Carbon emission projection of China's transportation industry based on support vector regression machine

As the global economy continues to develop, climate change has become a critical global issue, with increasing greenhouse gas emissions threatening ecosystems. Carbon emissions from the transportation sector account for approximately one-quarter of global emissions. This makes it crucial to analyze the characteristics of carbon emissions in China's transportation industry for energy conservation and emissions reduction. This paper uses the IPCC inventory guidelines to calculate the total carbon emissions of China's transportation sector from 1997 to 2020 using a top-down approach. Subsequently, we construct an extended STIRPAT model to assess how population, economy, technology, and other factors impact carbon emissions. Ridge regression analysis is employed to eliminate multicollinearity issues, identifying six key influencing factors: passenger turnover, freight turnover, per capita gross national product, motor vehicle ownership, urbanization rate, and carbon emission intensity. These six factors were subsequently used as features to evaluate the effectiveness of the Support Vector Regression (SVR) model in predicting carbon emissions in China's transportation sector. For comparison, two additional machine learning models, Random Forest Regression (RFR) and Back Propagation Neural Network (BPNN), were introduced. The results indicate that the SVR model exhibits high predictive accuracy, producing results closely aligned with actual values. Furthermore, under varying GDP growth rates, the SVR model generally provides higher predictions under high-growth conditions.