Researchers in Italy have accurately predicted crime outbreaks in London – and all without plunging Samantha Morton into a perma-bath. Unlike Minority Report, the University of Trento team didn’t need pre-cogs or Tom Cruise on the case; they just needed mobile phone data. As Engadget reports, Andrey Bogomolov and his colleagues build an algorithm using anonymised data from London-based O2 users (including age, gender, and when they were home) during December 2012 and January 2013.
They first used existing crime stats to train an algorithm to predict whether an area would be host to more or less crime during the following month, which it did with 62% accuracy. Adding phone data made it 68% accurate. Info on whether phone users were at home apparently proved the most crucial element. ‘The proposed model could be used to predict new crime occurrence areas that are of similar nature to other well-known occurrence areas,’ according to the team’s report, Once Upon A Crime (geddit).
While it’s clearly not a fool proof system, it could potentially be used help police (in London, Italy, or wherever) to decide how best to allocate their resources. Mobile phone data has the advantage of being cheap and easy to collect, but as Medium points out, it also raises some serious privacy concerns. It might be anonymised, but it’s open to abuse by overzealous police forces, and the idea of monitoring private citizens without their knowledge or consent, whether it’s anonymous or not, could possibly be interpreted as creepy…
But not quite as creepy perhaps as the crime-prevention software unveiled in Maryland and Pennsylvania last year, which local law enforcement is using to predict who is at risk of reoffending and so should be more carefully monitored. It can apparently predict a future murderer 8 times out of a hundred. So what do you think: should we give up some of our rights in order to be a little bit safer, or should police keep their hands off our data?
Image via Dave Crosby’s Flickr.