In his story “The Minority Report,” Philip K. Dick envisioned a future when police could arrest would-be perps before they struck. We don’t have cops armed with precogs quite yet, but law enforcement agencies throughout the country are adopting a statistical approach to policing that can help predict when and where crimes are likely to occur. Pulling together burgeoning public data sets on everything from school schedules to home foreclosures and correlating them with crime stats, analysts can spot surprising patterns that help departments anticipate problems and identify emerging hot spots. Here’s how four cities are using predictive policing to get a jump on the bad guys.
Lock your doors—it’s spring break!
University of Memphis criminologists and local police mashed up business-analytics software with geotagged data to create a tool they’ve dubbed Blue Crush. It compiles crime reports and layers in variables like weather, lighting conditions, and proximity to concert venues, along with reporting from PDA-equipped beat cops, to find connections. For instance, the system noticed that colleges’ spring-break week reliably spawns a rash of burglaries.
Car thefts spread like chemical reactions.
According to UCLA anthropologist Jeffrey Brantingham, analysts can use “reaction-diffusion” models—typically deployed to study molecular interactions and bee swarms—to explain the incidence of opportunistic crimes like burglary and car theft. By modeling thieves and victims as simple particles in Brownian motion, his system can predict actual crime patterns and suggest preventive strategies. The LAPD uses it to figure out where to deploy extra patrols.
Building code violations = increased burglaries.
Cops in this Dallas suburb mapped residential break-ins against building code violations and found that crime skyrockets around dilapidated structures. For every “unit” of physical decay—even cosmetic things like broken windows, graffiti, and abandoned cars—there were six burglaries. Police have begun working with other city agencies to clean up areas that the maps flag as “fragile neighborhoods.”
Need to find a gang banger? Check the library.
A special Crime Analysis Unit identifies locations where gun crimes have been reported—not just robberies and shootings but also gun thefts and illicit possession—then factors in geographic details on things like bus routes and proximity to parks, liquor stores, and public libraries. (Yes, libraries: Turns out that gang bangers frequent them for free Internet access.) Combining that with seasonal data enables them to predict, for instance, when certain parks will become trigger points for gun violence.
Illustrations: Miles Donovan