Abstract

Predictive Crime Hotspot Mapping
Track: Law Enforcement and Criminal Justice
Authors: Paul Zandbergen

Mapping crime hotspots is very useful for examining crime patterns, assisting in the allocation of resources, and analyzing the impacts of crime reduction policies. Increasingly crime hotspots are also used for predicting where and when future crime events will occur. This presentation will discuss the results of research on the reliability of various hotspot techniques for predictive crime mapping. Crime datasets from six different jurisdictions were used including assaults, auto theft, burglary, drugs, homicide and robbery. Hotspot techniques analyzed using GIS include thematic mapping, grid-based mapping, Local Indicators of Spatial Association (LISA), Getis-Ord Gi*, and kernel density. Results indicate that the reliability of hotspot techniques to predict future crime events varies by crime type and with the characteristics of the study area. Results are also very sensitive to hotspot parameter selection. Strengths and weaknesses of each hotspot technique are presented as well as general recommendations for predictive crime mapping.