Spatial patterns of human movement in relation to crime occurrences: Focusing on vehicle load theft cases in Sendai city

This study explored whether the number of inflow trips of past vehicle load theft offenders, and of the general population as estimated by the results of a person trip survey, could predict the number of future vehicle load thefts in Sendai city. A geographically weighted Poisson regression (GWPR) was applied to account for spatial heterogeneity in the relationship between explanatory variables and the objective variable, and its performance was compared with a global Poisson regression (PR). Results showed that the GWPR had a better model than the PR. The number of inflow trips of offenders showed a positive association with future vehicle load thefts, while those of general population showed a negative association in suburban areas. These results indicate that human spatial mobility, both criminal and general, plays different roles in shaping the spatial distribution of vehicle load thefts depending on urban structure.