An Ensemble Learning Approach for Accurate Energy Load Prediction in Residential Buildings

Reducing energy loads while maintaining the degree of hotness and coldness plays an essential role in designing energy-efficient buildings. Some previous methods have been proposed for predicting building energy loads using traditional machine learning methods. However, these traditional methods suffer from overfitting problems, which leads to inaccurate prediction results. To achieve high accuracy results, an ensemble learning approach is proposed in this paper. The proposed approach uses an extreme gradient boosting (XGBoost) algorithm to avoid overfitting problems and builds an efficient prediction model. An extensive experiment is conducted on a selected dataset of residential building designs to evaluate the proposed approach. The dataset consists of 768 samples of eight input attributes (overall height, relative compactness, wall area, surface area, roof area, glazing area distribution, glazing area, and orientation) and two output responses (cooling load (CL) and heating load (HL)). The experimental results prove that the proposed approach achieves the highest prediction performance, which will help building managers and engineers make better decisions regarding building energy loads.