Locating Criminals with Geo-Profiling

Requires a Wolfram Notebook System
Interact on desktop, mobile and cloud with the free Wolfram Player or other Wolfram Language products.
Geoprofiling is a technique that has been used to locate criminals (with small success) or shark attacks, created by Kim Rossmo. It is based on two assumptions: the criminal does not commit crimes in a "buffer zone" close to his home, and he does not travel too far to find a new victim. Locators show the locations of crimes or shark attacks and the regions in red are the regions where the criminal's or shark's home is most likely to be.
Contributed by: Enrique Zeleny (January 2014)
Based on a program by: Jeremy Kun
Open content licensed under CC BY-NC-SA
Snapshots
Details
The formula used to find the probability of the criminal's residence is
,
where the sum is over crime locations and
.
Here is a point in the grid and
is the distance to a crime location with a choosen metric, either the Manhattan or Euclidean distance. The "buffer zone" is the area where the criminal does not commit crimes;
is its radius. The powers
and
affect the decay of the curve.
[1] J. Kun. "Hunting Serial Killers." (Jul 20, 2011) jeremykun.com/2011/07/20/serial-killers.
[2] R. A. Martin, D. K. Rossmo, and N. Hammerschlag, "Hunting Patterns and Geographic Profiling of White Shark Predation," Journal of Zoology, 279(2), 2009 pp. 111–118. doi: 10.1111/j.1469-7998.2009.00586.x
[3] SuperJustimagine. Geo-profiling on Numb3rs [Video]. (Aug 15, 2010) www.youtube.com/watch?v=AUub-ch_ULM.
[4] "The Math behind Numb3rs: Pilot Episode." (2007) numb3rs.wolfram.com/season1.html.
Permanent Citation