A novel model averaging forecasting method for electricity consumption using electricity business expansion data

This paper introduces a novel model averaging method for electricity consumption forecasting that incorporates electricity business expansion application data. The proposed framework uses a two-step regression approach: first, electricity consumption is decomposed into long-term deterministic trends, temperature-related components, and short-term economic activity components. Second, a model averaging forecast is constructed for the short-term component using 15 forecasting variables derived from expansion application data, along with shortterm Gross Domestic Product (GDP), electricity prices, and other relevant factors. Using data from Fujian Province, we compare the forecasting performance of models with and without expansion application data, employing both model selection and the Akaike Information Criterion (AIC) based model averaging techniques. Empirical results show that incorporating expansion application data improves forecasting accuracy, the Mean Absolute Percentage Error (MAPE), by up to 14 % in out-of-sample predictions. These findings underscore the value of expansion application data as a resource for enhancing electricity consumption forecasting and supporting more effective energy management strategies.