Urban Resilience Assessment for Sustainable Development in Mountainous Cities: A Pressure–State–Response Integrated Machine Learning Framework

Enhancing urban resilience is a critical pathway for sustainable development in mountainous cities facing multi-hazard scenarios and rapid urbanization pressures. From an emergency management perspective, this study develops a data-informed urban resilience assessment framework to identify key drivers of urban resistance, response, and recovery capacities. Methodologically, a multidimensional indicator system is built on the Pressure–State–Response (PSR) framework, with quantifiable metrics selected to characterize urban operational features. Data-driven screening identifies core variables, and the Random Forest algorithm quantifies indicator importance to construct a comprehensive evaluation model. A representative mountainous city in China—Lishui—is analyzed to validate the approach, drawing on a 15-year panel dataset (2010–2024) that captures pre- and post-disaster dynamics. Results show that population agglomeration and industrial activities dominate risk pressure; economic development and public services underpin system stability; and medical resource allocation and public emergency response capacity significantly boost resilience. Temporal analysis reveals three distinct stages—initial development, fluctuating adjustment, and rapid improvement—reflecting changes in infrastructure, institutional capacity, and urban development. Urban resilience emerges from multidimensional synergies among pressure, state, and response factors, highlighting the importance of integrated infrastructure planning and cascading risk mitigation. The framework provides a reference for identifying mountainous cities’ key resilience factors, while the findings may help decision-makers identify risks, formulate resilience strategies, and optimize sustainable development pathways.