Comparative analysis of machine learning models for forecasting hydroelectric generation using socioeconomic indicators

Hydropower plays a significant role in Malaysia’s renewable energy mix, particularly in regions with abundant water resources such as Sarawak. Accurate forecasting of hydroelectric generation is increasingly important to support effective energy planning and the country’s sustainability objectives. This study explores the performance of four machine learning models: Artificial Neural Networks (ANN), Support Vector Machines (SVM), Random Forest, and XGBoost in forecasting Malaysia’s hydroelectric power output using socioeconomic indicators, including Gross Domestic Product (GDP), energy consumption, and population. ANN demonstrated the most promising results among these models, achieving a testing Mean Squared Error (MSE) of 1.1541×10⁴ and a correlation coefficient (R) of 0.9962. These results suggest that ANN can capture the underlying patterns within the data and may offer a valuable tool for improving the reliability of hydropower generation forecasts, thereby contributing to Malaysia’s ongoing efforts toward renewable energy development.