Differences in Geographic Profiles When Using Street Routing Versus Manhattan Distances in Buffer Zone Radii Calculations

Abstract

Geographic profiling (GP) is a technique used to uncover probable areas where an offender might be anchored based on a set of interconnected crime locations. GP applies a distance-decay function modulated by a buffer zone radius to the area under investigation to produce a probability surface indicating areas where an offender’s anchor point is likely located. Distance approximations are often utilized to create profiles, such as Euclidean and Manhattan distance. Though distance approximations have worked well in most applications of GP, existing research falls short in exploring the costs and benefits associated with a more accurate distance metric, such as street routing distances. This study examines the costs and benefits of street routing distance metrics by applying GP to 19 crime series data sets. Two profiles are generated for each series: one using Manhattan distances and the other using street routing distances in the buffer zone radius calculations with costs recorded at each step. The resulting probability surfaces are compared, and the results obtained conform with previous literature in determining that Manhattan distance approximations often underestimate the offender’s travel distances. The results indicate that street routing distances may be a viable alternative to distance approximations particularly in areas with low road density.