Multiple criteria assessment of methods for forecasting building thermal energy demand

•Alternative forecasting methods to predict the building energy performance.•Comparison of the three forecasting methods by applying a Multiple Criteria Analysis.•Complex Proportional Evaluation method to identify the most efficient forecast tool.•Extra effort in defining the criteria and identifying by a team of experts.•First application of a Multiple Criteria Analysis in building energy forecast. Nowadays worldwide directives have focused the attention on improving energy efficiency in the building sector. The research of models able to predict the energy consumption from the first design and energy planning phase is conducted to improve building sustainability. Use of traditional forecasting tools for building thermal energy demand tends to encounter difficulties relevant to the amount of data required, implementation of the models, computational costs and inability to generalize the output. Therefore, many studies focused on the research and development of alternative resolution methods, but the choice of the most convenient is not clear and simple. Single comparison of statistical quality indexes does not allow an adequate identification of the most efficient method, as the necessary efforts for implementation of the methods from the initial data collection to the use phase are not considered. In this work, the authors propose to apply, for the first time, the multicriteria assessment to determine the most efficient alternative method, used for forecasting of building thermal energy demand. Three alternative “black-box” methods, previously investigated by the authors, were compared by the multiple criteria Complex Proportional Assessment Method. Such a procedure revealed ranking of the methods in four phases, namely Pre-processing, Implementation, Post-processing and Use, as well as overall assessment and selection of the most efficient method in terms of evaluated criteria. This first application could represent an incentive for future multi-criteria analyses involving a growing number of alternative forecasting methods.