This repository gathers and expands results obtained with a Deep Learning system for solar irradiance forecasting. It is organised as follows:
-
Datasets: The data that was used in the experiments is subject to distribution restrictions. Quoting the website where the raw data was obtained from: These data and any subset may not be publically redistributed by any means. Thus, a Jupyter Notebook is provided to reproduce how the raw data was transformed to train the models. Detailed information can be found here.
-
Result tables: Skill scored by each model for every horizon alongside model hyperparameters. Also, robustness tests for all models that work with irradiance maps. Detailed information can be found here.
-
Graphs: Sensor distribution map, boxplots of the skill per horizon, skill maps, robustness tests, sample predictions, model representation as graphs, animated irradiance maps... Detailed information can be found here.