Residential Building Energy Usage Prediction Using Bayesian-Based Optimized XGBoost Algorithm

With the growing energy demand in residential buildings, selecting energy-efficient building designs is crucial for sustainable development. Therefore, engineers and designers try to find a reliable tool to help them analyze and predict the energy consumption of buildings in the early design stages before construction. The state-of-the-art eXtreme Gradient Boosting (XGB) algorithm was successfully used to estimate the Heating Load (HL) and Cooling Load (CL) energy usage based on building design characteristics. To enhance the performance of XGB and improve prediction reliability, we propose a novel Bayesian optimization approach using the Tree-Structured Parzen Estimator (TPE), leading to the development of the TPE-XGB model for HL and CL estimation. Also, a new fair evaluation method was proposed to have a fair evaluation. The evaluation results based on the proposed fair evaluation method showed that the proposed model has outperformed other state-of-the-art models and improved the prediction accuracy of HL and CL by 3.4% (from 0.175 to 0.167) and 10.4% (from 0.307 to 0.275), respectively. As a result, the new proposed model provides the opportunity to be used as a strong and reliable tool in the early stages of building construction and to evaluate different designs/plans regarding energy consumption efficiency to select the best design with high reliability.