Crowdsourced Crime Analysis, Redux

Earlier this week, I posted about the recent wave of burglaries of churches here in Saint Paul, and linked to a Google Map listing of the incidents, suggesting that armchair mathematicians, statisticians, criminologists, and any other interested party should take a crack at spotting patterns in the crimes, or, really, extracting any other information from it they could – call it crowdsourced crime analysis, if you will.

The Google map doesn’t contain a lot of details, and many of you, in comments or in emails, clamored for details, especially the times of the incidents and what was taken. So, I started making phone calls, and today became the recipient of many of the available pertinent details on these incidents.

Before we get to that, though, there are a few things that need to be covered.

This data – which is “public”, after all – was released by the SPPD, but that doesn’t mean it’s publication here is necessarily endorsed or supported by them. This whole “let’s see what random people on the internet can make of this” idea is mine and mine alone; I don’t work for them, by looking at this you’re not working for them (or for me); there’s no money in this for anyone, even if you do come up with an astounding insight. 🙂 Do this for the fun, and the challenge, as there’s little fame and even less money to be had, sorry.

That said, I think this is an interesting intellectual exercise – and for a good cause; I also think that crowdsourced analysis could be a viable, and valuable, tool for the law-enforcement community – if you look at this site, Entropic Memes, it gets read by several thousand people a day; if just fifty of you spend fifteen minutes looking at, playing with, and most importantly thinking about the information here, you will have – probably – collectively put in the equivalent of a week’s worth of effort by a single analyst, while bringing fifty new perspectives and sets of eyes to the problem. Get something like this Dugg or Slashdotted, and the case gets a couple man-years of effort in a day or two. Think of it as a proof-of-concept; if this is reasonably successful, this kind of thing could really catch on in these days of limited resources and tiny budgets. Collaborative web-based crimesolving? Why not; it has to be worth a try!

Secondly, a few words about what (I think) we should be looking for in this data. Patterns, for one – if we can identify a discernible pattern among these burglaries, there’s a chance future burglaries could be predicted, which is good; failing that, if and when someone eventually gets caught in the act, the details of the incident, if they fit a pattern of previous burglaries, could help connect them to past incidents. If nothing else, being able to identify what churches are at the highest risk of next being victimized would certainly be good.

Offender details would be useful, as well – some of these burglaries are almost certainly anomalous, random incidents, but the number and frequency suggest that one or more parties are responsible for a large number of them. How many groups are there, if any? What incidents are they responsible for, and what pattern or patterns do they follow?

Motivation would be really useful, but would require a lot of effort, even by folks who live in the area, here. Are these incidents truly random? Or are the churches being targeted for a specific reason? It’s not in the crime information released, alas, but an enterprising local investigative reporter (or citizen-reporter, in this age of “new media”) might want to talk to a couple of the churches in question, and try to find out what events if any took place at their facilities just before the burglaries. Is there a common thread connecting several of the incidents – a DJ, for example, who plays at weddings and uses the opportunity to “scout” the church for valuables? Digging up details like this is incredibly labor-intensive and time-consuming, and would actually make a lot of sense to crowdsource. (Anyone who wants to pursue this avenue of exploration might want to start with Progressive Baptist Church and Saint Paul Baylon, the only two churches in the city to be burglarized repeatedly; you get more data points for less effort that way. Yes, it seems fairly obvious, but don’t automatically assume it’s already been done. I don’t know, either way – like I said, it is ridiculously labor-intensive…)

Data concerning these events was released to me as two computer files, both Excel spreadsheets. The first contains details of all the church burglaries in the city over the last year, and can be downloaded here (31KB Excel spreadsheet). The second is a listing of all “commercial burglaries” (per the FBI’s uniform crime-reporting system, a burglary of a church is considered a “commercial burglary”) last year for which reports were written; it contains fewer details, but might be of interest to statisticians, or anyone who likes to play around with largeish real-world data sets, and can be downloaded right here (109KB Excel spreadsheet). Even if you’re not so interested in turning your analytical efforts to this endeavour, feel free to download these files and use them for whatever tickles your fancy – plot ’em on Google Earth, use ’em in some kind of brilliant mashup; whatever floats your boat.

I don’t usually try to promote anything I do here, but I’ll make an exception for this, because I think this is a fairly unique circumstance. If this kind of thing tickles your fancy, download the data and spend a few minutes playing with it. Tell your friends, share it with coworkers, your blog’s readers, people on forums you post to. Let’s find out what several hundred – several thousand? – smart people working together can do.

Published in: Geekiness, General, Security | on February 26th, 2009| 1 Comment »

Both comments and pings are currently closed.

One Comment

  1. On 2/27/2009 at 7:35 pm colin Said:

    if you’d like to share your data, we’ll put it on SpotCrime.