Strategic Enhancements in Electricity Price Forecasting: The Role of XGBoost and Error Correction Features

Abstract Due to the fact that the electricity markets are now driven by the integration of renewable energy sources, they experience very high price volatility. This paper presents the upgraded forecasting model with a view to managing such volatility by making use of an improved XGBoost algorithm with error-corrective features. Important factors such as holidays and dew point are considered in explaining their substantial influence on the dynamics of electricity pricing. The model ingests hour-based load and pricing information from the ISO New England energy market. Afterwards, Bayesian optimization of hyperparameters further makes the model very accurate and computationally efficient. Results in Mean Absolute Error (rMAE) and Symmetric Mean Absolute Percentage Error (sMAPE) show this model to be very resilient against contemporary energy market complexity.