Comparative Analysis of ML Models for Electricity Price Forecasting

Precise forecasting of electricity prices is crucial for optimizing energy consumption, ensuring market stability, and facilitating well-informed decision-making within the electricity sector. However, the intricate dynamics and inherent volatility of electricity markets present significant challenges to conventional statistical methods. Consequently, machine learning (ML) algorithms have emerged as powerful tools for addressing this complex forecasting task. This study undertakes a comprehensive comparative analysis of prominent ML algorithms for electricity price forecasting, assessing their performance across various forecasting horizons and market conditions. The research investigates the comparative performance of 18 ML algorithms, focusing primarily on regressors such as the Random Forest Regressor, Extra Trees Regressor, LGBM Regressor, and Decision Tree Regressor, among others. To thoroughly evaluate the forecasting accuracy and robustness of each model, a variety of evaluation metrics are employed, including mean absolute error (MAE) and root mean squared error (RMSE). An important aspect of this research involves examining exogenous factors, such as weather patterns and electricity generation sources, and their influence on the forecasting performance of ML algorithms.