Uncovering the multiple socio-economic driving factors of carbon emissions in nine urban agglomerations of China based on machine learning

Understanding urban carbon emissions (CEs) within China's urban agglomerations (UAs) is critical for effective climate action. This study applied machine learning models, particularly the Multi-Layer Perceptron (MLP) model, to analyze socio-economic driving factors of CEs across 144 cities in China's nine major UAs from 1990 to 2020. Results indicated that the Yangtze River Delta (YRD), Shandong Peninsula (SP), and Beijing-Tianjin-Hebei (BTH) UAs exhibited the highest total CEs by 2020, reaching 4.55, 3.53, and 3.90 times their respective 1990 values, primarily driven by regional economic structures and industrial activities. The MLP model demonstrates high predictive accuracy and interpretability in assessing CEs drivers, highlighting the roles of anchor institutions and the correlation of economy, investment and transportation with urban CEs. Based on these driving factors, this study classified UAs into three emission patterns: Socio-Economic Centric, Investment Centric and Balanced Type. This classification supports UA-specific policies, emphasizing comprehensive strategies that integrate technology, finance, and governance to advance green development. This study offers insights into the socioeconomic mechanisms of emissions, guiding tailored reduction targets to aid China in achieving "3060" target.