Algorithms can predict with an impressive 70% accuracy whether a given area will have high or low crime levels

Sep 23, 2014 14:46 GMT  ·  By
Sample visualization of the high-level information provided by the Smartsteps platform
   Sample visualization of the high-level information provided by the Smartsteps platform

Predicting unfortunate events like crimes or accidents still belongs to sci-fi or paranormal movies, but certain aspects may become reality sooner than we might think.

Many studies indicate that it’s easier to predict the percentage of crimes in a certain region than to actually know when, who and where a crime will be committed.

We’re basically talking about statistics, which are based on raw data. However, these statistics aren’t quite accurate and usually require a lot of money.

However, it appears that researchers from the University of Trento (Italy), Telefonica Research, Barcelona (Spain), Fondazione Bruno Kessler, Trento (Italy) and MIT Media Lab, Cambridge, MA (United States) have found a way to predict crimes using mobile data and demographics.

The algorithm they created is based on the data received from Telefonica carrier and can automatically predict crimes with almost 70% accuracy and foretell whether a given geographical area will have high or low crime levels next year.

The data received by the researchers is not anonymous though, as they have requested info on certain people who have been tracked for specific periods of time.

“While previous research reports have used either background historical knowledge or offenders’ profiling, our findings support the hypothesis that aggregated human behavioral data captured from the mobile network infrastructure, in combination with basic demographic information, can be used to predict crime.

“In our experimental results with real crime data from London we obtain an accuracy of almost 70% when predicting whether a specific area in the city will be a crime hotspot or not,” reads the report.

Sadly, the data used by the study isn't quite anonymous

Mobile phones have become human extensions and they can now be used by third-parties to research human behavioral data. Mobile phones are now used to monitor people’s mobility patterns and demographic interactions, as well as to understand individual behavior towards various aspects of life.

With a little bit of mobile phone surveillance, it’s easy to predict peoples’ behavior, but the report says it can also be used “to map the spreading of diseases such as malaria and H1N1, as well as to predict and understand socio-economic indicators of territories.”

Only time will tell if these algorithms based on mobile data surveillance can help police departments to deploy their resources in areas where high levels of crimes are expected.

Obviously, no one wants to be tracked through his/her mobile phone unless the data is anonymous and unobtrusive. It does seem like a good start, but the methodology may raise some concerns.