An optimum regression approach for analyzing weather influence on the energy consumption

In the modern society, energy consumption such as gas and electricity is closely related to the weather condition because of the large share of weather-sensitive electrical appliances. Investigating how weather influences the energy consumption is of great significance for energy demand forecasting. This paper proposes an optimum regression approach for analyzing weather influence on the energy consumption. It combines several regression algorithms such as Artificial Neural Network (ANN) and Support Vector Machine (SVM) with optimum weights. The weights of these regression algorithms are determined by formulation of an optimization model considering their corresponding fitting errors. Case studies on two international competitions on energy consumption forecasting are conducted. One is the 2015 Npower Forecasting Challenge, which focuses mainly on the gas consumption; the other one is the 2016 BigDEAL Forecasting Competition, which puts more attention on electrical load forecasting of a certain area. The proposed algorithm was ranked Top 2 in both competitions. It verifies the effectiveness and superiority of our proposed method.