NONLINEAR EFFECTS OF THE DIGITAL ECONOMY ON URBAN LOW-CARBON ECOLOGICAL HIGH-QUALITY DEVELOPMENT IN CHINA: A MACHINE LEARNING-BASED ANALYSIS AND PREDICTION

The digital economy development (DED) plays a crucial role in promoting urban low-carbon ecological high-quality development (ULCEHD).This study uses panel data from Chinese prefecture-level regions between 2014 and 2023.It employs a Bayesian Graph Convolutional Neural Network (BGCNN) model to explore the nonlinear and spatially heterogeneous effects of DED on ULCEHD while controlling confounding variables.A spatiotemporal dynamic entropy method is proposed to quantitatively measure DED and ULCEHD in this study, revealing significant upward trends in both metrics from 2014 to 2023.The findings demonstrate an "S-shaped" relationship between DED and ULCEHD, where initial increases in DED lead to ULCEHD improvements of 0.02 units per 0.10 DED increase; this effect intensifies to a 0.20 rise at medium levels of DED, followed by a plateau.Spatial analysis reveals significant regional disparities, with the strongest DED impacts on ULCEHD observed in central, southwestern, and northeastern China.Conversely, regions such as the North China Plain and parts of the eastern coast exhibit weaker DED effects on ULCEHD.By 2030, DED is expected to spread more evenly across China.This will lead to broader ULCEHD improvements, highlighting the need for region-specific policies to fully leverage digital economic growth for sustainable urban development.