Crime potential modelling: A GIS-based method using weights-of-evidence (Ontario).

摘要

This thesis was undertaken to evaluate the utility, effectiveness and accuracy of the GIS-based weights of evidence prediction model for the spatial prediction of residential break and enter crime in the City of Ottawa. The data-driven predictive weights of evidence (WofE) model was applied against six specific land uses of the urban environment as identified in the environmental criminology literature, including known occurrences of break and enters, in order to generate maps that could identify where future occurrences of residential break and enter might be located. Using the Arc-WofE extension written for the ArcView 3.2 GIS software package, a series of "crime potential maps" were generated that successfully indentified small areas of high potential (in the order of 60 to 95% probability) and were successful in placing future break and enter occurrences in these areas of high potential.