Combining a Genetic Algorithm and Support Vector Machine to Study the Factors Influencing CO2 Emissions in Beijing with Scenario Analysis

• Main Authors: Jinying Li, Binghua Zhang, Jianfeng Shi
• Format: Article
• Language: English
• Published: MDPI AG 2017-10-01
• Series: Energies
• Subjects: CO2 emissions prediction; genetic algorithm; support vector machine; scenario analysis; influence factors
• Online Access: https://www.mdpi.com/1996-1073/10/10/1520

In recent years, Beijing has been facing serious environmental problems. As an important cause of environmental problems, a further study of the factors influencing CO2 emissions in Beijing has important significance for the social and economic development of Beijing. In this paper, Cointegration and Granger causality test were proposed to select influencing factors of CO2 emissions prediction in Beijing, the influencing factors with different leading lengths were checked as well, and the genetic algorithm (GA) was used to optimize the initial weight and threshold values of a support vector machine (SVM) and the new SVM optimized by GA (GA-SVM) was established to predict the CO2 emissions of Beijing from 2016–2020 with scenario analysis. Through the comparison of 36 kinds of development scenarios, we found that economic growth, resident population growth and energy intensity enhancement were the major growth factors of carbon emissions, of which the contributions exceed 0.5% in all kinds of development scenarios. Finally, this paper put forward some reasonable policy recommendations for the control of CO2 emissions.