Police Forces Are Using Computer Algorithms To Predict Crime

The concept of pre-crime has been circulating among authorities, philosophers, and the general public since The Minority Report novel was turned into a Tom Cruise blockbuster in the early 2000s. What exactly is pre-crime? The idea is when an offense is going to occur and measures are taken to prevent its happening. BBC reports that thanks to tech firm PredPol, authorities in some cities are using this technology and data analytic algorithms to improve crime detection by a range of 10 to 50 percent.

According to anthropology professor Jeff Brantingham, PredPol — which stands for predictive policing — was inspired by experiments that were ran by the University of California and the Los Angeles Police Department.

“That study demonstrated that algorithmically driven forecasts could predict twice as much crime and, when used in the field, prevent twice as much crime as existing best practice.”

This innovative software attempts to predict approximately when and where specific types of crime will occur over a 12 hour period. The algorithm is reported as being updated every day as the new data rolls in. These predictions are displayed via a color coded map. On the map, red colored boxes of about a 500 square foot areas represent high risk crime areas. Officers are encouraged to spend at least 10 percent of their time in those areas.

Brantingham states that machine learning allows for PredPol to not only analyze data, but to draw conclusions, as well as make connections between greater amounts of data that even human analysts have not been able to cope with.

Of course, as with anything, there are critics and sceptics of PredPol. These sceptics are claiming that the technology is nothing more than a pseudoscience. Their reasons behind this remark claim that “crunching time data to make informed decisions on police deployment is nothing new.” This, they say, is the same “hot spot analysis” that many forces have already been using.

Those who are for the concept of PredPol state that traditional hot spot analysis is merely reacting to the events of yesterday, rather than anticipating the possibilities of crime tomorrow.


“Machine learning provides a suite of approaches to identifying statistical patterns in data that are not easily described by standard mathematical models, or are beyond the natural perceptual abilities of the human expert.”

More than 50 police departments in the United States are now using PredPol software, as are several forces inside the United Kingdom. According to Key Constabulary, it is reported that street violence fell 6 percent after only a four month trial use of PredPol.