Codes for A Cutting-Edge Conceptual Reservoir Operation-Based Deep Learning Framework

Training and analysing code for A Cutting-Edge Conceptual Reservoir Operation-Based Deep Learning Framework (CRO-LSTM). Descriptions for model training and analyzing steps can be found at “Steps to reproduce” in the link of source code.
Data description:
(1) runoff, rain and reservoir data in the testing stage collected from the management authority of the Minjiang basin. hydro_data.csv : Include basin runoff data and reservoir data. rain_data.csv: Include rainfall observation data.
(2) The processed data for model teseting. Resampling and averaging methods were used to handle missing values and mitigate fluctuations. data.csv: processed data. data.pth: processed data packed for model  development. shap_values.pth: computed shapley values from the test datasets.