Sorry, you need to enable JavaScript to visit this website.

Image CAPTCHA
Enter the characters shown in the image.

Predicting crime in the nick of time

Predicting crime in the nick of time
2010-07-082013-02-11innovationen
Can technology predict future crimes? If current research is to be believed, we may be able to use analytical techniques to create a Minority Report-style scenario, in which we know who is going to commit crimes, and in some situations, when it is likely to...
Published July 8, 2010 by Anthony Plewes in innovation

Can technology predict future crimes? If current research is to be believed, we may be able to use analytical techniques to create a Minority Report-style scenario, in which we know who is going to commit crimes, and in some situations, when it is likely to happen.

We already apply predictive analytics in a variety of areas thanks to advanced data analytics capability. We can predict the likelihood of demand increases for certain products and services in certain regions, for example. Why, then, can we not do the same for crime? Video analysis techniques are reaching a point where we can analyze peoples' gait, and even spot activities such as tailgating through doors. Surely, it is only a small jump to drawing conclusions based on those analyses?

Research is creating some interesting possibilities in the area of crime prediction. Nick Mallison, a Ph.D. student studying at the School of Geography within the University of Leeds, is creating what his thesis calls An Agent-based Model of Burglary in Leeds. He simulated the daily movements of potential burglars on two maps. The first depicted a portion of the city before a major urban development scheme, while the second simulated burglars' movements around a redeveloped city. "Interestingly, the model predicted that a few houses in particular will face a disproportionately high risk as a result of the regeneration, due to the changing behavior of the simulated burglars," he said.

Another group of researchers at the University of California, Los Angeles, have been studying human behavior to create mathematical models that can be used by law enforcers. They found hotspots in which crime naturally formed in an area where there was little or no previous crime. Although that particular work is not predictive, it is moving in that direction, the researchers say. Wilpen Gorr, a professor of public policy and management information at Carnegie Mellon, was using geographical information systems and large amounts of data mined from police departments in the US as early as 2003. His team assembled the data, plotted the offences on a digital map, and found itself able to predict monthly criminal activity before it happened, with 80% accuracy using standard business intelligence engines. Such information could conceivably help police departments to plan resource allocation, creating a better understanding of where to put their beat cops.

And as early as 2001, researchers were positing the use of smart cameras that could spot guilty parties based on an analysis of the way they walked. Researchers at Sussex University studied CCTV footage and found 'trigger' signals that showed when offenders were about to commit a crime. Car thieves would walk erratically and look about a lot, for example. Those about to commit acts of violence would walk more aggressively in a way that could be quantifiably measured.

As our video analysis techniques, combined with our geospatial information and number crunching improve, we are likely to gain ever greater insights into our behavior, before it happens. And that could lead to a reduction in certain kinds of crime while bringing its own problems. When technology begins to predict the way that people act and think, people will always cite Big Brother and the specter of an increasingly omniscient state. But, if it means that your house is less likely to be burgled while you are on holiday, is that something we should necessarily complain about?

Add comment

comments

  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd> <br>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

Email HTML

  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.
Image CAPTCHA
Enter the characters shown in the image.
Change the display