Application of Machine Learning in Forecasting Short-Term Electricity Load

The short-term electricity forecasting using the Extreme Gradient Boosting (XGBoost) approach is used in this research to predict electricity demand in Panama. We provide two approaches for comparison: Adaptive Boosting (AB) and Support Vector Machines (SVM). The objective is to predict the electricity demand due to the escalation of using electricity. Feature selection methods are also presented in this study, including the step-forward method (SFM), the step-backward method (SBM), the filter method (FM), and the principal components analysis method (PCA). The results indicate that the XGBoost model outperforms the other models.