|
Abstract:
|
The predictive policing policing movement is aimed at using quantitative models to support police decision-making. One popular predictive policing method, the self-exciting point process (sepp) model, uses past crime events to predict where and when future crimes will occur. Another approach, crime linkage analysis, models the similarities and differences between crimes to help police discover and group crime events that share a common offender. It turns out that there is a close connection between the assumptions behind the sepp and crime linkage models. Namely, both models are based on the observation that crimes tend to cluster in space and time. In this talk, we show how the sepp models can be used for crime linkage. We then develop a new sepp model that incorporates additional linkage related crime variables. We find that the new models perform better than the existing crime linkage models because they can adapt to spatiotemporal changes in offender preference.
|
