The industrial development in northwest China heavily relies on high-energy-consuming and high-emission sectors like coal and petrochemicals,leading to a persistent rise in manufacturing carbon emissions.Based on panel data from 2000 to 2020,a carbon emission prediction model is developed,integrating STIRPAT,Gradient Boosted Decision Tree(GBDT),and Extreme Gradient Boosting(XGBoost).Initially,carbon emissions are estimated using emission factors,analyzing the status quo and spatiotemporal evolution.Subsequently,an extended STIRPAT model selects 15 emission-related factors,refined further by Person and GBDT.Among five machine learning models tested,XGBoost exhibits superior predictive performance,with MSE,MAPE,and R2 of 0.54%,16.78%,and 93.87%respectively.Achieving R2 over 90%across northwest provinces,the model accurately forecasts manufacturing carbon emissions,offering insights for advancing towards the"dual carbon"goals in the region.
关键词
制造业碳排放/XGBoost模型/西北地区/影响因素/预测模型
Key words
manufacturing carbon emissions/XGBoost model/northwest region/influencing factors/prediction model