Weather to Grid

Subseasonal Energy Forecasting using ML

This project aims to advance subseasonal energy forecasting using a causal ML approach (based on a type of autoencoder termed CMM-VAE) designed to identify large-scale weather patterns that are both predictable and informative of a local variable. The first goal is identify weather regimes relevant to European electricity demand and wind and solar energy production, with forecasts communicated to users on a dedicated website. The second goal is to develop an open-source Python module that allows users to seamlessly train the model on their own data, thereby bringing a research tool to the broader community.