A Likelihood-Based Approach to Developing Effective Proactive Police Methods

Abstract

Proactive policing methods are crucial to ensuring safety and security in line with the UN Sustainable Development Goals. This chapter considers aoristic crime data, where an event occurs within a known time interval, but at an unknown time. We introduce a Bayesian likelihood-based approach to estimate occurrence times of property crimes given a known time interval by modelling victim and offender behaviour as stochastic processes. The model can capture non-homogeneous behaviour by both the victim and the offender and underlying factors leading to patterns in crime occurrence times. We test our model on an open-source aoristic crime data set from the USA, comparing our approach to previous approaches. The model determines the most likely occurrence times through parameter estimation methods, finding potential hot spots, and allowing police to adapt proactive policing strategies. This ties in with SDG 16, which involves strengthening institutions and working towards safe and secure societies.