Nonlinear relationships and interaction effects of an urban environment on crime incidence: Application of urban big data and an interpretable machine learning method

Abstract

While environmental criminology suggests that crime and the urban environment are closely related, some studies suggest a nonlinear relationship. This study analyzed the relationship between crime incidence and the urban environment using urban big data such as points-of-interest (POI), smart civil complaint data, and street image data from Naver Street View in Seoul, Korea. For analysis, the Light Gradient Boosting Machine (LightGBM) model and SHapley Additive exPlanation (SHAP) method have been used. The analysis results confirmed a nonlinear relationship comprising inflection points between crime incidence and the urban environment. Also, this study identified the interaction effects of urban environmental variables on crime incidence. Finally, the hierarchical clustering method was used to identify the contributions of various aspects of the urban environments to crime incidence. Then, this study provides policy implications to prevent potential criminal activities and promote public safety for sustainable cities and societies.