Analysis influence factors and forecast energy-related CO2 emissions: evidence from Hebei

With economic development and the acceleration of urbanization, China’s energy demand has gradually increased and brought a lot of energy-related CO2 emissions. Energy-related CO2 emissions are affected by a variety of factors. Quantifying the correlation between energy-related CO2 and driving factors and constructing the driving factor system are conducive to predict the future energy-related CO2 emissions and analyze the impact of driving factors. In this paper, the improved grey relational analysis (IGRA) was proposed to screen the influencing factors of energy-related CO2 emissions considering the sample difference, and the factor analysis (FA) was used to reduce dimensionality of the influencing factors. Then, a carbon dioxide emission forecasting model based on the bacterial foraging optimization algorithm (BFO) and the least square support vector machine (LSSVM) was proposed. Empirical analysis results of Hebei show that the LSSVM optimized BFO significantly improves the accuracy of energy-related CO2 emissions forecasting, and IGRA-FA-BFOLSSVM model is significantly better than BP, PSOBP, SVM, and LSSVM models. The mean absolute percentage error (MAPE) of the proposed model is 0.374%. The forecasting results of the supplementary case show that the model has better generalization ability. In addition, education and technological progress have proven to be important drivers of energy-related CO2 emissions. Simultaneously, the research results can also offer more breakthrough points for policy makers to control carbon emissions.