Multicriteria Decision Making for Selecting Forecasting Electricity Demand Models

Sustainable electricity consumption is considered a pivotal element in the effective governance and growth of any institution. Accurate electricity demand forecasting is essential for strategic planning and decision making. However, due to the numerous existing forecasting approaches, many forecasters find it challenging to select the best model. Currently, there is no robust approach for selecting the best forecasting model when considering conflicting error measures. This paper proposes a novel methodology using a multicriteria decision making (MCDM) approach to determine the most appropriate forecasting model for electricity demand, considering various interdependent error measures. The Analytical Network Process (ANP) was applied to determine the weights of evaluation criteria, while the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) was employed to select the best forecasting model. The proposed methodology was tested and validated with a real case study in Tunisia using the opinions of experts and stakeholders. The results show that multiple regression and exponential smoothing are the best alternatives and outperformed the other models. Additionally, a sensitivity analysis is presented to test the robustness of the final ranking. This serves to assist decision makers to select the best forecasting model.