Predictive Software May Forecast Crimes
Last week, I attended the NIJ Conference 2010 and sat in on a fascinating session: “How Predictive Policing is Changing the Law Enforcement Landscape.” Predictive Policing has interested us at Evans & Chambers for some time. We do a lot of work with law enforcement agencies, and Predictive Policing is the newest movement in policing and is the marriage of law enforcement and technology.Think for a moment of old police shows and movies in which police officers use pushpins and a big wall map to visualize the locations of previous crimes. This is the traditional method of policing: manually analyzing past events and looking for patterns to emerge. The new methods use the past events, too, but new software and technologies can be used to predict areas where future events might occur. This allows officers to focus attention in those “hot spot” areas, with the goal of focusing police efforts in the hot spot area and perhaps catching the criminal in the act.
At the Conference, Lieutenant Raymond Guidetti, Manager of the Regional Operations Intelligence Center, New Jersey State Police, explained how the predictive software operates and talked about successful applications of this software in New Jersey.The software combines areas of interest, past events, and factor data to determine the future hot spots where crime is more likely to occur. Factor data comes mainly from open-source data, and might be information about how far a robbery occurred from a landmark like a bar, bus stop, or even fire hydrant. The software then finds other areas on the map with “geospatially similar” features, which may be used to predict a higher likelihood of crimes in that area.This process has been successfully used to predict shootings in Jersey City, NJ. Jersey City officers plotted previous shootings on a map, and ran an assessment using predictive software. The software highlighted areas with geospatially similar features, highlighting the hot spots that indicate a higher likelihood of future shootings. Police were then able to allocate resources to those areas indicated by the software. In order to test the accuracy of the hot spots, the next three instances of shootings were plotted on the map. All three were located in the hot spot areas indicated by the predictive software. In one particular hot spot, no previous shooting had occurred. Without the software’s identification of the area as a hot spot, the police would have had no reason to suspect that a future shooting would occur.There are quite a few other success stories of the software’s accuracy in Philadelphia, PA and in cross-county commercial robberies in New Jersey. This research is promising for the future of law enforcement and Predictive Policing. Evans & Chambers will continue to follow the latest news as Predictive Policing technology develops.
