Predictive Policing

Defining Predictive Policing

In its simplest meaning, predictive policing helps law enforcement agencies respond more effectively to future crime by anticipating problems before they occur. In some ways, it is similar to how private businesses use social science to determine what customers might want to buy, and then offer these items to customers online. The methods of predicting shoppers’ behaviors are similar to methodologies used to determine criminal behavior. The differences lie only in datasets used. However, because of security and privacy laws and court decisions governing the public sector, private industry may be able to utilize more personal data than is available to law enforcement.

The next generation of police problem solving could combine existing technologies like computers, crime analysis, geographic information systems, crime mapping, and police reports with a few emerging technologies such as artificial intelligence. The future may well bring a system that can predict crimes before they happen. Of course, high-tech [Page 294]tools alone cannot solve crimes, but when trained crime analysts and police management combine the tools with their ingenuity, anything is possible in this high-tech world.

How Predictive Policing Works

The police know that strong-arm robberies near check-cashing businesses increase at the end of the month, that domestic and violent crime soars on hot days and drops during rainy days, that residential burglaries often occur on Sunday mornings while people are attending church services, and that Super Bowl Sunday is usually the slowest crime day of the year. But officers’ minds can store and remember only a limited amount of data. So when the police monitor crime data and query a computer system for historical and real-time patterns, they can predict, more systematically, over a bigger area, and across shifts and time spans, where crimes are likely to occur. More importantly, the crime-analysis software does not forget details, get sick, take vacation, or transfer to a different precinct (although it is susceptible to the usual kinds of technical problems). If commercial robberies were high in March 2010, crime analysis software may predict another spike in March 2011, and the police can then look at the types of businesses that were hit, their locations, and time of day. Such software can even analyze a robber’s modus operandi—what was said, type of weapon used, and so on.

The following are a few examples of how predictive policing (e.g., looking specifically at patterns of behavior) can be applied by patrol officers and crime analysts. Consider the example of a domestic dispute involving a man and woman. The officer arrives at the home, knocks on the door, is escorted into the living room, and all goes as it has gone on such calls 1,000 times—but now the officer is engaged in a struggle for his life. Had the officer’s mobile computer relayed information that the male subject was previously arrested for assaulting officers in another jurisdiction and recently arrested for driving under the influence, the officer might have approached this call differently and avoided such a struggle. In another example, there have been numerous copper thefts from construction sites in the city. With knowledge of the patterns of such thefts, the police can predict when and where similar thefts are likely to happen, rather than having to rely on parking a patrol car in various areas and hoping for blind luck. These are just a few simple examples of how predictive policing can be useful—simply by looking at patterns of behavior—to the patrol officer.

...