Machine learning prediction of building energy consumption by combining Bayesian optimization and the SHAP model

R2Accurate prediction of building energy consumption is essential for enhancing energy efficiency and reducing carbon emissions in the built environment. This study develops an interpretable, data-driven framework by integrating Principal Component Analysis (PCA) and SHapley Additive exPlanations (SHAP) to identify key driving factors. Furthermore, a novel Adaptive Bayesian Optimization-integrated Gradient Boosting Regression (ABO-GBR) model is proposed, employing a probabilistic mechanism for automatic hyperparameter tuning to improve generalization. Experimental results demonstrate that the ABO-GBR model significantly outperforms benchmark methods, achieving a coefficient of determination (R2) of 0.994 and the lowest Mean Absolute Error (MAE), providing practical insights for sustainable building energy management.