Evaluating different metrics for inter-model comparison of urban-scale building energy simulation time series

This article discusses different statistical methods to compare the predictions (in form of time series) urban-scale building energy models. It focuses on intermodel comparison without empirical measurements as a reference case. is thus suggested build series point-to-point mean average all models be compared (ensemble methods). Different metrics found in literature are then reviewed and tested simulation data from modelling common exercise (DESTEST project). Most those behave similarly. Finally, Python-based tool presented. uses three simple models: NMBE, CVRMSE hourly daily amplitudes.