Discovering microscale spatial scenes of crime with accounting for urban environmental heterogeneity

Abstract

The advent of urban big data has enabled fine-scale analysis of relationships between crime and urban facilities (typically represented by points of interest, POIs). The spatial configuration of diverse POI types can be conceptualized as crime-related spatial scenes. However, existing methods for detecting spatial scenes of crime often assume a homogeneous relationship between POIs and crime across entire urban environment, neglecting how crime risks associated with specific POIs vary within urban environmental backcloth. This research advances existing research by incorporating the spatial heterogeneity of urban environment when exploring the connections between crime and POIs. We develop an analytical framework that combines: (1) urban functional area identification and (2) spatial co-location pattern mining to detect spatial scenes of crime. This approach facilitates micro-scale spatial scene detection with improved sensitivity to functional urban area. An empirical analysis of robbery crime in New York City reveals spatial variation in crime-POIs relationship across different urban functional areas. These findings offer insights into microscale crime-environment associations and support developing spatial-scene specific crime prevention strategies.