Interpretable machine learning for analyzing the spatial-temporal evolution and influencing factors of green economic development

• Published in: Expert systems with applications Vol. 283; p. 127863
• Main Authors: Ding, Rui, Ran, Xiaofeng, Jiang, Shuyue, Zhang, Bowen
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.07.2025
• Subjects: Green economy development; Influencing Factors; Interpretable Machine Learning; SHAP; Spatio-temporal Transition; Green economy development; SHAP; Spatio-temporal Transition; Influencing Factors; Interpretable Machine Learning
• ISSN: 0957-4174
• DOI: 10.1016/j.eswa.2025.127863

Understanding the dynamics and influencing factors of green economy development (GED) is essential for fostering sustainable growth and environmental sustainability. This paper investigates the spatial–temporal evolution, transition and influencing factors of GED in 287 cities of China. A multidimensional evaluation framework combining the entropy weight method and Exploratory Spatio-Temporal Data Analysis (ESTDA) quantifies GED trajectories while revealing two critical dynamics: 1) a core-to-periphery diffusion pattern with persistent regional disparities, and 2) unstable development pathways of cities despite stable high-performance clusters in coastal major urban agglomerations. To decode complex influencing factors, we introduce SHAP (SHapley Additive exPlanations)-interpreted Random Forest modeling, identifying consumption, financial development, scientific and technological innovation and commercial development as dominant drivers. The results directly support evidence-based governance by operationalizing SHAP-driven insights: urban decision-makers should embed spatial SHAP heterogeneity analysis into cross-regional and adaptive policy systems that may trigger GED upgrades.