Since the days the investigating officers used ”pin maps” to locate and to think about crime events, crime mapping has become widespread thanks to spatial analysis mainly supplied by GIS-like software. In particular these methods suit well to geographic profiling devoted to crime series characterised by a single offender and hence limited space and time variability. Although spatial techniques are now regularly performed to delineate an offender’s area of residence, the temporal dimension is underemployed due to the wider uncertainty of time records. This paper proposes a methodology based on a least-squares adjustment in order to cope with this temporal issue for determining the most probable offender’s residence. Moreover, a chi-square test is described to check the significance of the solutions suggested by the method. The process is carried out on the real road network which has been discretised (rasterised) for computing convenience. Three simulations show the validity of the reasoning. Finally the main time and speed assumptions introduced in the model are discussed paving the way for further research.
展开▼