资源型城市碳排放影响因素研究:基于可解释的机器学习方法OACHSSCD

Under the constraint of dual-carbon targets,reducing carbon emissions is a key path to sustainable development in Chinese resource-based cities.Based on panel data of 114 cities from 2008 to 2023,this study constructs an interpretable machine learning model integrating Boruita feature selection method,integrated learning algorithm and SHAP to analyze the factors influencing per capita carbon emissions.The result indicate that:1)Machine learning model constructed based on 42 influencing factors shows good prediction performance,among which the LightGBM has the best performance.2)Importances of electricity consumption per capita,built-up area ratio,urbanization and GDP per capita are higher.3)Most of the factors show a non-linear relationship with their marginal contribution to per capita carbon emissions,and interactions exist among these factors.4)Importance of factors changes with city geographical location and development stage.This study provides a data-driven decision-making basis for resource-based cities to accurately formulate carbon emissions control measures.