Leveraging User‐Generated Spatial Content for Urban Crime Analysis: A Data‐Driven Approach

ABSTRACT Street crimes in urban areas are influenced by the physical characteristics and built environment of a city, including road networks, visibility, lighting, and land use distribution. Traditionally, authorized data sources such as city maps have been used to investigate these urban form parameters. However, with advancements in technology, user‐generated spatial content (UGSC) or volunteered geographic information has become available, offering advantages such as real‐time updates and unrestricted access. This research integrates both authorized and UGSC data sources to explore the relationship between the occurrence of street crimes and multiple criteria related to land use and road network characteristics. We employ three data‐driven methods: Support Vector Regression, Partial Least Squares regression, and Geographically and Temporally Weighted Regression (GTWR). Our findings reveal that GTWR is robust for modeling the correlation, achieving an accuracy of 55%. Notably, network indexes (e.g., Integration and Total Depth), commercial and miscellaneous land uses, and the population aged 18 to 44 significantly impact street crime occurrence. These insights can guide urban planners and decision makers in developing effective crime reduction strategies during the urban planning phase. By considering the effect of planned road network and land use mix development strategies in relation to crime occurrence, quantified through our proposed data‐driven approach, cities can enhance safety and security for their residents.