A Data-Driven Evaluation Method for Building Energy Efficiency Based on Multi-Kernel Support Vector Regression

This paper explores a data-driven evaluation method for building energy efficiency based on Multi-Kernel Support Vector Regression (MK-SVR). In response to the limitations of traditional energy-saving evaluation methods, this study aims to develop a more accurate and efficient model to support building energy-saving design and management decisions. The article first reviews the current application status of data-driven methods and support vector regression (SVR) in the field of building energy efficiency, and elaborates on the basic principles and advantages of MK-SVR. Subsequently, the process of model construction is described in detail, including data preprocessing, feature selection, kernel function selection, and parameter tuning. The accuracy and generalization ability of the MK-SVR model in energy-saving effect evaluation are verified through case analysis using actual building energy consumption datasets. The experimental results show that compared with existing methods, the MK-SVR model exhibits higher prediction accuracy and stronger adaptability. This study not only provides a new method for evaluating the energy-saving effect of buildings but also contributes to promoting green buildings and sustainable development.