Exploring the Nonlinear Response Patterns and Interaction Effects of Urban Resilience Using the XGBoost-SHAP Model: Evidence from the Yangtze River Economic Belt

Abstract

Urban resilience (UR) is shaped by multiple interacting factors, yet existing studies have paid limited attention to the differentiated nonlinear and interactive mechanisms across different resilience dimensions. Based on panel data from 110 prefecture-level and above cities in the Yangtze River Economic Belt (YREB) from 2015 to 2024, this study adopts a four-dimensional framework covering social, economic, ecological, and infrastructure resilience. Spatial analysis and the XGBoost-SHAP model are combined to examine the spatiotemporal evolution, key drivers, nonlinear responses, turning point characteristics, and factor interactions of overall UR and its four dimensions. The results show that overall UR increased steadily, while the four dimensions followed distinct evolutionary trajectories: economic and infrastructure resilience improved most rapidly, ecological resilience increased steadily, and social resilience improved relatively slowly. The importance and effects of key drivers also varied across overall resilience and its four dimensions. Major factors, including technological investment, non-registered population, unemployment insurance coverage, land use, and natural population growth, exhibited distinct nonlinear responses and turning point characteristics, with their effects changing across different variable levels. Furthermore, significant interactions were identified among population and industrial structure, population and transportation, industry and technology, and industry and social security across different resilience dimensions. Overall, UR in the YREB is characterized by differentiated development, nonlinear responses, nonlinear transition patterns, and joint influences of multiple factors.