Construction and analysis of China's carbon emission model based on machine learning

Area of Science:

• Environmental Science
• Machine Learning
• Climate Change Modeling

Background:

• China faces significant challenges due to high carbon emissions.
• Understanding the drivers of carbon emissions is crucial for effective climate policy.

Purpose of the Study:

• To develop a machine learning framework for analyzing and predicting China's carbon emissions.
• To identify key factors influencing carbon emissions and their impact.
• To forecast future carbon emission trends under various policy scenarios.

Main Methods:

• Utilized a "modelling + SHAP analysis + scenario prediction" research paradigm.
• Calculated Spearman correlation coefficients to identify significant explanatory variables.
• Employed SHAP (SHapley Additive exPlanations) analysis to quantify variable contributions.
• Conducted policy scenario simulations for future emission predictions.

Main Results:

• Identified nine variables significantly correlated with China's carbon emissions, including coal proportion and urbanization rate.
• Determined energy intensity (negative effect) and urbanization rate (positive effect) as key drivers.
• Projected carbon emissions to level off between 2022-2028 and peak in 2028.

Conclusions:

• The study provides a robust machine learning approach to analyze carbon emission drivers.
• Energy intensity reduction and sustainable urbanization are critical for mitigating emissions.
• China's carbon emissions are predicted to peak around 2028, reaching approximately 9.72 billion tons by 2030.