Predictive policing

Predictive policing is a phrase that is beginning to gather more and more usage in our field. It will be in the news even more, because it is being touted as the next big thing. I was among a group of practitioners and researchers invited to Los Angeles late last week by the National Institute of Justice for a conference on predictive policing. I served on a three-chief panel with Jim Bueermann from Redlands, CA, and George Gascon, from San Francisco to discuss what chief's expect to emerge from predictive policing.

In a nutshell, predictive policing is the practice of using data and analysis to predict future police problems and implement strategies to either prevent or ameliorate those problems. It borrows from the principals of problem-oriented policing, information-led policing, hot spot policing, community policing, situational crime prevention, evidence-based policing, and intelligence-led policing. Can you tell that we have a penchant for catch phrases in policing?

What distinguishes predictive policing from other paradigms is the emphasis on using analysis to anticipate problems--rather than responding to problems after they have occurred. At the simplest level, this might mean using crime analysis to determine the likely patterns of drive-by shootings, then deploying police officers to the areas and at the times these are most likely to occur in order to preempt the crime. At a more complex level, it might mean watching the trends in the spot copper price, and implementing strategies (such as legislation and scrap business monitoring) to reduce the marketability of stolen copper in advance of an anticipated spike in thefts.

None of this is exactly new, but predictive policing is gaining some steam because of the huge influx of data into police work, and our increasing ability to use these data to not only find existing trends and patterns, but to anticipate new ones. We know, for example, well in advance what the proliferation of a hot product like portable GPS devices will mean. When a new nightclub is planned, we can anticipate the impact on crime and police problems. You could build a pretty effective mathematical model to anticipate what happens when 500,000 square feet of retail space is developed, or when 600 two bedroom apartment units are built, because we have lots and lots of data about what goes on in similar situations.

Business has been using these analytics and models for a long time to make decisions: it's not just chance that there's a new Walgreen's on the corner, and Starbucks didn't just throw a dart at the map when deciding where to locate that new store. In policing, we are just starting to use our data to anticipate police problems. We are babes in the woods compared to the private sector.

Whether the predictive policing moniker persists and becomes part of the fabric of policing remains to be seen. There are all sorts of labels out there, some are a flash in the pan, some with great staying power. You might see articles, books, grant solicitations, and conferences galore on predictive policing. Conversely, the term might fade from use rather quickly. Regardless of whether the label gets sticky or not, these concepts are here to stay. Police departments will continue to improve in their ability to analyze data and formulate strategies based on these analyses.