Crime Distance and Background Probability in Geoprofiling

Posted on May 10th, 2010 by

Seminar by Chen Yu Yang, Gustavus Mathematics Major
Wednesday, May 12, 2010 at 11:30 AM in Olin 321

This work on geoprofiling builds on my collaboration with Sam Rethwisch and Eric Cox in the Mathematical Contest in Modeling 2010. Geoprofiling is a technique to locate criminals based on historical data, proposed by Rossmo in his doctoral thesis. This paper will first review the CGT algorithm, the most widely used model of geoprofiling, and some other models that have high potential of application. Next, the paper will introduce a modification of the CGT algorithm. The major contribution of this paper is the defining of a general concept of crime distance that incorporates a wide range of historical information. This distance is constructed based on a background probability, estimated through a Probit Model using Monte Carlo Markov Chain method.

Refreshments will be served at 11:15AM in third floor lobby of Olin Hall.


Comments are closed.