Electricity price forecasting utilizing machine learning in MIBEL

Abstract

Short term electricity price forecasts have become increasingly important in the last few decades due to the rise of more competitive electricity markets throughout the globe. Accurate forecasts are now essential for market players to maximize their profits and hedge against risk, hence various forecasting methodologies have been applied to electricity price forecasting in the last few decades. This dissertation explores the main methodologies and how accurately can three popular machine learning models, SVR LSTM and XGBoost, predict prices in the Iberian market of electricity. Additionally, a study on input variables and their relationship with the final price is made.