Big Data Revives Crime Prevention

This post is also available in: עברית (Hebrew)

In simpler times, policing was about maintaining presence in a community. A local cop could know everyone and everything in his neighbourhood. And that’s the thing, it was his neighbourhood, one he patrolled on foot day in, day out. These are no simple times. With stretched resources, understaffed police department, an ever growing number of criminal elements, regular law enforcement officers are having a hard time keeping up and keeping us safe.

New York City Police Commissioner William Bratton says that Sir Robert Peel, England’s home secretary in the early 19th century, in creating the British Metropolitan Police Force had “nine principles of policing, which were focused on the prevention of crime.” Now, police forces the world over are focused on solving crimes, rather than preventing them. This tide is now turning, with modern tools allowing officers to avert crime, rather than only clean up the mess criminals made.

Big Data tools like the PredPol system have radically altered the way policing is done in many cities in the western world. PredPol works by dividing a precinct into 150 by 150 metre cells. Each cell is monitored both by historic and current activity. A cell is marked “hot” if it’s likely to be a centre of crime due to a variety of factors: location, history, current activity. Are there any sites that are particularly attractive to burglars, like bars, shops, or schools? As most criminals won’t brave below freezing temperatures, even the weather is constantly analysed.

One of the biggest factors in predicting crime is past criminal activity. Criminals follow patterns, return to strike at the same area again (as a familiar area is easier to work in), and that makes future activity easier to predict. Statistically, a burglary is likely to happen within some 200 metres over a 14 to 28 day period of a previous robbery. A stolen car is likely to be used as a getaway car when stealing another car. Much like in an earthquake, a crime is likely to be followed by a series of “aftercrimes” that follow a predictable pattern. Knowing where the next strike is likely to happen makes policing more effective and efficient, and we already have the tools for it.

“Predictive policing used to be the future,” said William Bratton, “and now it is the present.”