Spatial Patterns and Multidimensional Drivers of Land Use Inefficiency in Quanzhou, China: Insights from Explainable Machine Learning

As urban development shifts from outward expansion to stock-based regeneration, inefficient urban land has become a critical constraint on land resource allocation, urban renewal, and sustainable spatial governance. To support differentiated urban renewal and sustainable land governance, this study examined the spatial patterns, factor contributions, and their interaction mechanisms of inefficient land use in 2022 within the urban development boundary of Quanzhou, Fujian Province southeastern of China. Based on the official 2022 citywide survey of inefficient urban land, multisource geographic and socioeconomic data were integrated with spatial statistical analysis, extreme gradient boosting (XGBoost), and Shapley additive explanations (SHAP). The results show that 5396 inefficient land parcels were identified in 2022, covering 7762.44 ha. Industrial land accounted for the largest area share (59.23%), followed by residential land (38.67%) and commercial land (2.10%). Inefficient land displayed a clear northwest-southeast distributional orientation and was mainly distributed across core urban areas, traditional industrial agglomeration zones, and county-level urban nodes. For total inefficient land, the Global Moran’s I reached 0.97, and High-High clusters were mainly located in Licheng District, southern Jinjiang City, and the county-level urban centers of Dehua and Yongchun, indicating concentration in old urban districts, mixed industrial-residential spaces, and county-level built-up areas. The SHAP results reveal type-specific but multidimensional mechanisms. Industrial inefficiency reflected the compatibility between production-space organization, transport and service support, ecological-location conditions, and market value; residential inefficiency was shaped by the coordination among parcel organization, community services, population demand, and mobility support; and commercial inefficiency depended on the matching among development intensity, road and public transport conditions, population distribution, and ecological-location conditions. These findings suggest type-specific renewal strategies, including parcel consolidation and production-supporting function improvement for industrial land, neighborhood-scale service coordination for residential land, and functional matching among development intensity, road and public transport capacity, population distribution, and ecological-location conditions for commercial land.