This paper discusses some of the geographic and computational challenges in implementing a new software for geographic profiling that is based on Bayesian methods.
The author of this paper reports on a project that aimed to solve the geographic profiling problem by creating an operationally useful estimate of the location of the home base of a serial criminal, based on the known locations of the offense sites. The researchers developed and released a new software that uses Bayesian methods to attempt to solve that problem. This paper examines some of the geographic and computational challenges in implementing the new method, starting with an introduction to the geographic profiling problem, the author continues with the report on the researchers’ creation of their Bayesian model, and discusses distance and distance decay, target attractiveness, anchor point density, geography, normalization function, and the completed and released prototype software which implements the methods discussed. The author concludes that in contrast to the spatial distribution strategies, the researchers’ approach did not produce a single point estimate for the offender, and that the researchers’ method is able to account for a number of salient features of the local geography.