Measuring Multi-Dimensional Urban Boundaries Influencing Theft: A Case Study of Guangzhou, China

Abstract

Identifying the environmental and social factors that influence crime is crucial for effective crime prevention and for building safer cities. Boundary areas are often neglected, despite being hotspots for criminal activity. However, previous studies have primarily focused on physical boundaries, with insufficient attention paid to social boundaries shaped by factors such as population composition and socioeconomic disparities. Additionally, existing research methods remain limited in scope, with models that struggle to capture the nonlinear and complex nature of these relationships. This study proposes a comprehensive approach by measuring both multidimensional physical and social boundaries within urban environments. Using machine learning models, we present three main findings: urban boundaries are strong predictors of crime; boundary variables show stronger correlations with crime than intra-area variables; and both social and physical boundaries warrant equal attention. These findings suggest that governments should enhance problem-oriented policing at boundary hotspots in urban boundary areas while also promoting social integration to address the root causes of crime.