Energy Demand Forecasting for an Office Building Based on Random Forests

A random forest (RF) is a classifier consisting of an ensemble of decision trees and has already been demonstrated to have effective predictive classification capabilities in many fields. Currently, however, few scholars have considered using RFs to forecast the energy consumption of buildings without historical load data. In this paper, RFs were applied to forecast the energy demand of an office building in Shanghai. A predictive model was obtained by training on 30 office buildings that were also located in Shanghai. Then, the trained predictive model was used to forecast the energy demand of an unknown building. The results demonstrated the feasibility of using RFs to forecast the energy demand of unknown office buildings.