Early experiments on predicting solar electricity generation over the next few hours, using deep learning, satellite imagery, and as many other data sources as we can think of :)

These experiments are focused on predicting solar PV yield.

Please see SatFlow for complementary experiments on predicting the next few hours of satellite imagery (i.e. trying to predict how clouds are going to move!)

And please see OCF's Nowcasting page for more context.

From within the cloned predict_pv_yield directory:

conda env create -f environment.yml
conda activate predict_pv_yield
pip install -e .