Forecasting in energy markets is identified as one of the highest leverage contribution areas of Machine Learning towards transitioning to a renewable based electrical infrastructure. The goal of the project is to predict the electrical price by the time of the day using generation, consumption and weather data and also analyze what features influence electrical price the most in Spain.
The project was collaborated by Weipeng Zhang, Soham Shinde and Mingni Luo. We performed classical regression model such as linear regressor, ridge regressor, lasso regressor, random forest regressor, LightGBM, XGBoost, and CatBoost, as well as time seires model like Long Short-term Memory (LSTM), one of the recurrent neural network (RNN) algorithm.
Data source comes from Kaggle.