Overlapped Bayesian spatio-temporal models to detect crime spots and their possible risk factors based on the Opole Province, Poland, in the years 2015–2019

Abstract

AbstractGeostatistical methods currently used in modern epidemiology were adopted in crime science using the example of the Opole province, Poland, in the years 2015–2019. In our research, we applied the Bayesian spatio-temporal random effects models to detect ‘cold-spots’ and ‘hot-spots’ of the recorded crime numbers (all categories), and to ascertain possible risk factors based on the available statistical population (demographic), socio-economic and infrastructure area characteristics. Overlapping two popular geostatistical models in the analysis, ‘cold-spot’ and ‘hot-spot’ administrative units were detected which displayed extreme differences in crime and growth rates over time. Additionally, using Bayesian modeling four categories of possible risk factors were identified in Opole. The established risk factors were the presence of doctors/medical personnel, road infrastructure, numbers of vehicles, and local migration. The analysis is directed toward both academic and police personnel as a proposal for an additional geostatistical control instrument supporting the management and deployment of local police based on easily available police crime records and public statistics.