Law Enforcement and Public Safety

Predictive Policing and Quantitative Techniques in Law Enforcement

A brand-new rookie can have the experience of a 20-year veteran

With a wealth of data available to law enforcements regarding who, what, where, and when crimes are occurring, quantitative techniques -- especially predictive analytics -- are being used by more police departments to prevent crime before it occurs.

The RAND Corporation, a not-for-profit research institution and think tank, breaks these predictions into four categories.

  1. Methods for predicting crimes: These are approaches used to forecast places and times with an increased risk of crime.
  2. Methods for predicting offenders: These approaches identify individuals at risk of offending in the future.
  3. Methods for predicting perpetrators’ identities: These techniques are used to create profiles that accurately match likely offenders with specific past crimes.
  4. Methods for predicting victims of crimes: Similar to those methods that focus on offenders, crime locations, and times of heightened risk, these approaches are used to identify groups or, in some cases, individuals who are likely to become victims of crime.

Source: Predictive Policing: Forecasting Crime for Law Enforcement, Research Brief, The RAND Corporation, 2013.

The first one, predicting crime, is perhaps the most commonly discussed. Several Information Builders predictive policing customers emphasized this in their analytics projects. When police departments can predict where, when, and what type of crime is likely to occur, they can deploy police officers to likely hotspots.

When that happens:

  • police officers reduce crime by maintaining a more effective police presence
  • police departments reduce staffing and overtime costs because the policing effort is concentrated appropriately
  • communities attain a higher quality of life at a lower cost than they would through larger-scale police deployments

One successful client said that predictive policing "gives a brand-new rookie the knowledge of a 20-year veteran."

Data Integration in Predictive Policing

Predicting crime requires many different kinds of data. Some of it is maintained by police departments: Arrest records, crime reports, and similar data is the bread and butter of law enforcement. Police departments may have a hard time using that data for analysis, however, because it’s difficult to extract from the systems that are used to manage police operations. This information also needs to be blended together to get the broadest possible picture. As a result, data integration is a core requirement for predictive policing.

Internal data isn’t enough to predict crime, however. There’s a strong need for external data, which is maintained by external parties, to get a clear understanding of the crime environment. These might include:

  • Weather data. A light rain can move crime patterns from a local park to nearby storefronts where the awnings shelter people from the rain.
  • Event data. Concerts, sporting events, rallies, and other events can increase crime or concentrate it in certain areas.
  • Calendar data. People may change their behavior patterns on paydays, weekends, and holidays.

External data is often purchased from commercial sources or downloaded on a scheduled basis from government organizations. Formats and frequencies of delivery can vary widely. The data integration processes that are so critical for predictive policing must be extended to this highly variable data as well.

An Ongoing Endeavor

A law enforcement agency might successfully deploy a predictive policing system, but they’re never done devoting energy to the predictive policing process.

For one thing, the redeployment of police officers will change people’s behavior over time. Crime patterns from three years ago probably won’t hold true today. Predictive policing thus requires law enforcement agencies to revisit crime behavior patterns continually.

By feeding a rolling three to five years’ worth of crime data into a predictive policing system, law enforcement agencies can continually train their predictive policing models to understand how crime is shifting over time as predictive policing achieves its successes.

Sophisticated Technology Kept Cop-Simple

Predictive policing requires sophisticated artificial intelligence (AI) and machine-learning (ML) technologies to help computer systems learn what crime patterns to expect. Predictive policing software vendors often talk about “predictive analytics” as the solution in law enforcement scenarios.

But the police officer on the street doesn’t need sophistication. They need simplicity.

This isn’t limited to predictive policing. The most groundbreaking products in history were those that used sophisticated technology to simplify specific tasks: Apple’s iPod simplified music listening, Google reduced its interface to a single search box, and Netflix eliminated trips to stores -- and, afterward, dropping DVDs in the mail -- to return movie rentals.

Similarly, police officers don’t need “predictive analytics.” They need predictions: Where crime is likely to occur, who the victims are likely to be, who the criminals most likely are.

In fact, the more the individual police officer has to know about predictive policing, the worse the system is.

Good predictive policing gives law enforcement personnel everything they need in the easiest possible way: areas on maps, locations where they should deploy their cars, simple statements about where crime may shift since it is starting to drizzle.

Learn More

At Information Builders, we’ve been building building predictive policing solutions for law enforcement for well over a decade.

To learn more about our capabilities for predictive policing, crime data analysis, and more, download the fact sheet at left, check out the webinar "Next-Generation Law Enforcement Software for Data-Driven Policing," or contact us.

Learn More About Our Next-Gen Policing Platform