Geographic profiling and criminal predictive analysis; investigative techniques that enable an analyst to identify where a serial murderer lives, or to predict, with surprising accuracy, where he will strike next. While it sounds like the plot of Minority Report, it's very real. Just ask Special Agent Regan Ross.
If you really want to get under Regan Ross’ skin, call it science fiction or compare her investigative area of expertise to a carnival side show. (I’ll just sit back and enjoy that verbal barrage.)
Regan is quick to point out the investigative techniques she uses to support local, state, and federal police departments all over the country are not ‘mumbo-jumbo,’ ‘voodoo,’ or ‘black magic.’ She can explain what she does a hell of a lot better than I can, and although the release of Jeopardy Surface is a couple of months away, I think it’s time to introduce you to her.
The following is the transcript of our recent interview.
SLH: So, Regan, why don’t you introduce yourself and tell Dear Reader what you do.
RR: At the Bureau?
RR: Sure. I’m Special Agent Regan Ross, and I’m a criminal intelligence analyst with the FBI, assigned to the National Center for the Analysis of Violent Crime at Quantico. I specialize in geographic profiling and criminal predictive analytics.
SLH: That’s quite a mouthful.
SLH: Let's start with geographic profiling? What is it?
RR: In general terms, it’s an analytic approach that in certain cases can help focus a criminal investigation. The techniques are based on statistically probability modeling, and the output of the analysis is a geographic profile. This geoprofile is a two-dimensional contoured map that predicts the probable area from which an offender operates. We call this location the anchor point.
SLH: That sounds—
RR: I’m not friggin’ clairvoyant.
SLH: I was going to say complicated.
RR: Oh. Right. It is complicated, but the complicated stuff, the algorithms, run in the background.
SLH: How does that work?
RR: The geoprofile models the psychology of spatial behavior and target selection to recreate an offender’s cognitive map from the distribution of his crime scenes.
SLH: You said crime scenes, plural. Does that mean it only works with a series of crimes?
RR: Yep, and only with marauder offenders.
SLH: Marauder, what’s—
RR: I’ll get to that, but I should first say that the primary goal of geographic profiling is to narrow the focus on an investigation. We don’t provide a big red ‘X’ on a map. Ideally, with enough data, we can provide a search area that can be used to prioritize leads and allocate resources.
Geoprofile used with permission of the author, Wesley English http://www.wesleyenglish.com
SLH: You mentioned, what was it, the psychology of spatial behavior? What did you mean by that?
RR: Good question, and that’s the foundation of everything, really. A serial offender’s spatial behavior is rooted in his operating location.
SLH: That’s the anchor point? Is that, like, his residence?
RR: Usually, but not always. It’s the place from which he leaves to commit a crime, or the place he returns to after committing a crime, or both. It’s typically his home, but could be a girlfriend’s house, workplace, favorite bar . . .
SLH: Where he hangs out.
RR: Right, and we know that serial offenders commit crimes close to their anchor point and the frequency of offending decreases as the distance from the anchor point decreases. That pattern is called the distance decay model. Wicked smart mathematicians have developed several distance decay functions to model the observed rate at which the probability decreases. But this model only works with marauder type offenders.
SLH: You were going to explain that.
RR: Right. Serial offenders can be broadly divided into two categories: marauders and commuters. Marauder offenders commit crimes outward from the anchor point. Commuter offenders travel outside of their normal activity space to commit crimes, usually in an attempt to limit the connection between their residence and the locations of the crimes. Geographic profiling algorithms don’t work with commuter type offenders.
SLH: So, what do you do when the offender is a commuter type?
RR: We have to ask a different question. Instead of taking a forensic approach to identify his anchor point, we have to shift to a predictive model. We ask the question, “Where will he likely strike next?” Simplistically, think of it like Wack-a-Mole. Instead of reacting to the mole as he pops out of the hole, we try to predict which hole he'll pop out of next. For a serial offender, we do that by geospatially characterizing and modeling the locations he chooses to commit his crimes to forecast where he may strike next.
SLH: I'm starting to get it, but to really understand it, I think Dear Reader may need to see how this works. Do you think you could give an example from an actual case?
RR: I’d love to. Sounds like a great topic for your next blog post. And let's start with a geoprofile for a series of cases Dear Reader is familiar with: Jack the Ripper.
SLH: Can’t wait.
Interview, Part 2 coming soon...