Using Random Forest Algorithm to Explore the Driving Factors of China Co2 Emission Based on Remote Sensing Data

China has become the world's largest emitter of carbon dioxide, and the government is under tremendous pressure to reduce carbon emissions. This paper analyzes the driving factors of carbon emissions in 281 prefecture-level cities in China from 2003 to 2019 based on the carbon emissions of prefecture-level cities obtained from the matching of nighttime light data and thermal power station emission/location data. Firstly, we compare the accuracy of linear and random forest regression and find that the latter has a better fitting effect. Then we rank the impact of these seven factors according to the regression results of random forests of population, economic development level, energy intensity, urbanization, industrialization, foreign investment, and environmental regulation. The results show that the top three factors affecting carbon emissions are population, economic development, and industrialization. Further, we use the partial dependency graph to reflect factors' complex relations to carbon emissions. In addition, we discuss the interaction between factors and regional differences of factors. The results indicate that the interaction between population and economic development can explain 68.5% of carbon emissions, and the influencing factors' ranking varies in different regions. Finally, this paper puts forward some suggestions for the government to implement carbon reduction policies. This study has significant reference value for studying the influencing factors of CO2 emissions in China and provides a scientific basis and decision support for carbon emission reduction.