Dataset for: Redefining Multi-Target Weather Forecasting with a novel Deep Learning model: Hierarchical Temporal Convolutional Long Short-Term Memory with Attention (HTC-LSTM-Attn) in Bangladesh

The climate data set was compiled monthly for Bangladesh, from January 1961 to December 2022; it was generated from the Central Climate Information Management System of the BARC. The initial data consisted of measurements from 35 weather stations, covering a multitude of weather parameters that include solar radiation, potential evaporation (PE), evapotranspiration (ETo), maximum temperature, rainfall, humidity, wind speed, cloud cover, and sunshine duration. In the scope of the research titled "Redefining Multi-Target Weather Forecasting with a Novel Deep Learning Model: HTC-LSTM-Attn in Bangladesh," the dataset underwent several pre-processing steps to ensure its quality and suitability for deep learning-based forecasting. Some of these were:Data consolidation: The merging of multiple CSV files (solar radiation, PET, sunshine, wind speed, cloud coverage, humidity, rainfall, and temperature) into one dataset keyed by station code, year, and month.Station filtering: Eleven stations were excluded due to incomplete or unreliable records, retaining 24 stations representing various climate regions.Outlier treatment: Anomalies are detected by the Interquartile range (IQR) method, and such values are replaced with the closest nearest valid value for the same station.Missing value imputation: For gap-filling, k-nearest neighbours (k=5) are applied.Feature engineering: Added seasonal indicators, lag features, and rolling averages to account for temporal dependencies.Feature Selection: By removing highly correlated variables (Pearson's r > 0.9), redundancy was reduced.Normalization: Normalize numerical columns between 0 and 1 scaling using statistics calculated over training sets.Usage:The processed dataset is optimized for deep learning–based weather forecasting models such as HTC-LSTM-Attn, but can also be used for climate trend analysis, seasonal prediction, and meteorological research.