A novel integrated optimization model for carbon emission prediction: A case study on the group of 20

Carbon emission is a central factor in the study of the greenhouse effect and a crucial consideration in environmental policy making. Therefore, it is essential to establish carbon emission prediction models to provide scientific guidance for leaders in implementing effective carbon reduction policies. However, existing research lacks comprehensive roadmaps that integrate both time series prediction and analysis of influencing factors. This study combines the environmental Kuznets curve (EKC) theory to classify and qualitatively analyzes research subjects based on national development patterns and levels. Considering the autocorrelated characteristics of carbon emissions and their correlation with other influencing factors, we propose an integrated carbon emission prediction model named SSA-FAGM-SVR. This model optimizes the fractional accumulation grey model (FAGM) and support vector regression (SVR) using the sparrow search algorithm (SSA), considering both time series and influencing factors. The model is subsequently applied to predict the carbon emissions of the G20 for the next 10 years. The results demonstrate that this model significantly improves prediction accuracy compared to other mainstream prediction algorithms, exhibiting strong adaptability and high accuracy.