Contribution to Predicting Electricity Load using Random Forest

The growing demand for sustainable energy solutions and the integration of renewable resources have increased the need for accurate electricity demand forecasting, which is an essential element in effective power system management. This study evaluates the application of the Random Forest (RF) algorithm, an ensemble learning technique, for predicting Short-Term Electricity Load (STEL). Using historical data including load, weather conditions, and calendar variables, we optimized the model through feature engineering and hyperparameter tuning, in an effort to capture the dynamic interactions between consumer behavior, climate effects, and seasonal patterns, all of which significantly influence electricity demand. The results demonstrate the RF model’s superiority over traditional methods, achieving high accuracy measured by Mean Absolute Percentage Error (MAPE).