Benchmarking Energy Quantification Methods to Predict Heating Energy Performance of Residential Buildings in Germany

Abstract To achieve ambitious climate goals, it is necessary to increase the rate of purposeful retrofit measures in building sector. As a result, Energy Performance Certificates have been designed as important evaluation and rating criterion EU Germany. Yet, today’s most frequently used legally required methods quantify energy performance show low prediction accuracy, recent research reveals. enhance community introduced data-driven which obtained promising results. However, there are no insights how far Quantification Methods particularly suited for prediction. In this article Artificial Neural Network, D-vine copula quantile regression, Extreme Gradient Boosting, Random Forest, Support Vector Regression compared with validated by real-world German residential buildings issued qualified auditors using engineering method law. The results, tested robustness systematic bias, that all exceed almost 50% terms accuracy. contrast existing literature favoring Networks Regression, similar accuracy marginal advantages Boosting Given higher methods, seems appropriate revise current legislation prescribing methods. addition, could support different organizations, e.g., asset management, decision-making order reduce financial risk cut expenses.