Exploring road safety through urban fabric characteristics and theory-driven prediction modeling with SEM-XGBoost

Abstract

This study addresses the critical issue of road safety in urban environments, with a specific focus on the Greater London Area. Utilizing a novel, theory-driven approach, the study investigates the multifaceted impact of urban fabric factors on road safety, operationalized through a severity-weighted index of road accident frequency per capita. Through factorial analysis, six key factors (Urban Integration, Socioeconomic Challenges, Urban Amenities, Commuter Patterns, Housing and Mobility Barriers, and Major Urban Infrastructure) are identified. These factors are examined in relation to road safety using a structural equation model to uncover theoretical relationships, which inform predictive modeling with an XGBoost machine learning framework, enhanced by SHAP value analysis. Our findings reveal significant insights into the interplay between urban physical and social environments and road safety, revealing that integrated urban development strategies—encompassing improved urban integration, enhanced sustainable density, robust infrastructure development, alleviation of socioeconomic disparities, fostering of local employment, and integration of residents with transportation—are imperative for increasing road safety.