The study that introduced the dataset also curated a subset of features. We leave it to the pipeline to select the relevant features for the task.\n- There is a \"time\" feature in the data which is the follow-up time in days. The authors used this in some part of the experiments and showed it improved the predictive performance. We also keep this feature and follow their preprocessing to transform it into months (from days). Note, this does not make the data temporal, as the split is across unseen patients (not across time) and we only have one entry per patient.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/heloc/019d7368-0a52-7460-abf5-b78381cf69f0/dataset_metadata.dataset-mold-v1.json b/heloc/019d7368-0a52-7460-abf5-b78381cf69f0/dataset_metadata.dataset-mold-v1.json index e5643d61aeba0292348615bcf3f37db28ce3db66..96207161bc51607c8a23fc602e8783566c8faeb5 100644 --- a/heloc/019d7368-0a52-7460-abf5-b78381cf69f0/dataset_metadata.dataset-mold-v1.json +++ b/heloc/019d7368-0a52-7460-abf5-b78381cf69f0/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- Anomaly: the dataset has time-related features. However, the task and features are preprocessed to be time-invariant.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/hepatitis_c_prediction/019d736b-9eba-79b8-8c5a-78d95163e286/dataset_metadata.dataset-mold-v1.json b/hepatitis_c_prediction/019d736b-9eba-79b8-8c5a-78d95163e286/dataset_metadata.dataset-mold-v1.json index 5cc7944bfac9ac392431de2eed9e1d8f329c9940..3c026d778b7a87140f1b054a4019e2bb506488b5 100644 --- a/hepatitis_c_prediction/019d736b-9eba-79b8-8c5a-78d95163e286/dataset_metadata.dataset-mold-v1.json +++ b/hepatitis_c_prediction/019d736b-9eba-79b8-8c5a-78d95163e286/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- The hepatitis related classes were defined in the paper, but the blood donor class is not mentioned in it. It seems the paper might be related to a different task than the UCI release in the end. But it has the same features. Thus, after looking more into it, we adjust the classes. We drop the \"suspect Blood Donor\" class as we cannot map it to being part of a reasonable task. We thus treat the task as predicting whether a patient has one of three versions of hepatitis, or is \"just\" a blood donor.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/hepatitis_survival_prediction/019d736b-afda-7edf-9f73-3cd8e3799329/dataset_metadata.dataset-mold-v1.json b/hepatitis_survival_prediction/019d736b-afda-7edf-9f73-3cd8e3799329/dataset_metadata.dataset-mold-v1.json index f88503da79915484d71c353d51ebe3f7bd361c6c..9269be76cfc3e8e24b69d0769a6ca2dfea804e53 100644 --- a/hepatitis_survival_prediction/019d736b-afda-7edf-9f73-3cd8e3799329/dataset_metadata.dataset-mold-v1.json +++ b/hepatitis_survival_prediction/019d736b-afda-7edf-9f73-3cd8e3799329/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- We encode missing values as NaN.\n- We convert numeric features to float and categorical features to category dtype.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/hiva_agnostic/019d7368-2dd6-7aa8-b7f3-42a416dc1315/dataset_metadata.dataset-mold-v1.json b/hiva_agnostic/019d7368-2dd6-7aa8-b7f3-42a416dc1315/dataset_metadata.dataset-mold-v1.json index 903d11b7d51f96e2a4283a1c5984666b7818fd10..bdb2499cb689c8e7da5036ef562531bb311d0877 100644 --- a/hiva_agnostic/019d7368-2dd6-7aa8-b7f3-42a416dc1315/dataset_metadata.dataset-mold-v1.json +++ b/hiva_agnostic/019d7368-2dd6-7aa8-b7f3-42a416dc1315/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We take the original data with the original labels with three classes.\n- We only take the training data from the challenge, as the original labels are not available for the validation data.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/home_credit_default_risk/019d736d-6a16-7087-999b-e953d48f7516/dataset_metadata.dataset-mold-v1.json b/home_credit_default_risk/019d736d-6a16-7087-999b-e953d48f7516/dataset_metadata.dataset-mold-v1.json index 1a9a0bc3e8cf6d52b9fb0d4cd8ab88294a657183..82f93765c39fc76152b8325106efe239698e667a 100644 --- a/home_credit_default_risk/019d736d-6a16-7087-999b-e953d48f7516/dataset_metadata.dataset-mold-v1.json +++ b/home_credit_default_risk/019d736d-6a16-7087-999b-e953d48f7516/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the SQL tables from Kaggle. We then merge these tables into a single file following the top Kaggle solutions.\nWe start from this file.\n\n- The top solutions start with various features from public notebooks/kernels. When selecting the kernel to use as our starting point, we looked among the most popular and best performance on the leaderboard. We found https://www.kaggle.com/code/jsaguiar/lightgbm-with-simple-features?scriptVersionId=6025993 and its follow-up https://www.kaggle.com/code/ogrellier/lighgbm-with-selected-features. We use the later one, because it performance better and comes with a reduced set of features from the initial raw merges. Note, this may introduce a slight bias, as the feature selection was done for LGBM.\n- We do not follow the script exactly. We do not ordinal encode binary features. Moreover, we do not apply one-hot encoding to all categorical columns. Instead, we do it only when need for aggregations and otherwise keep the categorical column with the original semantics. Moreover, we fix some bugs in the original script that cause it to crash when run out of the box.\n- We drop a duplicated column after preprocessing.\n- We replace values that became inf after aggregation with NaN. This occurred across several aggregated columns.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/home_credit_default_stability/versions/019d7383-f45a-72f7-ac99-09447cf6d41f/dataset_metadata.dataset-mold-v1.json b/home_credit_default_stability/versions/019d7383-f45a-72f7-ac99-09447cf6d41f/dataset_metadata.dataset-mold-v1.json index 4c217a9521cd010f01b242aab1c43ac50b8f04c3..e409c1ab82c41773f535bb26b0a14f519d4cfbda 100644 --- a/home_credit_default_stability/versions/019d7383-f45a-72f7-ac99-09447cf6d41f/dataset_metadata.dataset-mold-v1.json +++ b/home_credit_default_stability/versions/019d7383-f45a-72f7-ac99-09447cf6d41f/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the data from Kaggle and follow the preprocessing from TabRed (https://github.com/yandex-research/tabred/tree/main/preprocessing#homecredit-default-stability-homecredit-20), which in turn follows two Kaggle solutions (https://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/507946, https://www.kaggle.com/code/yuuniekiri/fork-of-home-credit-catboost-inference).\n\n- We follow TabRed and use temporal splits for the task. The Kaggle experts used various strategies and most commonly StratifiedGroupKFold to align offline CV with the temporal-split data on the leaderboard. That is, the split used for training can be StratifiedGroupKFold, but not for testing.\n- Note, the original competition was heavily influenced by metric hacking, which is not relevant for our offline benchmark tasks. Thus, results are not directly comparable ot transferable.\n- We change the preprocessing from Kaggle and TabRed in specific steps: (A) we depart from the TabRed and Kaggle solutions in that we do not drop high-cardinality string columns, (B) we do not drop month and week, as it was only dropped due to the metric leak in the competition, (C) we only drop constant columns if they are constant w.r.t. NaN and non-NaN values, (D) we do not perform ordinal encoding and drop minor categories, (E) we do not sub-sample the data, (F) we keep the date column without transformations for the pipelines to handle, and (G) the code from TabRed is outdated and does not function the same anymore when it comes to down-casting string and categorical data of the data from Kaggle, we fix this by treating categorical data as categorical following the original Kaggle scripts, moreover, we had to remove getting the std() of a date column, which is not supported anymore.\n- We drop case_id as it is already collapsed and thus does not contain extra information.\n- We cast all object and string columns to be categorical. Note that they are anonymized and string preprocessing likely does not add a lot of value as a result.\n- We drop name-related columns as they do not hold relevant information. These were otherwise dropped as high-cardinality string columns in TabRed or on Kaggle.\n", "version_from_unique_name": "home_credit_default_stability", "version_comment": "\nWe randomly sub-sample the train data to 1 million rows. We follow TabReD and use random sub-sampling. The idea behind this instead of a time-based subsampling is to keep data from various time periods and model the distribution shift across the full time horizon.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/homeq_default_prediction/019d7374-cab8-700e-97f3-9c55ea980153/dataset_metadata.dataset-mold-v1.json b/homeq_default_prediction/019d7374-cab8-700e-97f3-9c55ea980153/dataset_metadata.dataset-mold-v1.json index 5ef104bf1a908d511e4cfdd7b1764355f015145d..95fac0f3bd812b26c0e306794127e26dbedd405f 100644 --- a/homeq_default_prediction/019d7374-cab8-700e-97f3-9c55ea980153/dataset_metadata.dataset-mold-v1.json +++ b/homeq_default_prediction/019d7374-cab8-700e-97f3-9c55ea980153/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We rename the target column and make its values more semantically meaningful (0 -> \"no_default\", 1 -> \"default\").\n- We noticed several columns missing a large number of features. In all these cases the loan also did not have a value for \"REASON\". We drop 252 rows that dont have a reason for their loan in the data as these are most likely data errors that we want to ignore.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/homesite_quote_conversion/019d736d-c045-7597-b0e9-665db9c61dac/dataset_metadata.dataset-mold-v1.json b/homesite_quote_conversion/019d736d-c045-7597-b0e9-665db9c61dac/dataset_metadata.dataset-mold-v1.json index b7a3fcf5ce026c66886c5bde940a116d733bc61c..cda34925171542d56eaa507a7e119306bceaecd6 100644 --- a/homesite_quote_conversion/019d736d-c045-7597-b0e9-665db9c61dac/dataset_metadata.dataset-mold-v1.json +++ b/homesite_quote_conversion/019d736d-c045-7597-b0e9-665db9c61dac/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized, so feature meanings are unknown.\n- The data contains decoded spatial information.\n- We fix the Field10 column to be integer instead of string with ',' characters.\n- We convert the Original_Quote_Date to datetime.\n- IMPORTANT NOTE: this dataset was used as a non-IID task by TabRed. We keep it as an IID task, because: (A) the Kaggle discussions regarding validation converged on doing IID cross-validation, (B) based on our understanding of the prediction task IID seems most appropriate, and (C) the official test data on Kaggle contains timestamps from the same period as the train. Concluding from this, we think that treating the data as IID is appropriate and that potential temporal leakage is insignificant (due to A and B) or intended (due to C).\n- We drop the ID number and constant columns.\n- We replace -1 with NaN (https://www.kaggle.com/competitions/homesite-quote-conversion/discussion/17417) and keep ordinals as numeric features following TabRed's preprocessing.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/horse_colic_survival/019d736d-db15-7ad1-b407-b937aa958bd9/dataset_metadata.dataset-mold-v1.json b/horse_colic_survival/019d736d-db15-7ad1-b407-b937aa958bd9/dataset_metadata.dataset-mold-v1.json index f9c0854961e4c40e0b241a2db746d86efa63441f..74e5bdb373130b5f58409b1b1709d0d7ec233379 100644 --- a/horse_colic_survival/019d736d-db15-7ad1-b407-b937aa958bd9/dataset_metadata.dataset-mold-v1.json +++ b/horse_colic_survival/019d736d-db15-7ad1-b407-b937aa958bd9/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI and merge the train and test data.\n\n- We encode missing values as NaN.\n- We drop features that leak the target.\n- We keep only the first instance (hospital number) per horse, as they can be treated multiple times and otherwise leaks information about the target.\n- We use the target \"outcome\" to model a task where we aim to predict if the horse survives or not. We drop cases that have not outcome recorded.\n- We keep the duplicates occurring in the data.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/hotel_booking_demand/019db498-7402-74ea-a334-69938355ae2c/dataset_metadata.dataset-mold-v1.json b/hotel_booking_demand/019db498-7402-74ea-a334-69938355ae2c/dataset_metadata.dataset-mold-v1.json index 7f73ec0e3596f250eed67162f94f8293405fdaca..cce66d94c2cfc1bc9dfea85571ea2fb612e51360 100644 --- a/hotel_booking_demand/019db498-7402-74ea-a334-69938355ae2c/dataset_metadata.dataset-mold-v1.json +++ b/hotel_booking_demand/019db498-7402-74ea-a334-69938355ae2c/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n - We concatenate the two datasets H1 (resort hotels) and H2 (city hotels). We add a column \"hotel\" to indicate the hotel type.\n - Note: The paper shares several thoughts on data curation, looking at them can be insightful for understanding the data and its limitations. \n - We remove 26.80% duplicates from the data. It is nearly impossible that so many bookings have are booked the same time (# days) in advance on the exact same date for the exact same date. As the paper mentioned that the data was joined from multiple databases, it can be assumed that many duplicates are a data collection artifact, rather than systematic. Therefore, although some duplicates might occur realistically, we remove all duplicates to avoid data leakage.\n - We remove all bookings with a \"LeadTime\" of 0, assuming that bookings that are booked and canceled on the same day cannot be predicted.\n - Note: The data would allow for a second target to predict no shows.\n - We reconstruct the arrival date from the three columns \"ArrivalDateYear\", \"ArrivalDateMonth\", and \"ArrivalDateDayOfMonth\".\n - We create a \"booking_date\" column by subtracting the \"LeadTime\" from the \"arrival_date\". This is necessary to further define the prediction point in time and to filter the data to avoid data leakage. \n - To define train/test splits, we define 9 points in time each starting on the 1st of a month. Using the available time stamps we separate bookings that lie in the future, and were not canceled yet as the test data.\n - To account for the fact that, the available bookings suddenly stop at the latest month, the latest prediction point is set to three months before the end of the recorded period. This is necessary to avoid an unrealistic distribution shift in the test data, as the later prediction points don't include bookings for later time points that would typically be available. \n - To make sure that the training data reflects the forecasting horizon, we remove all historical bookings that were canceled more than three months in advance. Note that we made this choice based on assumptions without actually testing the impact of this decision. It would be interesting to test the impact of this decision on prediction performance.\n - Bookings that have arrival dates in the future, but were already canceled at the given prediction point are included in the train data. \n - After splitting, we remove \"ReservationStatus\" and \"ReservationStatusDate\" to avoid data leakage.\n - Anomaly: The first recorded month for the city hotels shows an unusually high amount of cancellations.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/houses/019d7368-4493-7f1a-a1e6-d784558ee09a/dataset_metadata.dataset-mold-v1.json b/houses/019d7368-4493-7f1a-a1e6-d784558ee09a/dataset_metadata.dataset-mold-v1.json index 1c37bf446cc4062000cff99a988b159e77f90544..4c48c85b1b11aeb0cea482d374bbcfba2fcd2c54 100644 --- a/houses/019d7368-4493-7f1a-a1e6-d784558ee09a/dataset_metadata.dataset-mold-v1.json +++ b/houses/019d7368-4493-7f1a-a1e6-d784558ee09a/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- The data contains rows whose values were capped artificially at a price of 500001. This creates a censored target, punishing model that learn to extrapolate from the data. We remove rows with this censored value in the target variable to obtain a more realistic task.\n- We kept the latitude and longitude features as they are.\n- We log scaled the target variable (with base e) as intended by the original task.\n- Anomaly: As always, we randomly shuffle the data before uploading. If one does not randomly shuffle the data, there would be a distribution shift from the longitude and latitude based on the original order of data samples.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/hr_analytics/019d7368-56c6-75a9-b489-eb6faf376ce5/dataset_metadata.dataset-mold-v1.json b/hr_analytics/019d7368-56c6-75a9-b489-eb6faf376ce5/dataset_metadata.dataset-mold-v1.json index c019b0bf34f2a2276c12bd4e3fa4e24c4ed2ada0..0c96e3d4351ab3b460973401d118550d9a90c1ed 100644 --- a/hr_analytics/019d7368-56c6-75a9-b489-eb6faf376ce5/dataset_metadata.dataset-mold-v1.json +++ b/hr_analytics/019d7368-56c6-75a9-b489-eb6faf376ce5/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- We renamed the target feature and its values to be more descriptive.\n- We drop the ID column.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/ieee_fraud_detection/019db516-2f8e-7e50-a8c4-f1c57754f52c/dataset_metadata.dataset-mold-v1.json b/ieee_fraud_detection/019db516-2f8e-7e50-a8c4-f1c57754f52c/dataset_metadata.dataset-mold-v1.json index aad503aaebde2a8bb1b6c762de698b19ac3cf076..f4159d5f5a354e155e2b28c2f8e772383b53e515 100644 --- a/ieee_fraud_detection/019db516-2f8e-7e50-a8c4-f1c57754f52c/dataset_metadata.dataset-mold-v1.json +++ b/ieee_fraud_detection/019db516-2f8e-7e50-a8c4-f1c57754f52c/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- We use insights of the first place solution on Kaggle for conceptualizing the task: https://www.kaggle.com/competitions/ieee-fraud-detection/discussion/111284\n- The data is given as transactions, but the task is to predict fraudulent clients. Once a transaction is detected as fraud, the entire account is considered fraudulent.\n- The competition host commented on the labeling logic: \"The logic of our labeling is define reported chargeback on the card as fraud transaction (isFraud=1) and transactions posterior to it with either user account, email address or billing address directly linked to these attributes as fraud too\" (https://www.kaggle.com/c/ieee-fraud-detection/discussion/101203#589276)\n- We load train_transaction.csv and train_identity.csv and merge them on TransactionID.\n- We derive Transaction_date from TransactionDT using the competition reference start date 2017-11-30.\n- The Kaggle competition also provides test_transaction.csv and test_identity.csv, but they are unlabeled and are therefore excluded from df for curation checks.\n- We normalize the D-columns, since they correspond to days since a certain event, we can normalize them by the transaction date to get the actual day of the event. This is also what the first place solution on Kaggle did.\n- We add a uid feature used by the first place solution on Kaggle to identify unique users. Note that this feature is not perfect and just an approximation by the Kaggle users. Moreover, due to missing values, for ~15% of the samples no uid could be reconstructed.\n- Following the first place Kaggle solution, which used month-based cross-validation, we create splits using forecasting horizons of 1 month.\n- The competition test data left a planning window of one month, so do we.\n- Anomaly: Missing values.\n- Anomaly: Several features might allow to approximate to the uid and can lead to overfitting.\n- Anomaly: Grouped data with members of a group often sharing the same label, depending on whether fraud occurred previously or not.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/immoscout_german_house_prices/019d7374-a0e4-7646-b11b-f762fe9b693b/dataset_metadata.dataset-mold-v1.json b/immoscout_german_house_prices/019d7374-a0e4-7646-b11b-f762fe9b693b/dataset_metadata.dataset-mold-v1.json index 1e8494f72375b182d9e4d4bedffec025f8e50443..468773a725973a49fbe13ed1ce322ca54f75628e 100644 --- a/immoscout_german_house_prices/019d7374-a0e4-7646-b11b-f762fe9b693b/dataset_metadata.dataset-mold-v1.json +++ b/immoscout_german_house_prices/019d7374-a0e4-7646-b11b-f762fe9b693b/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe treat this as an IID time-independent task, because the prices is not sold prices but instead the offer price, so the snapshot here represents a task where someone would want to predict the price they should offer for their house given other houses that are currently on the market. Thus, it seems like a real IID task where someone would want to build a model to know how much they should offer their houses at in the current market (with the limitation that past trends are ignored)\n\n- We log scale it, as is appropriate for house price prediction tasks. We do not normalize by lot or living space, as both houses with a lot of land and small living space and houses with a lot of living space and small land are interesting to predict the price of, and normalization would bias towards one of them, as both cannot be homogenised.\n- We drop Free_of_Relation as it indicates a (mostly) price-independent variable; showing logistic information. Note that this column also indicates that the houses are most likely from the same time, further justifying our use case of an IID task.\n- We drop all cases where the price is less than 10k (most of which have a price of 0) as we think this might be a data error, or a case where the price is put to 0 to be marked as \"price on request\" (which is common in house listings).\n- We drop all duplicates as these are most likely repeated entries given the feature space.\n- We drop all rows where the living space is less than 10 (in Germany, this would likely not be enough living space for a real house, and thus might be a data error.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/in_vehicle_coupon_recommendation/019d7368-68cc-754c-99e4-7e145a96e287/dataset_metadata.dataset-mold-v1.json b/in_vehicle_coupon_recommendation/019d7368-68cc-754c-99e4-7e145a96e287/dataset_metadata.dataset-mold-v1.json index 09ff0d571926829479662c836c43af5460800db8..b133738603c14b71a3545338aa15024e5ec318b0 100644 --- a/in_vehicle_coupon_recommendation/019d7368-68cc-754c-99e4-7e145a96e287/dataset_metadata.dataset-mold-v1.json +++ b/in_vehicle_coupon_recommendation/019d7368-68cc-754c-99e4-7e145a96e287/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the target feature and its value to be more descriptive.\n- We removed text data from the \"time\" feature to make it numeric.\n- We drop constant columns.\n- We fixed a typo in the feature names.\n- Anomaly: the data has many binned numeric features that are treated as categorical features with text describing the bins.\n- Anomaly: the numeric features in the dataset are low-cardinality (<25)\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/indian_liver_patient_dataset/019d736d-ec03-75de-b8e9-075fcdde7993/dataset_metadata.dataset-mold-v1.json b/indian_liver_patient_dataset/019d736d-ec03-75de-b8e9-075fcdde7993/dataset_metadata.dataset-mold-v1.json index 53e6de548924de1006610ebbd5c88ff63953a661..23581becf02d7185be533a12df50a07dc01b2e96 100644 --- a/indian_liver_patient_dataset/019d736d-ec03-75de-b8e9-075fcdde7993/dataset_metadata.dataset-mold-v1.json +++ b/indian_liver_patient_dataset/019d736d-ec03-75de-b8e9-075fcdde7993/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe keep the data as is as it does not require any further cleaning.\nNote, the data contains a few (n=13) duplicates, that are likely naturally occurring duplicates given the features.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/iranian_churn/019d7374-7b3c-7eed-940f-e8ac3c68e2f4/dataset_metadata.dataset-mold-v1.json b/iranian_churn/019d7374-7b3c-7eed-940f-e8ac3c68e2f4/dataset_metadata.dataset-mold-v1.json index e465ad37453dda0ea4c79509cbc39c83e16f5a91..1265dec2215ef1c6617f97feb3118b30da45851e 100644 --- a/iranian_churn/019d7374-7b3c-7eed-940f-e8ac3c68e2f4/dataset_metadata.dataset-mold-v1.json +++ b/iranian_churn/019d7374-7b3c-7eed-940f-e8ac3c68e2f4/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- The data collection process for the task is not ideal. The data was collected for each individual right up to the churn month. That means, for the churners we observed varying length periods, while for the other individuals we observed the full period.\n- Nevertheless, the features are not collected after the churn and the task is still valid, if we conceptualize it as \u201cidentify customers close to churn\u201d rather than \u201cpredict churn ahead of time\u201d.\n- We remove exact duplicates (9.52% of the samples) from the data to avoid leaks. The data contains features like \"Seconds of Use\", \"Frequency of use\", \"Frequency of SMS\", \"Distinct Called Numbers\", and most importantly a customer value assignment. Given just 3000 samples, it is reasonable to assume that if feature values with so many degrees of freedom are exactly the same, they belong to the same customer.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/jm1/019d7368-7f0b-74da-a01f-3c5e7dd31dab/dataset_metadata.dataset-mold-v1.json b/jm1/019d7368-7f0b-74da-a01f-3c5e7dd31dab/dataset_metadata.dataset-mold-v1.json index 338ec422376a43168ec9e6d376667218b1327865..f6cb9b4f6506aa3e45cb250fb59787bf61658bf0 100644 --- a/jm1/019d7368-7f0b-74da-a01f-3c5e7dd31dab/dataset_metadata.dataset-mold-v1.json +++ b/jm1/019d7368-7f0b-74da-a01f-3c5e7dd31dab/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We selected this dataset as representative of a set of predictive modeling tasks of software engineering (other examples of such tasks are available on OpenML under the names mozilla4, pc1, pc2, pc3, mc1, kc1, kc2). This dataset was selected as it has the largest sample size.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/kdd_cup_09_appetency/019d7368-96e4-777a-bfbb-4fcb8602465c/dataset_metadata.dataset-mold-v1.json b/kdd_cup_09_appetency/019d7368-96e4-777a-bfbb-4fcb8602465c/dataset_metadata.dataset-mold-v1.json index 6fa927b1e81fcfc26545dfea4f2373a33b271887..0180e11435d9139698d19769e81785bed584c153 100644 --- a/kdd_cup_09_appetency/019d7368-96e4-777a-bfbb-4fcb8602465c/dataset_metadata.dataset-mold-v1.json +++ b/kdd_cup_09_appetency/019d7368-96e4-777a-bfbb-4fcb8602465c/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We use the small training data from the original data (230 features).\n- We use appetency as label.\n- We dropped empty columns: \"Var8\", \"Var15\", \"Var20\", \"Var31\", \"Var32\", \"Var39\", \"Var42\", \"Var48\", \"Var52\", \"Var55\", \"Var79\", \"Var141\", \"Var167\", \"Var169\", \"Var175\", \"Var185\", \"Var209\", \"Var230\".\n- Anomaly: the feature names and categorical feature values have no semantic meaning.\n- Anomaly: this dataset has many missing values.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/kick/019d744f-147d-721e-affc-94b58645c0ad/dataset_metadata.dataset-mold-v1.json b/kick/019d744f-147d-721e-affc-94b58645c0ad/dataset_metadata.dataset-mold-v1.json index c7b1f579a350415bc4bfdcaf3bfc998125eaf0e1..c1d1470576175e64e992eee7bedbe0eafaea52e0 100644 --- a/kick/019d744f-147d-721e-affc-94b58645c0ad/dataset_metadata.dataset-mold-v1.json +++ b/kick/019d744f-147d-721e-affc-94b58645c0ad/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n - The data is from a kaggle competition that used grouped splits based on \"Auction\" and \"VNZIP1\".\n - The data is temporal, but also grouped by auctions. Shifts can be expected from both, but auctions reappear at later time points, so it is unclear whether a grouped split truly reflects deployment conditions. Therefore, we use temporal evaluation.\n - The test features from the Kaggle competition are available and could be used for unsupervised approaches.\n - We transform \"PurchDate\" to datetime.\n - We drop the \"RefId\" column, as it is just an identifier and not useful for modeling.\n - We assign the categorical features \"WheelTypeID\", \"BYRNO\", and \"VNZIP1\" to category dtype, as they are given as numbers.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/kickstarter/019db497-6b35-71e0-9c6b-f4a7286f8f6f/dataset_metadata.dataset-mold-v1.json b/kickstarter/019db497-6b35-71e0-9c6b-f4a7286f8f6f/dataset_metadata.dataset-mold-v1.json index 2fc516585189ba0074a4848c1393579bcaf9a566..5bc776a0ed95be2910ca937f3b5147ede67ae78f 100644 --- a/kickstarter/019db497-6b35-71e0-9c6b-f4a7286f8f6f/dataset_metadata.dataset-mold-v1.json +++ b/kickstarter/019db497-6b35-71e0-9c6b-f4a7286f8f6f/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nSimilar to the Kaggle task, we aim to predict whether a Kickstarter project will be funded successfully.\n\n- The data we used only had 2 samples for 2026, so we stick to data until 2025.\n- The original data comes in .csv files per scrape data. We first combine all files into one file.\n- We remove all columns that are known only after the outcome and thus could leak information.\n- The fx_rate seems to show the conversion rate at the time of crawling the data and not at the time of the project. Thus, if it is used as on Kaggle, the numbers are wrong. We use the \"usd_exchange_rate\" to transform the goal into USD currency.\n- We drop the two entries with \"disable_communication\" as they point to other issues.\n- We drop photo and video based references as we do not include these modalities.\n- We decode the category and creator name from JSON strings into usable columns.\n- Note, the crawl does only include blurbs and not the full-text descriptions of the projects. Also some blurb repeat from similar projects or orders.\n- We decode the profile blurb, when it exists.\n- We decode the location display name from the location JSON string, which can include state and city name.\n- The data contains spatial information (city, state, country). We do not decode this but leave it to the pipelines.\n- We found 150 rows without location information. We drop these rows as it is unclear which issue caused this and how the rows' data might be affected by this.\n- We drop duplicates (20%) which seems to have occurred from multiple scrapes of the same project.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/labour_inspection_compliance/019d736e-5a12-720a-9281-b652d1acae27/dataset_metadata.dataset-mold-v1.json b/labour_inspection_compliance/019d736e-5a12-720a-9281-b652d1acae27/dataset_metadata.dataset-mold-v1.json index 835a797706badc13be0756eaf16d85317403a6c4..bf1e8b3ce4af8f20cd9edb9e5f8eb0e8c6583922 100644 --- a/labour_inspection_compliance/019d736e-5a12-720a-9281-b652d1acae27/dataset_metadata.dataset-mold-v1.json +++ b/labour_inspection_compliance/019d736e-5a12-720a-9281-b652d1acae27/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with the data from dataverse. We create a task for the Non-compliance Classification Problem (NCP).\n\n- We resolve the industry codes to their true names.\n- We drop the checklist ID, as we use the checklist text as predictive signal.\n- We remove two rows with missing values for IsRegisteredVATregister, as these seem to be data artifact given the data state.\n- The data has spatial information (County).\n- There are lot of undescribed numeric values. Thus, we do some general purpose preprocessing and drop all columns with less than 5 not nan values\n- As it turns out, the data does not contain temporal information, although it is clearly a temporal task and has data from 2021 to 2019. It seems the curators have removed the temporal information. This can mean that there is not temporal leakage or relevance for the task, or that it was forgotten. In any case, we must treat the task as IID now.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/lending_club/versions/019d738c-5edd-7437-9116-e3bc87a8b0c5/dataset_metadata.dataset-mold-v1.json b/lending_club/versions/019d738c-5edd-7437-9116-e3bc87a8b0c5/dataset_metadata.dataset-mold-v1.json index c4e1a440a43c21b55b0f1f78b9ac08d5e2b08fac..8645740704411cb8d7b36b1e20db45281740f99d 100644 --- a/lending_club/versions/019d738c-5edd-7437-9116-e3bc87a8b0c5/dataset_metadata.dataset-mold-v1.json +++ b/lending_club/versions/019d738c-5edd-7437-9116-e3bc87a8b0c5/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start withe data from Sanz-Guerrero et al. (2025). This data is already well preprocessed and curated for the task and used for tabular-text learning. However, we noticed a lot variables from the original data are missing that can be added to the task without invalidating the task or introducing data leakage. Thus, we also merge new features from the original data into the version from Sanz-Guerrero et al. (2025).\n\n- This is a datasets where the description is very often empty or a standard phrase. So the model needs to be able to handle cases with a lot of missing data in the text modality, and also cases where the text is rarely informative. This is a common case in real-world tabular-text datasets, and it is important to have it represented in our benchmark.\n- We reverse the ordinal encoding of the target.\n- We reverse the name change of \"revenue\" back to \"annual_inc\".\n- We also create a sec_app_fico_n like the fico_n from anz-Guerrero et al. (2025).\n- We drop rows where \"application_type\" is missing as these rows have consistent missing values across features and likely represent some data loading artifact.\n", "version_from_unique_name": "lending_club", "version_comment": "\nTo sub-sample this dataset, we only take the first year to obtain 814751 train and 276431.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/ljubljana_breast_cancer/019d736e-6ea7-7889-b0ff-20fbdd628b8a/dataset_metadata.dataset-mold-v1.json b/ljubljana_breast_cancer/019d736e-6ea7-7889-b0ff-20fbdd628b8a/dataset_metadata.dataset-mold-v1.json index f6793e2ac9bda5cacf43c959c75e4db26293417c..48f1810cdaa9ab1c5bb2546d64177abb490b2bb7 100644 --- a/ljubljana_breast_cancer/019d736e-6ea7-7889-b0ff-20fbdd628b8a/dataset_metadata.dataset-mold-v1.json +++ b/ljubljana_breast_cancer/019d736e-6ea7-7889-b0ff-20fbdd628b8a/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- We encode missing values as np.nan instead of \"?\".\n- We reverse the discretization of several numeric features and make them integers again.\n- We keep duplicates as they seem naturally occurring.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/ljubljana_primary_tumor/019d736e-819e-7f44-a8b2-5b93d4cce230/dataset_metadata.dataset-mold-v1.json b/ljubljana_primary_tumor/019d736e-819e-7f44-a8b2-5b93d4cce230/dataset_metadata.dataset-mold-v1.json index a0958e9b3d1da1cc796b09f6a674c149b5f49b62..14cb950bff001ea90a53f9aa5d7714fb9a0a3811 100644 --- a/ljubljana_primary_tumor/019d736e-819e-7f44-a8b2-5b93d4cce230/dataset_metadata.dataset-mold-v1.json +++ b/ljubljana_primary_tumor/019d736e-819e-7f44-a8b2-5b93d4cce230/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- We encode missing values as np.nan instead of \"?\".\n- We reverse the ordinal encoding of all features, adding the text back to the categories.\n- We drop classes with less than 10 samples to have sufficient data per class for a robust benchmark task. In reality, this would call for collecting more data on these classes, or treating it as an individual few-shot prediction task.\n- The data contains many duplicates (10%), which seems to be naturally occurring given the limited amount of features. Moreover, the duplicates sometimes have different targets. We keep the duplicates. While this might bias evaluation a bit, it represents the real-world task with ambiguous features better.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/lung_cancer/019d7375-0c48-7822-8f4e-5e071cbcabe5/dataset_metadata.dataset-mold-v1.json b/lung_cancer/019d7375-0c48-7822-8f4e-5e071cbcabe5/dataset_metadata.dataset-mold-v1.json index e9dec80b755c27daa794887bc547b0c36cd9fca9..9b90ac963839f7071dfbd8957ce9ebe8df9f5d91 100644 --- a/lung_cancer/019d7375-0c48-7822-8f4e-5e071cbcabe5/dataset_metadata.dataset-mold-v1.json +++ b/lung_cancer/019d7375-0c48-7822-8f4e-5e071cbcabe5/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- The original data comes with lung adenocarcinoma and other adenocarcinomas merged into one class already. We keep the same and were not able to find a way to reverse this from the public data.\n- We also drop the class \"small-cell lung carcinoma\" as it only has 6 samples, which is not a meaningful sample size for benchmarking, or the predictive task. We could add another version with all classes in the future.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/lung_cancer_epithelial_genexp/019d7376-82a5-76cc-b1ce-92dc200f3173/dataset_metadata.dataset-mold-v1.json b/lung_cancer_epithelial_genexp/019d7376-82a5-76cc-b1ce-92dc200f3173/dataset_metadata.dataset-mold-v1.json index 1a0481641253013127d10cc252f07e6f2402c218..1215139cc765437d7bb61adcb3cd38fe30fc0024 100644 --- a/lung_cancer_epithelial_genexp/019d7376-82a5-76cc-b1ce-92dc200f3173/dataset_metadata.dataset-mold-v1.json +++ b/lung_cancer_epithelial_genexp/019d7376-82a5-76cc-b1ce-92dc200f3173/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We start with an extended dataset of 192 patients. The original reference uses 129 samples as \"Individuals without final diagnoses as of May 2005 were excluded from this primary dataset\". We remove 5 samples without a clear label (\"Smoker with suspect lung cancer Sample\"), yielding a dataset of 187 samples.\n- We use \"Sample_title\" from the metadata as labels for the prediction task.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/maps_router_eta/versions/019d7407-606f-7147-b041-c7d0a3847c71/dataset_metadata.dataset-mold-v1.json b/maps_router_eta/versions/019d7407-606f-7147-b041-c7d0a3847c71/dataset_metadata.dataset-mold-v1.json index 29cda014f10b3c1dc8e23940b8cf3d5b6d0acffb..ddac14c4f05a1d7c5fb86afddc834cc62d81ec0c 100644 --- a/maps_router_eta/versions/019d7407-606f-7147-b041-c7d0a3847c71/dataset_metadata.dataset-mold-v1.json +++ b/maps_router_eta/versions/019d7407-606f-7147-b041-c7d0a3847c71/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with data from TabRed, which already comes preprocessed.\nWe perform no additional preprocessing steps.\n", "version_from_unique_name": "maps_router_eta", "version_comment": "\nWe randomly sub-sample the train to 1 million and test data 250k rows. We follow TabReD and use random sub-sampling. The idea behind this instead of a time-based subsampling is to keep data from various time periods and model the distribution shift across the full time horizon.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/marketing_campaign/019d7441-e83d-7bb2-bed5-7eb8a1ad36f6/dataset_metadata.dataset-mold-v1.json b/marketing_campaign/019d7441-e83d-7bb2-bed5-7eb8a1ad36f6/dataset_metadata.dataset-mold-v1.json index 4d1fd7a0333d9e5e52c593820899ef2f94aa7c96..7b5c59b01e386ea42926842ad7a565322fe4bc11 100644 --- a/marketing_campaign/019d7441-e83d-7bb2-bed5-7eb8a1ad36f6/dataset_metadata.dataset-mold-v1.json +++ b/marketing_campaign/019d7441-e83d-7bb2-bed5-7eb8a1ad36f6/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We ensure Dt_Customer is a pandas datetime.\n- We drop the ID column.\n- We rename the values of the target variable to be more descriptive.\n- We drop constant columns.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/maternal_health_risk/019d7368-bc0d-723e-a4f9-abfd36e085e9/dataset_metadata.dataset-mold-v1.json b/maternal_health_risk/019d7368-bc0d-723e-a4f9-abfd36e085e9/dataset_metadata.dataset-mold-v1.json index a5dd6d425751b1f7182b90bdc1d5a8e9b16fb708..c968ea83935800b8ed23f37e24e45e16b4b44519 100644 --- a/maternal_health_risk/019d7368-bc0d-723e-a4f9-abfd36e085e9/dataset_metadata.dataset-mold-v1.json +++ b/maternal_health_risk/019d7368-bc0d-723e-a4f9-abfd36e085e9/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- Anomaly: the data has a lot of duplicates (55%).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/mercari_price_suggestion/versions/019d736f-26fd-73bd-8853-dc219e5f4ed5/dataset_metadata.dataset-mold-v1.json b/mercari_price_suggestion/versions/019d736f-26fd-73bd-8853-dc219e5f4ed5/dataset_metadata.dataset-mold-v1.json index af7b08f28a3c88a18bb2260ee5308da2cc681e7f..a452acbe2b585b4592ab60157eeb818e13ddfe11 100644 --- a/mercari_price_suggestion/versions/019d736f-26fd-73bd-8853-dc219e5f4ed5/dataset_metadata.dataset-mold-v1.json +++ b/mercari_price_suggestion/versions/019d736f-26fd-73bd-8853-dc219e5f4ed5/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start from the Kaggle competition dataset.\n\n- We log1p scale the target to match the target metric of the competition by default RMSE.\n- The dataset was solved by Kaggle experts by tabular models and a lot of custom string preprocessing. So it is a great use case to test encoding pipelines! The dataset also has mostly string data.\n- The task was treated and solved as an IID task on Kaggle.\n- We replace \"No description yet\" with NaN.\n", "version_from_unique_name": "mercari_price_suggestion", "version_comment": "\nWe randomly sub-sample to 1.25 million rows such that train and test splits are roughly 1M and 250k rows respectively.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/mercedes_benz_greener_manufacturing/019d7384-bae5-71ab-ab7c-361f23e277dc/dataset_metadata.dataset-mold-v1.json b/mercedes_benz_greener_manufacturing/019d7384-bae5-71ab-ab7c-361f23e277dc/dataset_metadata.dataset-mold-v1.json index 5a9bffd7a698bc9d4addb7b02a1d5849f721cb56..77214b6d3a3c764ed8d5875c044727dc9adc5828 100644 --- a/mercedes_benz_greener_manufacturing/019d7384-bae5-71ab-ab7c-361f23e277dc/dataset_metadata.dataset-mold-v1.json +++ b/mercedes_benz_greener_manufacturing/019d7384-bae5-71ab-ab7c-361f23e277dc/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized, so feature meanings are unknown.\n- There is generally weird behavior related to the ID and potentially a time column (see X5 here https://www.kaggle.com/competitions/mercedes-benz-greener-manufacturing/discussion/34949). However, it is hard to parse the exact details due to anonymization. We generally take from this that the data is sorted sequentially by ID and that there may be time dependencies. Also checkout https://www.kaggle.com/code/sudalairajkumar/simple-exploration-notebook-mercedes which shows relation between index and y, showing that y goes up with higher index and it shows that X5 (the time group feature), has the highest importance. Plus, it has one outlier.\n- We remove the one outlier from y.\n- We drop X5 as it leaks time group information in temporal splits (besides having a full categorical-value distribution shift with a temporal split).\n- We tread the ID as a time_index.\n- Some of the other constructed features (X0-X8) may leak across time as well. We remove all of them for which we do not have an explanation from experts on Kaggle.\n- There are exactly 4 rows that have value that is not equal to \"d\" in column \"X4\". This indicates some kind of sub-group or data different. Given the lack of semantics and anonymity, we cannot parse the meaning of \"X4\". Thus, to avoid this being an indicator of data leakage, we remove these rows and subsequently drop \"X4\" as well.\n- We drop other constructed features from up to X10 for which we have no information about them and they may leak group information across rows.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/miami_housing/019d7368-cda5-7177-b500-4986a5e77a82/dataset_metadata.dataset-mold-v1.json b/miami_housing/019d7368-cda5-7177-b500-4986a5e77a82/dataset_metadata.dataset-mold-v1.json index d58b099919638d68a1d594f930668d7ce70c63fd..cb4ec389fc458c66ac7d578cf42cb0eb5b5134c7 100644 --- a/miami_housing/019d7368-cda5-7177-b500-4986a5e77a82/dataset_metadata.dataset-mold-v1.json +++ b/miami_housing/019d7368-cda5-7177-b500-4986a5e77a82/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- We log scale the target.\n- We drop duplicated homes (non-unique identifiers and location) (~1% of the data).\n- We drop the ID column \"PARCELNO\".\n- Anomaly: while the original paper describes data with 57k samples, we only have 13k on OpenML or Kaggle.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/mic/019d7444-22cb-7d04-940f-6fa8b0436ea2/dataset_metadata.dataset-mold-v1.json b/mic/019d7444-22cb-7d04-940f-6fa8b0436ea2/dataset_metadata.dataset-mold-v1.json index c359ee134304b257dcb2261839ed67db477ece3b..3b945657277702209fc8f2bd6303f63d61d728b7 100644 --- a/mic/019d7444-22cb-7d04-940f-6fa8b0436ea2/dataset_metadata.dataset-mold-v1.json +++ b/mic/019d7444-22cb-7d04-940f-6fa8b0436ea2/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- There are 12 possible targets and four possible time moments to predict the targets for this dataset. We use the categorical target as it summarizes the other possible targets. Furthermore, we use the last possible time point for prediction to utilize all available data.\n- We treat \"?\" as missing values.\n- We dropped the \"id\" column.\n- We reversed the ordinal encoding of the target feature.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/mice_protein_trisomy_discriminant/019d738d-99a8-7850-b911-f244ac4ad485/dataset_metadata.dataset-mold-v1.json b/mice_protein_trisomy_discriminant/019d738d-99a8-7850-b911-f244ac4ad485/dataset_metadata.dataset-mold-v1.json index 9d52fa7f72987992a45b6f9f169302d76c0c108b..6737188e278eaf1df8577b55d7f36a227db9f503 100644 --- a/mice_protein_trisomy_discriminant/019d738d-99a8-7850-b911-f244ac4ad485/dataset_metadata.dataset-mold-v1.json +++ b/mice_protein_trisomy_discriminant/019d738d-99a8-7850-b911-f244ac4ad485/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the dataset from UCI.\n\n- We drop the three feature that make up the target to remove the leakage.\n- We reduce the mouse ID to represent the ID of the mouse instead of the experiment for the mouse.\n- We drop the duplicated feature pS6_N.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/micro_mass/019d738d-c82e-796a-9343-e30919f324d5/dataset_metadata.dataset-mold-v1.json b/micro_mass/019d738d-c82e-796a-9343-e30919f324d5/dataset_metadata.dataset-mold-v1.json index dbc2070a4e69835126fb8852f89d18318ad63eaa..861a597e35fac2786dd16ef0d4fdf31514fd6abf 100644 --- a/micro_mass/019d738d-c82e-796a-9343-e30919f324d5/dataset_metadata.dataset-mold-v1.json +++ b/micro_mass/019d738d-c82e-796a-9343-e30919f324d5/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe use the pure spectra version (without mixing) and replicate a grouped-based task to predict the species of strains.\n\n- We drop columns with the same value across all samples.\n- We drop constant columns (all 0).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/musk/019d738d-ae72-7a0a-a2e4-6b797ab153fd/dataset_metadata.dataset-mold-v1.json b/musk/019d738d-ae72-7a0a-a2e4-6b797ab153fd/dataset_metadata.dataset-mold-v1.json index 74ea3d1d3935cab37f71644d28116a12f3b718f4..947eaa63fae1c628cb1845d5b8102b367f7df2d3 100644 --- a/musk/019d738d-ae72-7a0a-a2e4-6b797ab153fd/dataset_metadata.dataset-mold-v1.json +++ b/musk/019d738d-ae72-7a0a-a2e4-6b797ab153fd/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- We rename the molecule IDs to remove the target leakage from the names.\n- We drop the conformation name as it leaks information that the real task should not have (the correlation between specific conformations across samples).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/mutual_funds_india/019d736f-4e6b-7953-9a01-a5476b6708ef/dataset_metadata.dataset-mold-v1.json b/mutual_funds_india/019d736f-4e6b-7953-9a01-a5476b6708ef/dataset_metadata.dataset-mold-v1.json index b2e35c08c04184b97b2c4f2e9a9f7f1ded67973b..86b965c093530d59aa50be26d9fe01d62b92dbd4 100644 --- a/mutual_funds_india/019d736f-4e6b-7953-9a01-a5476b6708ef/dataset_metadata.dataset-mold-v1.json +++ b/mutual_funds_india/019d736f-4e6b-7953-9a01-a5476b6708ef/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from Kaggle.\n\n- We aim to predict the return over 3 years and drop all samples that do not have data for 3 year return.\n- We drop potentially leaking columns about returns (alpha, beta, sd, sharpe, sortino).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/naticusdroid_android_permissions_dataset/019d7368-f37d-77e2-be2e-0d8d94e9c27d/dataset_metadata.dataset-mold-v1.json b/naticusdroid_android_permissions_dataset/019d7368-f37d-77e2-be2e-0d8d94e9c27d/dataset_metadata.dataset-mold-v1.json index cf9f05212772420f24f009f56e85d80fb9388e99..c1338b52b8681e7cc8675a84edfcc7d892c69be2 100644 --- a/naticusdroid_android_permissions_dataset/019d7368-f37d-77e2-be2e-0d8d94e9c27d/dataset_metadata.dataset-mold-v1.json +++ b/naticusdroid_android_permissions_dataset/019d7368-f37d-77e2-be2e-0d8d94e9c27d/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- Originally, the data has 75% duplicates and a perfect class balance. The authors never mention duplicates in their paper. Given this large amount of duplicates, we dropped all row-duplicates, keeping only the first entry.\n- We drop one duplicated column.\n- Anomaly: all features are binary categorical features.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/obesity_estimation/019d7374-8d75-752a-aca4-1f311044d316/dataset_metadata.dataset-mold-v1.json b/obesity_estimation/019d7374-8d75-752a-aca4-1f311044d316/dataset_metadata.dataset-mold-v1.json index fb779636b82738f73a16b1b09a8607a297c8eb76..8f79a5d8eaab644673b258f428e87d47c844a499 100644 --- a/obesity_estimation/019d7374-8d75-752a-aca4-1f311044d316/dataset_metadata.dataset-mold-v1.json +++ b/obesity_estimation/019d7374-8d75-752a-aca4-1f311044d316/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We use only the real data, throwing away the SMOTE samples.\n- We make the task a regression task, re-creating the body mass values underlying the discretized obesity levels, and drop weight and height which leak the target partially. Thus, we create a regression task of estimating the body mass of individuals based on their eating habits and other questions from the questionnaire. The original task is trivial to solve due to the leak and thus not interesting.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/online_shoppers_purchasing_intention_dataset/019d7445-801c-7a91-831e-5f6725779505/dataset_metadata.dataset-mold-v1.json b/online_shoppers_purchasing_intention_dataset/019d7445-801c-7a91-831e-5f6725779505/dataset_metadata.dataset-mold-v1.json index e80be8100e774eb96ced26ee302317d0428e7be9..34eb06a8752619e921b9712f07a43feacd58c166 100644 --- a/online_shoppers_purchasing_intention_dataset/019d7445-801c-7a91-831e-5f6725779505/dataset_metadata.dataset-mold-v1.json +++ b/online_shoppers_purchasing_intention_dataset/019d7445-801c-7a91-831e-5f6725779505/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- Anomaly: the data contains time-based features that were preprocessed to create a time-invariant predictive task.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/otto_group_product_classification_challenge/019d736f-6330-78c4-91bb-c37e46162244/dataset_metadata.dataset-mold-v1.json b/otto_group_product_classification_challenge/019d736f-6330-78c4-91bb-c37e46162244/dataset_metadata.dataset-mold-v1.json index ce68d1f24c8cb44b741aa531289eb4cf35b4e0e8..0d165ee5233bedf02a38199d876da6f6fc627a56 100644 --- a/otto_group_product_classification_challenge/019d736f-6330-78c4-91bb-c37e46162244/dataset_metadata.dataset-mold-v1.json +++ b/otto_group_product_classification_challenge/019d736f-6330-78c4-91bb-c37e46162244/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized, so feature meanings are unknown.\n- Features and data is clean. Top Kaggle experts applied no relevant preprocessing.\n- We drop the index column.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/pancreatic_cancer_mouse_detection/019d7391-358e-725a-bc70-bbee346e9267/dataset_metadata.dataset-mold-v1.json b/pancreatic_cancer_mouse_detection/019d7391-358e-725a-bc70-bbee346e9267/dataset_metadata.dataset-mold-v1.json index 597da70ecfb9f5d97aa8bfbf78732e657f4b766e..86ba8f2286aeafdc88fba256637f1946dbd9ee20 100644 --- a/pancreatic_cancer_mouse_detection/019d7391-358e-725a-bc70-bbee346e9267/dataset_metadata.dataset-mold-v1.json +++ b/pancreatic_cancer_mouse_detection/019d7391-358e-725a-bc70-bbee346e9267/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nThe task in the original data is grouped as multiple serums per mouse were taken.\n- We merge all samples located in the .csv files into one DataFrame and recover target labels, group IDs, and feature names.\n- We label the samples based on name of the folder they are located in: \"PanINCTL\" translates to healthy animals cases while \"03-PanIN\" to individuals with a pancreatic cancer.\n- Anomaly: The data from the website has one more mouse than accounted for in the paper (73 vs (33 + 39))) but the same number of serums (181).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/parkinsons_biomedical_voice_measurements/019d738d-db39-7b5b-83dc-664e66da1854/dataset_metadata.dataset-mold-v1.json b/parkinsons_biomedical_voice_measurements/019d738d-db39-7b5b-83dc-664e66da1854/dataset_metadata.dataset-mold-v1.json index 65fe119af57ae6cf6cdcf8a848a3550e696b47da..1a0209379f8f07d13816c2705f60cd58c40f4bd4 100644 --- a/parkinsons_biomedical_voice_measurements/019d738d-db39-7b5b-83dc-664e66da1854/dataset_metadata.dataset-mold-v1.json +++ b/parkinsons_biomedical_voice_measurements/019d738d-db39-7b5b-83dc-664e66da1854/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- We decode the group ID and session ID (time index per patient) from the name column.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/physiochemical_protein/019d7369-1810-7d24-8e20-84da94babb3e/dataset_metadata.dataset-mold-v1.json b/physiochemical_protein/019d7369-1810-7d24-8e20-84da94babb3e/dataset_metadata.dataset-mold-v1.json index 07681b88a313d5b36f3532b8fcc87d14a5e5af1f..d0a7399e194108fc94adbbbd81df5904f943f42b 100644 --- a/physiochemical_protein/019d7369-1810-7d24-8e20-84da94babb3e/dataset_metadata.dataset-mold-v1.json +++ b/physiochemical_protein/019d7369-1810-7d24-8e20-84da94babb3e/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the features to be more semantically meaningful.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/polish_companies_bankruptcy/019d7369-2c20-709c-83cd-e27fab38e009/dataset_metadata.dataset-mold-v1.json b/polish_companies_bankruptcy/019d7369-2c20-709c-83cd-e27fab38e009/dataset_metadata.dataset-mold-v1.json index 87e84149f533d55af316418245224bd27f6ad687..56618cfafba242cc181edc492afc4235239bd8c8 100644 --- a/polish_companies_bankruptcy/019d7369-2c20-709c-83cd-e27fab38e009/dataset_metadata.dataset-mold-v1.json +++ b/polish_companies_bankruptcy/019d7369-2c20-709c-83cd-e27fab38e009/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We only use data from year 5 (5year.arff), because it is the newest data and the target is bankruptcy status after only 1 year.\n- We created semantically meaningful feature names.\n- We removed duplicates.\n- Anomaly: the data contains a lot of features created by feature engineering.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/porto_seguro/019d736f-86ed-700f-96c9-8f4be8546db1/dataset_metadata.dataset-mold-v1.json b/porto_seguro/019d736f-86ed-700f-96c9-8f4be8546db1/dataset_metadata.dataset-mold-v1.json index f062ac0c855cc104ee31a4bddc1bbf4c7b633e9d..e195403df506d3920ad453e1aa2fe21d7007850d 100644 --- a/porto_seguro/019d736f-86ed-700f-96c9-8f4be8546db1/dataset_metadata.dataset-mold-v1.json +++ b/porto_seguro/019d736f-86ed-700f-96c9-8f4be8546db1/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized, so feature meanings are unknown.\n- Following the expert solutions on Kaggle, we drop *calc features.\n- In general, the 2nd and 3rd place did a lot of interesting feature engineering for interactions. We do not copy these as the data is clean as it is.\n- We drop the index as it seems to be uninformative for this dataset.\n- There exist only 6 duplicates in the training data. We drop them as the portion is too small to be meaningful.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/predict_students_dropout_and_academic_success/019d7369-3ffd-7889-91d3-46e5b502f4ec/dataset_metadata.dataset-mold-v1.json b/predict_students_dropout_and_academic_success/019d7369-3ffd-7889-91d3-46e5b502f4ec/dataset_metadata.dataset-mold-v1.json index e0b67cab96edfbf90c8f09b14e14d4a28af79285..b5ea39c5942809fc8b23c30aca1ae947f48b6abd 100644 --- a/predict_students_dropout_and_academic_success/019d7369-3ffd-7889-91d3-46e5b502f4ec/dataset_metadata.dataset-mold-v1.json +++ b/predict_students_dropout_and_academic_success/019d7369-3ffd-7889-91d3-46e5b502f4ec/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the target variable and fixed some typos in the feature names.\n- We reversed the ordinal encoding of all categorical features.\n- We removed whitespaces and special characters from the feature names.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/prostate_cancer_detection/019d7375-de7b-7b43-8340-ea47165661d3/dataset_metadata.dataset-mold-v1.json b/prostate_cancer_detection/019d7375-de7b-7b43-8340-ea47165661d3/dataset_metadata.dataset-mold-v1.json index c913ab31e6057881153d6a9462dda7375ef45b0f..dbf3f4529a0bd525252ef4e3d7afc324d4654188 100644 --- a/prostate_cancer_detection/019d7375-de7b-7b43-8340-ea47165661d3/dataset_metadata.dataset-mold-v1.json +++ b/prostate_cancer_detection/019d7375-de7b-7b43-8340-ea47165661d3/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We merge all samples located in the .csv/.CSV files into one DataFrame. We label the samples based on name of the folder they are located in: \"Benign PSA Greater 4\" and \"No Evidence of Disease PSA Less 1\" constitute healthy patient cases while \"Prostate Cancer (..)\" samples point to individuals with a prostate cancer. This distinction is made based on the original reference and the prediction task mentioned in it (tumor vs healthy).\n- Note, we use data that is equivalent to the merged train and test data, as we cannot differentiate what was train/test from the data we got. This should still be a valid task.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/pva_revenue_prediction_kddcup98/019db49b-9c05-72b8-b4a5-7b967a18225b/dataset_metadata.dataset-mold-v1.json b/pva_revenue_prediction_kddcup98/019db49b-9c05-72b8-b4a5-7b967a18225b/dataset_metadata.dataset-mold-v1.json index 115349fdbe090c327d108c720406c569265aaecd..335f5687ea2eaa80c539bc4da6536b61f962427e 100644 --- a/pva_revenue_prediction_kddcup98/019db49b-9c05-72b8-b4a5-7b967a18225b/dataset_metadata.dataset-mold-v1.json +++ b/pva_revenue_prediction_kddcup98/019db49b-9c05-72b8-b4a5-7b967a18225b/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the train and validation data from the KDD Cup 1998 website and combine them.\n\n- The description of the task and data for the KDD Cup (https://kdd.ics.uci.edu/databases/kddcup98/epsilon_mirror/cup98doc.txt) is among the best real-world dataset and task descriptions I have ever seen.\n- The task was made IID through feature engineering and would usually used to predict the next year. We only predict from the training data from the current year, other samples from the current year (as to see who else to contact in the same year). Features contain information from the previous year.\n- The task is revenue generation and the score function is profit-based with a cost of $0.68 per prediction. The higher the revenue generated, the better. We use the binary classification target (TARGET_B) to predict if a person would donate.\n- The TCODE contains titles now in the data dictionary. We map large amount of missing values to placeholder titles and the smaller ones to a new \"other\" title.\n- We treat data as string or category based on the data dictionary suggestions. In cases where the column is from a predefined set of choices, it becomes categorical. We treat various free-text categorical as string columns (they are also high-cardinality otherwise).\n- We transform the date columns to pandas datetime.\n- The data contains spatial information (ZIP, STATE). We do not resolve them and leave them for the pipeline to handle.\n- We remove a faulty postfix (\"-\") in the ZIP code.\n- We drop rows with PVASTATE='E' as this represent a group of donors represent by a different organization chapter. These are 8 rows only.\n- The data contains two entries with a faulty day of birth (born in month 00). We drop both rows, as we assume this is indicative of faulty data.\n- We noticed that nan values and dtypes mismatch for NOEXCH do correlate with the target. We add these cases as their own categories.\n- We resolve the ordinal encoding and nan encoding for all columns.\n- Note, that data has a lot of features that contain data from a similar origin that was transformed into individual columns.\n- We do not decode the promotion codes across years but keep them as string columns as it would blow up the dimensionality too much otherwise.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/qsar_aquatic_toxicity/019d7370-0517-71a1-9e18-dc2a826fa01a/dataset_metadata.dataset-mold-v1.json b/qsar_aquatic_toxicity/019d7370-0517-71a1-9e18-dc2a826fa01a/dataset_metadata.dataset-mold-v1.json index 5ee478ba115fbc66265598255e0d78b2355907ee..551e2c5bbe6baa166f88eb714d66ab9cd4d5a1d6 100644 --- a/qsar_aquatic_toxicity/019d7370-0517-71a1-9e18-dc2a826fa01a/dataset_metadata.dataset-mold-v1.json +++ b/qsar_aquatic_toxicity/019d7370-0517-71a1-9e18-dc2a826fa01a/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\n- Note, the dataset has only a subset of features of the original data. The original data was reduced through feature selection with kNN by the authors in the paper above. We only have the 8 descriptors available in the UCI version. This might leak target information into the feature selection (or behave like an expert giving us the best features). Moreover, it biases the datasets and might make it trivial.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/qsar_biodeg/019d7369-5170-7d75-8932-e2a5778ee70b/dataset_metadata.dataset-mold-v1.json b/qsar_biodeg/019d7369-5170-7d75-8932-e2a5778ee70b/dataset_metadata.dataset-mold-v1.json index 15be0cbc9cbeb86180468b4abbbfdfc05976bc9f..89f709f56e8433dd0e9ac57a6f18301d1d62986d 100644 --- a/qsar_biodeg/019d7369-5170-7d75-8932-e2a5778ee70b/dataset_metadata.dataset-mold-v1.json +++ b/qsar_biodeg/019d7369-5170-7d75-8932-e2a5778ee70b/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We added semantic meaningful feature names.\n- Anomaly: several features are numeric-ordinal in nature but it is unclear if they are categorical features.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/qsar_fish_toxicity/019d7369-62fc-7ea1-9c58-8ac89144f820/dataset_metadata.dataset-mold-v1.json b/qsar_fish_toxicity/019d7369-62fc-7ea1-9c58-8ac89144f820/dataset_metadata.dataset-mold-v1.json index a5ddd8a24e64f769a9e47cf6cb076e6281962144..fedcbecbd250b61ede43ceb9bfa8b3f1832fbc98 100644 --- a/qsar_fish_toxicity/019d7369-62fc-7ea1-9c58-8ac89144f820/dataset_metadata.dataset-mold-v1.json +++ b/qsar_fish_toxicity/019d7369-62fc-7ea1-9c58-8ac89144f820/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We assigned descriptive column names following the original dataset documentation.\n- Anomaly: the data contains a lot of duplicates (15%) when ignoring the target feature.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/qsar_tid_11/019d7369-7ffa-7d6a-b578-fa8c63c7adb6/dataset_metadata.dataset-mold-v1.json b/qsar_tid_11/019d7369-7ffa-7d6a-b578-fa8c63c7adb6/dataset_metadata.dataset-mold-v1.json index a9f64ff7473942c391d88def9a79d170ce314f50..762086f2a1b1ef9e140ebb02f4823c14814987ba 100644 --- a/qsar_tid_11/019d7369-7ffa-7d6a-b578-fa8c63c7adb6/dataset_metadata.dataset-mold-v1.json +++ b/qsar_tid_11/019d7369-7ffa-7d6a-b578-fa8c63c7adb6/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We drop the \"MOLECULE_CHEMBL_ID\" column.\n- We shuffle the data as it has an original distribution shift.\n- Only one row has a negative value (-6.24) and is thus extremely out-of-distribution, which is likely results from the target being log scaled in the original domain. We remove this row as it is unclear if this is a scaling or labeling error, but we judge it is in all cases a data artifact.\n- We do not log scale the target (it could make sense for this distribution), as the original label is already log transformed.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/regensburg_pediatric_appendicitis/019d7370-1dc6-7c1e-8ff1-80ddeb4c4207/dataset_metadata.dataset-mold-v1.json b/regensburg_pediatric_appendicitis/019d7370-1dc6-7c1e-8ff1-80ddeb4c4207/dataset_metadata.dataset-mold-v1.json index d28bc5034e39dbe58d47d712c4cded9f7884dd48..060ddf64afac248407dcb57a66ba9635108439d1 100644 --- a/regensburg_pediatric_appendicitis/019d7370-1dc6-7c1e-8ff1-80ddeb4c4207/dataset_metadata.dataset-mold-v1.json +++ b/regensburg_pediatric_appendicitis/019d7370-1dc6-7c1e-8ff1-80ddeb4c4207/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the tabular meta-data used in the paper. We create only one task out of the dataset using one of the three available classes. We note that the main use case of the paper was image-based modelling and comparing to tabular baselines with image-to-tabular-radiometric. The authors also published the tabular data, which we use here.\n\n- Note, for this dataset is was hard to choose the label given the tabular data. We picked \"Severity\" in the end. \"Diagnosis\" is a noisy ground truth for all cases but the ones that had surgery. It was either deterministically determined by the AS score, or during surgery. \"Management\" reflects the decision and opinion of senior pediatric surgeon. Only \"Severity\" represents a ground truth as all cases that were server, were either clearly servery or did not even have appendicitis.\n - Warning: We tried to clean the data as good as possible but we suspect it might still contain some features leaking the severity, but we lack the medical expertise to identify them. Thus, I would not be surprised if we had to remove this dataset later again.\n- The data contains string data, but only for a very small number of patients and the text is often seemingly very useless. The features add most likely no value, but we keep them in for the pipeline/model to ignore if needed.\n- We drop the few patients that did not have an ultrasound to avoid them being leaking due to missing features.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/rossmann_store_sales/019db4fb-6d9e-7c15-b8cb-1bf97ea88782/dataset_metadata.dataset-mold-v1.json b/rossmann_store_sales/019db4fb-6d9e-7c15-b8cb-1bf97ea88782/dataset_metadata.dataset-mold-v1.json index 7120c5d03cde2faa6f04fee5456b1d03d3756260..665bd7a257897c305c168af8af6cd8de4fc3953c 100644 --- a/rossmann_store_sales/019db4fb-6d9e-7c15-b8cb-1bf97ea88782/dataset_metadata.dataset-mold-v1.json +++ b/rossmann_store_sales/019db4fb-6d9e-7c15-b8cb-1bf97ea88782/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\n- We merge the store data with the train data.\n- The data is from a Kaggle competition with a temporal split, so the test data cannot be used for unsupervised approachs.\n- The winning solution is public: https://storage.googleapis.com/kaggle-forum-message-attachments/102102/3454/Rossmann_nr1_doc.pdf.\n- The competition used 48 days for testing. The description says \"Rossmann store managers are tasked with predicting their daily sales for up to six weeks in advance\". Therefore, we define the same horizon for defining our test splits and add one day between train/test as a planning gap.\n- We transform the date column to datetime format.\n- We exclude days when the store is closed since they all have zero sales and are trivial to predict. We drop the Open column.\n- We assign categorical data type to the Store column and the DayOfWeek column.\n- For the train data, the # of customers per day is given, which is not available at the prediction point for the test data. The 1st-place solution says that one model variation computed recent-history features on the number of customers instead of sales, and also used store-level aggregates such as average sales per customer. However this would mean that we have to define split-specific splits. Since many high performing solutions did not utilize customer information, we drop the column.\n- The StateHoliday column has two encodings for the entry 0, one as number and one as object. We combine them into one category.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/santander_customer_satisfaction/019d7370-56f7-7b35-ba45-c77b1e3c531b/dataset_metadata.dataset-mold-v1.json b/santander_customer_satisfaction/019d7370-56f7-7b35-ba45-c77b1e3c531b/dataset_metadata.dataset-mold-v1.json index 19f5f4614d0d8d8a827e039ce795926f1752d674..0a8a6c44513d484a70f55601e8e2d1c5d56de3f9 100644 --- a/santander_customer_satisfaction/019d7370-56f7-7b35-ba45-c77b1e3c531b/dataset_metadata.dataset-mold-v1.json +++ b/santander_customer_satisfaction/019d7370-56f7-7b35-ba45-c77b1e3c531b/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized, so feature meanings are unknown.\n- We remove duplicated columns and zero-variance columns.\n- We add a zero-count feature (number of zeros per row). #1 Kaggle experts also applied a lot of other features and feature selection but each experts used something different (https://www.kaggle.com/competitions/santander-customer-satisfaction/writeups/1-leustagos-3rd-place-solution).\n- We drop duplicates as their origin is unclear and they could create data leaks when splitting the training data.\n- We replace -999999 in var3 with with nan.\n- We keep the data otherwise as it. Several columns might be categorical and contain encoded nan values but we are not able to determine this due to anonymization.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/santander_customer_transaction_prediction/019d7373-33b4-7002-948b-c14d1fffe341/dataset_metadata.dataset-mold-v1.json b/santander_customer_transaction_prediction/019d7373-33b4-7002-948b-c14d1fffe341/dataset_metadata.dataset-mold-v1.json index 13943946cc4c7b598bd1f340ddb42f345ee262d5..7c51cb3e9d756e4f743de9e448161fd2fe063f61 100644 --- a/santander_customer_transaction_prediction/019d7373-33b4-7002-948b-c14d1fffe341/dataset_metadata.dataset-mold-v1.json +++ b/santander_customer_transaction_prediction/019d7373-33b4-7002-948b-c14d1fffe341/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized, so feature meanings are unknown.\n- The data seems to be generated or created for the competition. Most features are perfectly normally distributed.\n- Kaggle experts found various uniqueness-based features and had to filter fake samples from the test data. We apply the uniqueness-based features.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/santander_transaction_value/019d7373-d139-7c8f-be3e-579006c05fad/dataset_metadata.dataset-mold-v1.json b/santander_transaction_value/019d7373-d139-7c8f-be3e-579006c05fad/dataset_metadata.dataset-mold-v1.json index 4df82ce8751ae219728a70572ad7830ce99c75f8..4904281f88348cce0b281a557db33423d1db4220 100644 --- a/santander_transaction_value/019d7373-d139-7c8f-be3e-579006c05fad/dataset_metadata.dataset-mold-v1.json +++ b/santander_transaction_value/019d7373-d139-7c8f-be3e-579006c05fad/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the train.csv from Kaggle.\n\n- The data has been anonymized.\n- We follow the preprocessing by Tschalzev et al. (https://arxiv.org/abs/2407.02112), which follows https://www.kaggle.com/competitions/santander-value-prediction-challenge/discussion/63919\n- Since we restrict to the train.csv, we do not exploit the test leak of the competition.\n- The expert preprocessing aggregates over feature groups per row to create the new features for the final dataset. Feature groups for aggregation are found by searching for chains of features (?).\n- The data contains a very small number of duplicates. We remove them to avoid leakage in our own splits.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/sat11_hand_algo_runtime/019d7392-268a-7998-a20d-29e9a2236dcd/dataset_metadata.dataset-mold-v1.json b/sat11_hand_algo_runtime/019d7392-268a-7998-a20d-29e9a2236dcd/dataset_metadata.dataset-mold-v1.json index bdf303b21788256f6822dbc1b8df1723a5d9a980..259299a503cbaa4832acaa2ac03588dc64a81228 100644 --- a/sat11_hand_algo_runtime/019d7392-268a-7998-a20d-29e9a2236dcd/dataset_metadata.dataset-mold-v1.json +++ b/sat11_hand_algo_runtime/019d7392-268a-7998-a20d-29e9a2236dcd/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe get the data from ASlib and merge them into one file.\n\n- Algorithm selection can be solved with many methods (pairwise classification, hierarchical regression, multi-label, ...). We follow the concept from Pulatov et al. (https://proceedings.mlr.press/v188/pulatov22a.html) and learn one single, unified regression model that goes from (instance_features, algorithm_features) -> runtime.\n- The description says 'If features are \"?\", the instance was solved during feature computation.' but from the status file we can see that no task was presovled. So it is unclear what this is referring to. We believe that in all cases we have nan values for features, it is a case of the feature running into a memout, timeout, or crash as stated by the description.\n- This regression task is special as it contains censored regression values (capped to some max value) and the model needs to learn this.\n- We log scale the target.\n- We drop duplicated columns and remove features that leak the target or are not relevant for the predictive task.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/sberbank_housing_market_forecasting/019d9279-a91a-73ac-a221-72a6aab17df0/dataset_metadata.dataset-mold-v1.json b/sberbank_housing_market_forecasting/019d9279-a91a-73ac-a221-72a6aab17df0/dataset_metadata.dataset-mold-v1.json index faf824c02acfd1ced56d93c9491f81a635ce123c..452d7a7d9eea356a66d69d17f7dc1af45c4621f8 100644 --- a/sberbank_housing_market_forecasting/019d9279-a91a-73ac-a221-72a6aab17df0/dataset_metadata.dataset-mold-v1.json +++ b/sberbank_housing_market_forecasting/019d9279-a91a-73ac-a221-72a6aab17df0/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe follow the preprocessing from TabRed (https://github.com/yandex-research/tabred/tree/main/preprocessing#sberbank-housing-market-forecasting).\n\n- We drop two duplicated columns with the same values as other columns: \"0_6_all\", \"7_14_all\"\n- We drop the ID column as it does not provide information.\n- Note, the dataset contains a lot of spatial feature and even more already decoded spatial information (like distances to various points of interest).\n- We drop all rows that have a build_year higher than their own timestamp of being sold. It is unclear from the original data what kind of house market data this comes from. Plus this drops around 150 samples for which the build_year is incorrectly encoded as integer derived from a range like \"2000-2005\". \n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/sdss_17/019d7369-942d-72c0-b327-e812c4b56b7e/dataset_metadata.dataset-mold-v1.json b/sdss_17/019d7369-942d-72c0-b327-e812c4b56b7e/dataset_metadata.dataset-mold-v1.json index c9ce9f9ebaa549912f6b2f0c5021a302d2419a5b..e7f1dca524c3f247d36c2c29b30a2f858589dd8e 100644 --- a/sdss_17/019d7369-942d-72c0-b327-e812c4b56b7e/dataset_metadata.dataset-mold-v1.json +++ b/sdss_17/019d7369-942d-72c0-b327-e812c4b56b7e/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the target feature.\n- We dropped duplicates based on \"obj_ID\" to avoid target leakage from subgroups. \n- We dropped several (ID-like) meta-features that seem to be not part of the predictive task.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/seismic_bumps/019d7369-a6b4-7060-a145-dc996b59cf1b/dataset_metadata.dataset-mold-v1.json b/seismic_bumps/019d7369-a6b4-7060-a145-dc996b59cf1b/dataset_metadata.dataset-mold-v1.json index de611f09fbac5de6a20d609b12e73762f3cab29e..7bf0eb658cc6e350dadb214526bec909c6e49638 100644 --- a/seismic_bumps/019d7369-a6b4-7060-a145-dc996b59cf1b/dataset_metadata.dataset-mold-v1.json +++ b/seismic_bumps/019d7369-a6b4-7060-a145-dc996b59cf1b/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We renamed the target feature and reversed the values' ordinal encoding.\n- We drop the constant columns \"nbumps6\", \"nbumps7\", and \"nbumps89\".\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/sepsis_prediction/versions/019d7391-f36e-72f8-89d7-f7ca71725034/dataset_metadata.dataset-mold-v1.json b/sepsis_prediction/versions/019d7391-f36e-72f8-89d7-f7ca71725034/dataset_metadata.dataset-mold-v1.json index 3519e0235c3ee1a588988b310300087d9d545248..b1a92eed068fc2492833733d454bbe195559825f 100644 --- a/sepsis_prediction/versions/019d7391-f36e-72f8-89d7-f7ca71725034/dataset_metadata.dataset-mold-v1.json +++ b/sepsis_prediction/versions/019d7391-f36e-72f8-89d7-f7ca71725034/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with all files from Kaggle.\n\nThe original data has one file per user that was already preprocessed to one .csv file by the competition creators. Here we start with the preprocessed single .csv file. The data is non-IID in nature based on the groups of patients from different hospitals. The data that is grouped per patient (\"Patient_ID\"). These groups are also temporal in nature, but this temporal dependency is irrelevant as teh task is to predicts for one full patient (i.e., no refitting given patient information).\nIn the original competition, one had to predict for unseen patients from the existing hospitals and also for a new hidden hospital. In the public data, we only have two hospitals, (A) and (B). Given the limited data, we decide not to simulate a domain shift as we could not \"train\" for domain shift. Thus, we simulate only a normal grouped non-IID scenario. That is, we use all patients from both hospitals for training and testing, but ensure that the splits are grouped by patient ID. Thus, we simulate what would happen if someone trains a model on data from two hospitals and uses this to predict for other patients from these hospitals. This decision is also amplified by the large gap in performance for the unseen hospital in the competition (Report, Table 3), clearly pointing to a domain shift that is out-of-scope for this task. Note, we do not have temporal information of the order of patients, thus, we cannot simulate to only predict for patients \"from the future\". We believe this does not introduce any data leakage for this dataset.\n\n- Patients with an ID larger than 100_000 are from hospital (B), while patients with an ID smaller than 100_000 are from hospital (A). We add this indicator into our data and then reset the Patient_IDs to be continuous increasing integers.\n- The data consists of a lot of missing values due to missing measurements.\n- Note, hour is a reset time index per patient. ICULOS is similar, but with an offset. We keep both as this can show that we have truncated data for a patient.\n- We reverse the ordinal encoding of Gender.\n- We mark features as categorical where appropriate.\n- For each patient, we predict the sepsis status over time. So it is more or less a transformed survival task (also in the original challenge).\n", "version_from_unique_name": "sepsis_prediction", "version_comment": "\nWe sub-sample to 1 million train rows after splitting and test 250k rows.\n", - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/sepsis_survival_minimal_clinical_records/019d7373-e475-7208-bd58-accca4615b01/dataset_metadata.dataset-mold-v1.json b/sepsis_survival_minimal_clinical_records/019d7373-e475-7208-bd58-accca4615b01/dataset_metadata.dataset-mold-v1.json index 8a33e6224a02a23775b20e984a3060e58faf2051..ecf63bc00305987ecf479db2dfe6fcf5d2942148 100644 --- a/sepsis_survival_minimal_clinical_records/019d7373-e475-7208-bd58-accca4615b01/dataset_metadata.dataset-mold-v1.json +++ b/sepsis_survival_minimal_clinical_records/019d7373-e475-7208-bd58-accca4615b01/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe use only the primary cohort from Norway and ignore the subset and the small data (137 samples) from the South Korea cohort.\n\n- The data has only 3 features. Moreover, any other information (such as timestamps from the collection) are not provided. But this should not limit the applicability of the data, as there should be no temporal leakage.\n- We reverse the ordinal encoding of gender and the target variable.\n- The data has a lot of naturally occurring duplicates. We keep them in as they make sense they would exist and leave it to the pipeline or methods to handle.\n- The alternative subset (the study cohort in the original paper) contains only patients that had a sepsis according to the official Sepsis-3 definition. The paper discusses this in more detail and opts for using/testing all cohorts. We could introduce several datasets from this source but this would bias the benchmark too much towards this specific data distribution. We stick to using all data from the primary cohort.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/sf_permit_time/019d738d-80c6-71a6-99f0-e3e0955ce2d0/dataset_metadata.dataset-mold-v1.json b/sf_permit_time/019d738d-80c6-71a6-99f0-e3e0955ce2d0/dataset_metadata.dataset-mold-v1.json index 086bff7c2487f53394bd8b76fea410c36068c7a6..6867a69167511edb5ba0ceb9d160bc74580c5aa1 100644 --- a/sf_permit_time/019d738d-80c6-71a6-99f0-e3e0955ce2d0/dataset_metadata.dataset-mold-v1.json +++ b/sf_permit_time/019d738d-80c6-71a6-99f0-e3e0955ce2d0/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe simulate the task of predicting the days (in float) it takes to issue the permit. We add the special use case, that we assume the model is only used to predict for permits that take longer than one day to be issued. We do this, as waiting for one day seems very reasonable. Plus, the data contains several unresolvable data errors when the permit was issued on the same day it was filed.\n\n- ALthough we donwloaded data\n- We drop miscellaneous permits (those that start with \"M\") as they are not of interest for our task. Compared to normal permits, these are usually automatically or very quickly approved and thus are not relevant to predict the time it takes to issue the permit. Moreover, the represent a significant distribution shift compared to the rest of the data.\n- We drop the permit type ordinal encoding and the creation date of the permit in the tracking system as other dates are more accurate.\n- A large number of descriptions are from standard phrases (they appear in the same way multiple times). so we add a new column that indicates whether the description is from a standard phrase or not through a categorical variable. We define a standard phrase as a description that appears more than 100 times in the dataset.\n- The 'Current Status' is the last status update of the permit. Note, that for miscellaneous permits, the permit is never completed. We filter to permits that have been issued for our task. Thus, we select all permits that are issued or completed.\n- There are several features that have almost no information (up to only 20 non nan values). We keep this and let the pipeline decide what to do with them.\n- We only keep on permit per primary address (filter based on Primary Address Flag) following https://data.sfgov.org/Housing-and-Buildings/Building-Permits-Deduplicated-on-Primary-Address/f2jc-ivnc\n- We drop all permits without a location, as every permit requires a location by definition, thus these are likely data errors.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/south_africa_coronary_heart_disease/019d7373-f750-7ecd-8cda-f16d5d36be62/dataset_metadata.dataset-mold-v1.json b/south_africa_coronary_heart_disease/019d7373-f750-7ecd-8cda-f16d5d36be62/dataset_metadata.dataset-mold-v1.json index a8e7cd7d1834035096dad0a074b129d5f60a0ee3..73028f4cc27070969cadadaa09054ee4b0422f42 100644 --- a/south_africa_coronary_heart_disease/019d7373-f750-7ecd-8cda-f16d5d36be62/dataset_metadata.dataset-mold-v1.json +++ b/south_africa_coronary_heart_disease/019d7373-f750-7ecd-8cda-f16d5d36be62/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the data from Kaggle.\n\n- We remove the row id column.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/splice/019d7369-b922-718a-adbf-099484b95908/dataset_metadata.dataset-mold-v1.json b/splice/019d7369-b922-718a-adbf-099484b95908/dataset_metadata.dataset-mold-v1.json index 05f333337c3d394344adb01746095f1f6096dab2..619b26d1ea14459fb8f17dc10823da2d7ebb0575 100644 --- a/splice/019d7369-b922-718a-adbf-099484b95908/dataset_metadata.dataset-mold-v1.json +++ b/splice/019d7369-b922-718a-adbf-099484b95908/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We tabularized the fixed-size DNA sequences from the original dataset based on the nucleotide position.\n- We removed the instance ID.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/student_portuguese_performance/019d7374-09f4-79ea-a27f-1ad571967f90/dataset_metadata.dataset-mold-v1.json b/student_portuguese_performance/019d7374-09f4-79ea-a27f-1ad571967f90/dataset_metadata.dataset-mold-v1.json index 8806fde45c47c70171460e2686c6f10cccda1e5c..c294e96b598e7bde14b2870ff26338b7bbbf919c 100644 --- a/student_portuguese_performance/019d7374-09f4-79ea-a27f-1ad571967f90/dataset_metadata.dataset-mold-v1.json +++ b/student_portuguese_performance/019d7374-09f4-79ea-a27f-1ad571967f90/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe use the data from UCI.\n\n- We drop G1 and G2 to simulate the task to predict the final grade from the base characteristic.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/superconductivity/019d7369-cca6-7baa-b958-4ec60c44e1c1/dataset_metadata.dataset-mold-v1.json b/superconductivity/019d7369-cca6-7baa-b958-4ec60c44e1c1/dataset_metadata.dataset-mold-v1.json index 286aa65c557c4dfacef58214bf326aa0ff9b2373..7cba2ab84c1cb275b1f184e359eef5c802664ce6 100644 --- a/superconductivity/019d7369-cca6-7baa-b958-4ec60c44e1c1/dataset_metadata.dataset-mold-v1.json +++ b/superconductivity/019d7369-cca6-7baa-b958-4ec60c44e1c1/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- Anomaly: the data has a lot of duplicates (29%) when ignoring the target feature\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/taiwanese_bankruptcy_prediction/019d7369-df9e-716f-974e-19cc910078ee/dataset_metadata.dataset-mold-v1.json b/taiwanese_bankruptcy_prediction/019d7369-df9e-716f-974e-19cc910078ee/dataset_metadata.dataset-mold-v1.json index 1224ea297a73e4ba442a2ba4af12cbf86dc5ab60..c293ab5113261b07f2ee5eaf6d9ace94c3642d6f 100644 --- a/taiwanese_bankruptcy_prediction/019d7369-df9e-716f-974e-19cc910078ee/dataset_metadata.dataset-mold-v1.json +++ b/taiwanese_bankruptcy_prediction/019d7369-df9e-716f-974e-19cc910078ee/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We map binary Bankrupt column to yes/no\n- We rename the features to be shorter and have no special characters.\n- We drop the constant column \"Net_Income_Flag\".\n- We drop two duplicated features (same values as other features).\n- Anomaly: the paper introducing the dataset has more features, but these do not seem to be in the dataset. \n- Anomaly: the data is temporal but all features are time-invariant. This might still have a latent, non-verifiable temporal component, which we ignore for our predictive IID task.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/telemonitoring_parkinsons_biomedical_voice_measurements/019d7391-4b65-7d0e-a950-684a098c3df9/dataset_metadata.dataset-mold-v1.json b/telemonitoring_parkinsons_biomedical_voice_measurements/019d7391-4b65-7d0e-a950-684a098c3df9/dataset_metadata.dataset-mold-v1.json index 0fa741e31b7ef00a396b12804e00ed48ed9ae145..0e528dc98166b3539f7f0a057479b66e4263193e 100644 --- a/telemonitoring_parkinsons_biomedical_voice_measurements/019d7391-4b65-7d0e-a950-684a098c3df9/dataset_metadata.dataset-mold-v1.json +++ b/telemonitoring_parkinsons_biomedical_voice_measurements/019d7391-4b65-7d0e-a950-684a098c3df9/dataset_metadata.dataset-mold-v1.json @@ -16,6 +16,5 @@ "curation_comments": "\nWe start with the data from UCI.\n\nWe aim to simulate the task of predicting the UPDRS score of Parkinson's patients based on their voice measurements over time. This represents the task of only getting the voice measurements to judge the UPDRS. We assume we have no measurement of a patient to make this call. Thus, we have a grouped data task, where we have to holdout entire patients over time. We thus test for one patients multiple time points and repeated measurements at once.\n\n- We use total_UPDRS as target. Note that real UPDRS values were were obtained at baseline, three-month and six-month trial period and all other weekly cases were interpolated. Thus, the ground truth is not perfect and just an estimate based on knowing the future. We reduce it to three values (first, medium <130 days, last), as we have no other ground truth.\n- We have several measurements per day for each patient when they were measured. We have some edge cases with negative test time as they got measured before the study started.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/thyroid_discordant/019d7374-1c19-79e6-986d-2afd2c16101e/dataset_metadata.dataset-mold-v1.json b/thyroid_discordant/019d7374-1c19-79e6-986d-2afd2c16101e/dataset_metadata.dataset-mold-v1.json index 6dc13eb64569c33392405d74843bd00593ac6552..95f793220e52d7b16ed1033783395cab21e6c958 100644 --- a/thyroid_discordant/019d7374-1c19-79e6-986d-2afd2c16101e/dataset_metadata.dataset-mold-v1.json +++ b/thyroid_discordant/019d7374-1c19-79e6-986d-2afd2c16101e/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the .data file from UCI and add columns and encode nan values.\n\n- Note, the data contains NaN indicators by default (e.g. TSH_measured). We keep them.\n- We drop/ignore the referral_source columns as it indicates the original of the data and thus might leak information or makes the model learn sub-group related behavior that is not the target/task of interest.\n- We correct the label column by removing the patient IDs.\n- We drop 60 duplicated rows and remove constant columns.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/tour_travels_churn/019d7374-2df2-7b85-94a6-3ec9d1796737/dataset_metadata.dataset-mold-v1.json b/tour_travels_churn/019d7374-2df2-7b85-94a6-3ec9d1796737/dataset_metadata.dataset-mold-v1.json index 6f2befbbecbaa800b0886d73a0b73b3003a7c41e..adde8be5b560ad0c09decbac6aaf7f3a92a9693b 100644 --- a/tour_travels_churn/019d7374-2df2-7b85-94a6-3ec9d1796737/dataset_metadata.dataset-mold-v1.json +++ b/tour_travels_churn/019d7374-2df2-7b85-94a6-3ec9d1796737/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nThe source of the data from Kaggle is unknown but the data distributions look reasonable enough to use. But I would also not be surprised if we find out the data is artificially created.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/video_game_fps_prediction/019d7392-6ce5-72bc-b598-193597825a66/dataset_metadata.dataset-mold-v1.json b/video_game_fps_prediction/019d7392-6ce5-72bc-b598-193597825a66/dataset_metadata.dataset-mold-v1.json index 50309a411a110fd58b43365291fa52cde6253d70..d036625842466d56a02b238e4ca18d50661c60c5 100644 --- a/video_game_fps_prediction/019d7392-6ce5-72bc-b598-193597825a66/dataset_metadata.dataset-mold-v1.json +++ b/video_game_fps_prediction/019d7392-6ce5-72bc-b598-193597825a66/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe start with the version from OpenML.\n\n- Following the original work, we do not use CPU and GPU name as features to learn to generalize across the features of CPUs and GPUs.\n- The OpenML version is only data from the FPS benchmark. Thus, we perform on a subset of data points from the original study.\n- We diverge in the way we split from the original work. The task could be split on games (predict FPS from past values of other games to a new game), but there are not features to allow generalization across games. Instead, we split on hardware configuration (CPU-GPU combination). That is, the task is to be given a set of hardware configurations, for which we have values for a set of games, and then we want to predict the performance of a new hardware configuration on the same set of games. The features in the dataset allow this kind of prediction task and it makes sense to use to save on benchmark costs and interpolate missing hardware combinations -- a surrogate model for hardware performance. Along these lines, it might also be a valid task when one assume we have samples of some hardware configurations for some games and want to interpolate the rest. But predicting for a new, unseen hardware configuration seems more realistic.\n- The data has measurement for two cases, the game on medium and the game on max settings. It is unclear if one would build one or two models for this case. In theory, this could also be seen as two labels. We drop the medium setting only keep one task, as this is a less ambiguous task definition.\n- We normalize the FPS to ensure that all target values are on the same scale. By default, FPS values can be in vastly different scales depending on the game, having different distributions. Since, we care about comparing hardware, we can simply predict performance relative to a baseline, as in, how much better is the hardware than some default hardware setting. We normalize and then drop this configuration. In other words, one assumes we always have benchmarked the FPS values for this configuration and all others could be potentially a target during one of the splits. We normalizes with respect to \"NVIDIA GeForce GTX 1060 5 GB\" and \"Intel Core i5-8400\". We compute the relative ratio to this baseline configuration. We then log scale this ratio.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/video_transcoding_time_prediction/019d7392-553f-7bb2-b0f4-0d80ea446f63/dataset_metadata.dataset-mold-v1.json b/video_transcoding_time_prediction/019d7392-553f-7bb2-b0f4-0d80ea446f63/dataset_metadata.dataset-mold-v1.json index 7798c7feee3a50cb665ec1c1ae78294fc27a67f5..076074cd035051c195733c09df602be3083f97bf 100644 --- a/video_transcoding_time_prediction/019d7392-553f-7bb2-b0f4-0d80ea446f63/dataset_metadata.dataset-mold-v1.json +++ b/video_transcoding_time_prediction/019d7392-553f-7bb2-b0f4-0d80ea446f63/dataset_metadata.dataset-mold-v1.json @@ -15,6 +15,5 @@ "curation_comments": "\nWe use the tabular prediction task from the UCI archive, which is a video transcoding time prediction task. The data is non-IID and grouped by video, with multiple transcoding measurements per video. The target variable is the transcoding time, and the features include various characteristics of the videos and transcoding settings.\n\n- Some features on the videos (like URL and category) are not part of this dataset as we got it from UCI, but this sounds like a good idea for the task.\n- The data includes several transcoding per video (also with different output target formats), which creates a non-IID setting. We will use the video id as the group label, and we will make sure to split the data such that all transcoding measurements for a given video are in the same split (train/test).\n- We log scale the target variable.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/website_phishing/019d7369-f10c-78e6-bd19-b8b0a72399fe/dataset_metadata.dataset-mold-v1.json b/website_phishing/019d7369-f10c-78e6-bd19-b8b0a72399fe/dataset_metadata.dataset-mold-v1.json index fb6e6f21c56419d72953d38fec8d60728308a972..344fc91cd4ae2ec93c459cd83156caab1671f2fe 100644 --- a/website_phishing/019d7369-f10c-78e6-bd19-b8b0a72399fe/dataset_metadata.dataset-mold-v1.json +++ b/website_phishing/019d7369-f10c-78e6-bd19-b8b0a72399fe/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We reversed the ordinal encoding of the original data.\n- We renamed the target feature to be more meaningful.\n- Anomaly: the data has many duplicates (~50%).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/wids_diabetes_mellitus/019d7374-50ba-7416-9d10-a526040930ee/dataset_metadata.dataset-mold-v1.json b/wids_diabetes_mellitus/019d7374-50ba-7416-9d10-a526040930ee/dataset_metadata.dataset-mold-v1.json index b4e8e6383a25ac2344e47392b03ab04d0d214a7c..81b15035da1f3e7c0cf3f1b248185eb4be19f758 100644 --- a/wids_diabetes_mellitus/019d7374-50ba-7416-9d10-a526040930ee/dataset_metadata.dataset-mold-v1.json +++ b/wids_diabetes_mellitus/019d7374-50ba-7416-9d10-a526040930ee/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the Kaggle dataset.\n\n- Data from the prior year's hackathon (2020) also used the very similar data but with older data and different splits. But it seems the feature engineering was transferable across years.\n- We drop encounter_id as it is an identifier with no temporal information, and from the competition there seems to be no temporal or index-based leakage.\n- We follow the top solutions and use normal random splits, instead of stratifying by hospital_id or ICU ID. Moreover, with the data for which we have labels, we have an even larger overlap for hospital_id and ICU ID even more than the train-test split on Kaggle. Lastly, we do not need to decode the hospital_ID based on the ICU ID.\n- A hospital can have multiple ICUs, so there is a strong co-correlation between hospital_id and ICU ID, but it is not identical.\n- We drop rows with age=0, as these seems to be 30 faulty entries (patients have a large height at age 0).\n- Kaggle experts found that several features were likely clipped to a min/max value based on investigating the histograms of the data. We noticed this too. Following the experts, we add a flag for these features indicating if the value was clipped or not.\n- Several Kaggle experts also re-computed the BMI. We also noticed that for 2770 rows, the BMI did not match the real BMI based on the weight and height. We drop all of these cases, as it is unclear to us what the cause of this error is. Given the small fraction, there is either a data entry error, or some values got corrected manually.\n- We found and dropped three duplicated columns that provided no new information.\n- We keep medical codes as string, as these could be more meaningfully decoded.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/wine_quality/019d736a-03ee-7297-95da-6bfa50c97f79/dataset_metadata.dataset-mold-v1.json b/wine_quality/019d736a-03ee-7297-95da-6bfa50c97f79/dataset_metadata.dataset-mold-v1.json index 0f913272c9adccbe8e8c0d0bc8278cd644b49fd7..31710f4db18289a7abefa4875f1b7a8ec99b7012 100644 --- a/wine_quality/019d736a-03ee-7297-95da-6bfa50c97f79/dataset_metadata.dataset-mold-v1.json +++ b/wine_quality/019d736a-03ee-7297-95da-6bfa50c97f79/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\n- We combine the original datasets for red and white wine into a single dataset with an additional column indicating the type of wine (red or white).\n- We treat the task as a regression problem, following the original work and because the target is the median wine quality (of at least 3 evaluations by experts).\n- Anomaly: the data has a high number of duplicates (18%).\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +} diff --git a/wine_world_cost/019d7374-6961-7fb3-9ca1-2e20e35fbbb0/dataset_metadata.dataset-mold-v1.json b/wine_world_cost/019d7374-6961-7fb3-9ca1-2e20e35fbbb0/dataset_metadata.dataset-mold-v1.json index 792999f87dce180b0d1a93ce2756f43449a12d43..8285672310cfb9f036c74a9176b40cd7b07c0250 100644 --- a/wine_world_cost/019d7374-6961-7fb3-9ca1-2e20e35fbbb0/dataset_metadata.dataset-mold-v1.json +++ b/wine_world_cost/019d7374-6961-7fb3-9ca1-2e20e35fbbb0/dataset_metadata.dataset-mold-v1.json @@ -14,6 +14,5 @@ "curation_comments": "\nWe start with the Kaggle version. We resolve several issues from the web scrapper.\n\n- We create a task to predict the price per bottle. We remove the 11 rows that have price information per case or each.\n- Create the price target by parsing the price column.\n- We map capacity to numeric values in ML.\n- We parse the ABV to a numeric value\n- We parse the vintage to a numeric date value, ensuring that each wine maps to the oldest year in its special syntax.\n- We log scale the target, as the price distribution is very skewed.\n", "version_from_unique_name": null, "version_comment": null, - "local_data_directory_base": "/home/lennart_priorlabs_ai/code/large_data_ensemble/data-foundry/local-data-warehouse", "type_adapter_id": "dataset-mold-v1" -} \ No newline at end of file +}