A Novel Industry-Classification Final Energy Consumption Structure Clustering Method Based on Improved K-Means Algorithm

Abstract

The industry-classification final energy consumption structure is inextricably linked to economic development. Due to the goal of carbon neutrality, China's industry-classification final energy consumption structure is undergoing profound changes. It is challenging to analyze the industry-classification final energy consumption structure in multiple dimensions using analytical tools such as line charts. In order to illustrate the variation of the final energy consumption structure in different industry sectors at different times, a novel industry-classification final energy consumption structure clustering method based on an improved $K$-means algorithm is proposed in this paper. Three methods, including the elbow method, the silhouette coefficient method, and the Calinski-Harabasz (CH) index method, are used to optimize $k$ values in the $K$-means algorithm. The classification results are evaluated through the empirical analyses from the China industry dataset. The simulation results demonstrate that the proposed method can accurately classify the industry-classification final energy consumption structure of industrial sub-sectors into four categories. Moreover, the trend of the evolution of the industry consumption structure shows that most industries have shifted from coal-based consumption to electricity-based consumption.