Multi-perspective assessment of ecological civilization progress across cities in the Yangtze River Economic Belt

Ecological civilization, China’s national strategy, promotes sustainable development through green growth, harmony between humans and nature, and improved well-being. The Yangtze River Economic Belt (YREB), with diverse urban trajectories, plays a key role in this goal. However, conventional static indices fail to capture dynamic changes, goal deviations, and multidimensional linkages. While the SDGs set ambitious targets, they lack explicit benchmarks, and current evaluation methods, relying on subjective or static data structures, limit goal-oriented assessments. Based on multi-criteria decision-making (MCDM) theory, this study integrates multi-source data to create a multidimensional indicator system for ecological environment (EE), ecological economy (EEC), ecological living (EL), and ecological framework (EF). An ecological civilization profile is developed using the mean-trend-fluctuation method. Dynamic weighting is achieved through K-means, hierarchical clustering, SOM, and machine learning models (RF, SVM, CatBoost). The goal-oriented S-P-G (Status-Progress-Gap) framework evaluates 108 YREB cities (2013-2022), identifies development stages, reveals spatial patterns via Getis-Ord Gi*, and quantifies indicator contributions using RF-SHAP. Results show that the SI increased from 65 in 2013 to 70 in 2022, the PI rose rapidly (average annual growth >8% in 2014–2017) before stabilizing with a –4.84% drop in 2020, and the GI declined from 38.9% to 27.0%. The “east-high-west-low” pattern was observed, with provincial capitals catching up (PI >55 by 2022). RF-SHAP analysis shows EEC drives SI, EL enhances PI and narrows GI, while EE is a bottleneck in advanced cities. Based on these findings, we offer differentiated policy recommendations tailored to cities at various development stages. • A goal-oriented S–P–G framework for assessing ecological civilization progress. • Ecological civilization profile-based machine-learning weight allocation method. • SI, PI, and GI indices show an east-high–west-low pattern in YREB. • Ecological civilization progress in YREB shifts from Lagging to Mature Optimization.