XgBoost based Short-term Electrical Load Forecasting Considering Trends & Periodicity in Historical Data

The effective planning and management of residential electricity demand requires precise forecasting of the short-term electrical load. A novel approach is proposed for short-term electrical load forecast employing the XGBoost algorithm in this research work which is based on feature selection by considering the trends and patterns in the historical dataset. This XGBoost algorithm combines several weak learners to create a strong predictive model. The trained model is capable of capturing the nonlinear relationships and complex patterns present in real-world residential load data. The simulation results reveal that the XGBoost algorithm outperforms traditional forecasting techniques in terms of accuracy when trained on feature selection based on data analytics. The accuracy of the forecast is assessed using standard metrics such as root mean square error (RMSE), mean absolute percentage error (MAPE), mean square error (MSE), and mean absolute error (MAE). The proposed approach achieves significant improvements for a 24-hour-ahead electrical load forecast for domestic users.