Short Term Load Forecasting using Regression Trees: Random Forest, Bagging and M5P

Abstract

Decision making in the energy market has to be based on accurate forecasts of the load demand. Therefore, Short Term Load Forecasting (STLF) is important tools in the energy market. In this paper, load forecasting using regression tree methods (Random Forest, Bagging and M5P) are used to effectively forecast the load. The usefulness of the proposed methods has been authenticated through extensive tests using real load data from the Australian electricity market. A comparison of these methods shows that there is an edge in M5P in relation to accuracy.