Analyzing and predicting CO2 emissions in China based on the LMDI and GA-SVM model

Abstract

With the effect of CO2 emissions being the primary cause of the greenhouse effect, a selection and analysis study of driving factors of CO2 emissions is vital to controlling growth from the source. This paper decomposes CO2 emissions based on the logarithmic mean division index (LMDI) from three industries and residential consumption in China during the period 2000-14. A genetic algorithm-support vector machine (GA-SVM) was established. The eight driving factors as input variables have been innovated to apply the forecasting model. In the case study, the data set of driving factors from 2000 to 2009 is selected as training samples, and the other data set of driving factors from 2010 to 2014 is regarded as test samples. The results show that the factor decomposed based on the LMDI method of CO2 emissions is very rational and can greatly improve forecast accuracy. The effectiveness of the GA-SVM model has been proven by the final simulation, which indicates that the proposed model outperforms a back propagation neural network (BPNN) model and a single SVM model in forecasting CO2 emissions.