Nate Berg is a freelance reporter and a former staff writer for CityLab. He lives in Los Angeles.
Santa Cruz is testing a new system that attempts to predict likely locations of future crimes
An algorithm has joined the fight against crime in the California beach town of Santa Cruz. Using data from reported crimes in the city, the program tries to predict where and when similar crimes are likely to occur, as a recent article from New Scientist explains.
The geographical predictions are being used by officers to add extra patrols in areas where crimes seem likely to occur. It’s part of a field test of a predictive policing program developed by George Mohler at Santa Clara University, who notes that some crimes, like burglary, have predictable patterns. One burglary in a neighborhood one day tends to be followed by more in the same neighborhood days later. Using data to back this assertion, the program utilizes daily crime statistics in Santa Cruz to suggest where and at what time of day crimes are likely to strike and where police should consider patrolling. The software will also be tested this fall in Los Angeles.
The program focuses on three types of crimes prone to spatial patterns: residential burglary, auto burglary, and auto theft. Each day it creates a list of 10 hotspots – each 150 square meters – where each of these crimes is likely to occur. The hotspots are added to a Google map, which is given to officers each day. And the maps can be wildly different depending the day.
“A key feature of the experiment is that, because crime is dynamic, hotspot areas can change on a daily basis. Police patrols therefore must adapt to this dynamic risk,” notes the Santa Cruz Experimental Predictive Policing Software website.
The maps are intended to help stop crime before it happens, but also to increase policing efficiency. Like a lot of local law enforcement agencies, the Santa Cruz Police Department has had to trim its rolls in recent years. The algorithm could serve as a way to lower the cost of crime fighting by making it more accurate.