Crime is No Match for Math
Published Jan 1, 2016 8:00 AM
UCLA professors' algorithms help police departments predict when and where robberies and other felonies may occur.
When you think of successful policing, math may not be the first thing that comes to mind. But UCLA researchers are finding that algorithms can be key to preventing major crimes, such as car theft and robbery. The success of their method has led the Los Angeles Police Department (LAPD) to adopt an algorithm-based program for 19 of its 21 divisions that helps them deploy officers most effectively.
Traditionally, large police departments have turned to dedicated crime analysts to predict and prevent crime. These analysts weigh multiple factors each day to anticipate areas with a high risk for criminal activity.
But the algorithms developed by UCLA Anthropology Professor Jeffrey Brantingham, along with UCLA Mathematics Professor Andrea Bertozzi and other colleagues, improve the prediction process. The researchers take into account what types of crimes have recently occurred in an area, and at what times, in order to predict where and when crime is likely to occur. Their work enables patrolling police officers to be in the right place at the right time, deterring criminal activity in high-risk areas.
A recent study led by Brantingham focused on the LAPD and the Kent Police Department in Great Britain to verify that the algorithms were more accurate than trained analysts in predicting crime hot spots. In Los Angeles, over a period of 117 days, the algorithm program predicted the location of 4.7 percent of the crimes that occurred, results 2.2 times more accurate than the analysts’ predictions. The more precise predictions translated to an average of 4.3 (or 7.4 percent) fewer crimes per week. In Kent, the algorithm predicted 9.8 percent and 6.8 percent of crimes in two regions, compared to 6.8 percent and 4.0 percent predicted by the analysts.
To promote UCLA's predictive-policing math approach across the U.S. and overseas, Brantingham co-founded a company named PredPol. To date, more than 50 police departments, including those in Atlanta, Seattle and several towns in California, have implemented the program. In Alhambra, Calif., Police Chief Mark Yokoyama notes that after the department began using PredPol, the city’s car burglaries and thefts decreased substantially.
LAPD Chief Charlie Beck says, “I’m not going to get more money. I’m not going to get more cops. I have to be better at using what I have, and that’s what predictive policing is about.”