from typing import Any, Dict, List, Tuple import numpy as np import pandas as pd import streamlit as st def validate_and_formate_columns(datasets, columns, load_options): _validate_selection(columns, load_options) datasets = _validate_date_and_format(datasets, columns["date_column"], load_options) _validate_numeric(datasets, columns["quantity_column"]) datasets = _combine_datetime(datasets, columns) datasets = _merge_files(datasets, columns, load_options) def filter_selected_options(datasets, columns, load_options): data = datasets["data"] if load_options.get("forecast_item") != None: data = data[data[columns["item_column"]] == load_options["forecast_item"]] data = data.reset_index(drop=True) if load_options.get("forecast_branch") != None: data = data[data[columns["branch_column"]] == load_options["forecast_branch"]] data = data.reset_index(drop=True) datasets["data"] = data return datasets def valid_rows_number(datasets, load_options): if load_options["forecast"]: if len(datasets["forecast_df"]) < 5: st.sidebar.error("The number of rows should be 5 or more to forecast.") st.session_state.clicked = False st.stop() elif load_options["evaluate"]: if len(datasets["train_df"]) < 5: st.sidebar.error("The number of rows should be 5 or more to forecast.") st.session_state.clicked = False st.stop() # if there are two seperate files, merge them def _merge_files(datasets, columns, load_options): if load_options["two_files"]: try: datasets["data"] = pd.merge(datasets["recipe_data"], datasets["data"], left_on=columns["recipe_column"], right_on=columns["sale_column"]) except Exception as e: st.sidebar.error("Cannot merge the two files, make sure the correct ID columns are selected.") st.session_state.clicked = False st.stop() return datasets # Combine date and time if in seperate columns def _combine_datetime(datasets, columns): data = datasets["data"] if columns["time_column"] != '': if pd.api.types.is_numeric_dtype(data[columns["time_column"]]): data[columns["date_column"]] = data[columns["date_column"]] + pd.to_timedelta(data[columns["time_column"]], unit='H') else: try: data[columns["time_column"]] = pd.to_datetime(data[columns["time_column"]]).dt.time data[columns["date_column"]] = data.apply(lambda x: pd.datetime.combine(x[columns["date_column"]], x[columns["time_column"]]), axis=1) except Exception as e: st.sidebar.error("Time column is not in datetime format.") st.session_state.clicked = False st.stop() datasets["data"] = data return datasets def _validate_selection(columns, load_options): if columns["quantity_column"]=='' or columns["date_column"]=='': st.sidebar.warning("Select all of the required columns.") st.session_state.clicked = False st.stop() if load_options["two_files"]: if (columns["quantity_per_item_column"]=='' or columns["recipe_column"]=='' or columns["sale_column"]==''): st.sidebar.warning("Select all of the required columns.") st.session_state.clicked = False st.stop() def _validate_numeric(datasets, column): data = datasets["data"] if not pd.api.types.is_numeric_dtype(data[column]): st.sidebar.error(f"{column} is not a numeric column.") st.session_state.clicked = False st.stop() def _validate_date_and_format(datasets, column, load_options): data = datasets["data"] if not pd.api.types.is_datetime64_any_dtype(data[column]): if pd.api.types.is_numeric_dtype(data[column]): st.write("point 1") st.sidebar.error(f"{column} is not in datetime format.") st.session_state.clicked = False st.stop() try: data[column] = pd.to_datetime(data[column], **load_options["date_params"]) except Exception as e: st.sidebar.error(f"{column} is not in datetime format.") st.session_state.clicked = False st.stop() datasets["data"] = data return datasets