As China’s urbanization transitions from incremental expansion to stock-based renewal, industrial land redevelopment has become a key pathway for promoting high-quality urban development. However, existing studies mostly assess redevelopment potential from a single dimension and lack a systematic framework integrating ecological function (E), spatial structure (S), economic conditions (C), and building foundations (B). Taking the built-up area of Xi’an as a case study, this study adopts a functional–structural optimization perspective and constructs a four-dimensional ESCB assessment framework based on 13 indicators covering ecological function, spatial structure, economic conditions, and building foundations. GIS-based spatial quantification, MiniBatchKMeans clustering, and the XGBoost algorithm were employed to identify the redevelopment potential of industrial land, while SHAP analysis was used to interpret indicator contributions and determine the core influencing factors. The results show that industrial land in the study area can be classified into four types: vitality–density dominant, transport–scale coordinated, scale–facility lagging, and topography–vegetation sensitive, with significant differences in spatial distribution and indicator characteristics. The interpretable machine learning model further identifies road network density, block-level economic vitality, and land-use suitability as the three principal drivers of redevelopment potential, among which road network density plays the most critical role. By integrating clustering analysis with interpretable machine learning, the ESCB framework effectively reveals the synergies and trade-offs among multidimensional indicators and provides differentiated and precise support for industrial land redevelopment strategies.
Lin et al. (Fri,) studied this question.