A Computational Model for Predicting the Location of Crime Attractors on a Road Network

Crime Pattern Theory argues that offenders often commit their crimes at major criminal attractors or along the routes that lead to these attractors from other activity nodes in their awareness space. Without knowledge about each offender's awareness space, however, it is not possible to know the nodes they travel to. In this paper, it is assumed that each offender commits their crime along the way to a particular end-destination that they frequent. It thus follows that, for each crime location of an offender, there is an activity path nearby which starts at the offender's home location and ends at a nearby activity node. Since the activity node is not known, but the crime and home locations of offenders are, the method presented in this paper determines the locations of the attractors by statistically extending the path along the road-network from the offender's home location to their crime location. Each offender's movement pattern is based on probabilities assigned to each road segment that are determined by examining the angle of the road in relation to an offender's journey from home, and how frequently the road segment is taken by commuters on the road network. After generating paths for all offenders, attractor locations are identified by calculating the most frequently travelled nodes in the network. The aim of the model is to investigate the influence that crime attractors and other major activity nodes have on the locations of crimes to better understand the target selection behaviour of known offenders.