Predicting Electricity Market Price Using Machine Learning and Quantifying Dependency Beyond Renewable Energy

Electricity price dynamics are influenced by various factors, making accurate price prediction a challenging task. In the UK, just like in other countries, electricity prices are affected by renewable energy sources, supply and demand imbalances, system selling prices, and other variables. Building an accurate electricity prediction system requires the integration of several features beyond renewable energy sources. In this study, we use machine learning techniques to predict electricity market prices by considering these diverse factors and later quantify the dependency. We explore several models, including ARIMA, Linear Regression, Decision Trees, Random Forest, Support Vector Regression, XGBoost, and Neural Networks, to identify the most effective model in forecasting electricity price. Our findings indicate that Random Forest, Neural Networks, and our proposed hybrid model deliver the best performance in terms of prediction accuracy. We further evaluated these models to assess how incorporating features beyond renewable energy can improve prediction accuracy. Our results suggest that relying solely on renewable energy is insufficient for making accurate predictions. Additionally, we analyzed and observed the seasonal variations in electricity prices.