The mainstream advertising of new capabilities for public safety data analysis may be leading to a smarter new world where police officers can arrive at the precise location of crimes before those crimes are committed. IBM’s current television spot for the company’s Software Package for Statistical Analysis (SPSS) shows a police officer “hanging out” at a convenience store (presumably sent there based on a prediction made by the SPSS system); when the prospective thief arrives, he sees the officer and promptly leaves the scene – crime averted.
The multiple benefits of such a capability are immediately obvious: Reduced crime, decreases in both violence and personal injuries, and lower insurance rates are just a few of the most obvious examples.
In addition, this next level of “smart” policing may enable law enforcement to truly do more with less by targeting enforcement in areas where crimes are the most likely to be committed. Today, in an environment of continuing fiscal austerity, new budget reductions may be inevitable. For that reason alone, a police force that is better informed by using analytics may be the most effective way to maintain a high level of public safety.
CompStat & Command Central
IBM’s offering represents one of several predictive-analytics solutions that have entered the public-sector marketplace over the past decade. Building on the traditional “CompStat” (computer statistics) geospatial analysis/“heat” maps, these new solutions add value by analyzing traditional factors – including but not limited to the time of day, the day of the week, weather conditions, and modus operandi. Such solutions are now possible because of the huge growth of structured and unstructured data already compiled from new cross-jurisdictional datasets. The resulting analysis then can helpentify the precise location of anticipated crimes – and can do so in near-real time – thereby providing a potential wealth of future tactical benefits.
In Caroline County, Maryland, for example, Captain James Henning used CrimeReports.com’s “Command Central” analytical tools to map a series of burglaries that had taken place over a three-month period. After that information had been mapped, then grouped in accordance with the modus operandi of the crimes themselves, Captain Henning applied a spatial-analysis algorithm – developed from the CrimeReports toolkit – that correlated the days, times, and locations of the burglaries that had already occurred.
Henning then was able to create a progression map of where the perpetrators were most likely to commit a crime during the next several weeks and months. By overlaying the spatial and time predictions on a map of his local police patrol routes, Henning was able to focus his department’s resources on specific areas and at specific times. “The rest was good old-fashioned police work,” he commented. Armed with the predictive analysis, the task force conducting the investigation entified the most likely suspects in those areas. Eventually, by using traditional physical and electronic surveillance of the most likely suspects as the case progressed, they were able to make several arrests.
Countering Crimes in Real Time The next challenge in the use of predictive analytics may be to engage law enforcement officers in the field, in real time, by entifying the most likely criminal activity through an automated alert system. Instead of relying on the traditional analysis of investigators reviewing data from various spatial-analysis tools, a “smart prediction system” could automatically alert officers already in the field by using specific locational information based on the real-time processing and analysis of the large volume of data constantly being ingested from multiple sources. Such a system could also be used to receive and analyze data in advance – received, for example, from incident reports, corrections and booking files, license-plate readers, suspicious activity reports, and electronic citations.
Coupled with other related factors – including date, time, and even weather conditions – the system could send a geographically defined alert to officers in a specific area (even across police jurisdictions) warning them that a particular type of crime may be likely to occur in a specific location. In addition, an officer conducting a routine traffic stop could receive a “predictive hit” based on the electronic submission of the driver’s name and/or vehicle tag.
A smart analytical system also may entify the person or vehicle as a possible “triggering event” for such an alert – again, based on the time of day and day of the week, the specific location, weather conditions, and similar data – and use it to inform the officer on-scene that the person stopped and/or the vehicle may be associated with a particular crime – either current or future. The alert may then lead the officer to look for additional “clues” that the person or vehicle may be engaged in some type of criminal activity.
One example of how this situation could develop: An officer might be alerted to the fact that the vehicle or person fits a particular modus operandi for the theft of copper wire. Using that data and/or other information – e.g., equipment usually associated with commercial power maintenance – the officer then might look into the trunk and/or backseat of the same car or truck.
The Bright Future of “Predictive Hits”
Although various technical solutions now exist that already can be used to trigger an automated alerting capability, many important procedural issues also must be addressed before such systems become routinely used in a tactical environment. Most importantly, the notion of “probable cause” may take on a more literal meaning if an officer were to receive a “predictive” alert.
Two key questions that might be asked are the following: (a) What actions, if any, would an officer be allowed to take based on this type of alert? (b) If hearsay from a reliable source can serve as a probable cause, would a so-called “smart” analytical system alert be considered reliable enough for an officer to take preventive action?
It may be quite some time before these and other issues are fully considered and subjected to legal scrutiny. In the meantime, new technologies will continue to be developed and police departments will almost certainly feel the operational effects caused by smaller budgets and reduced work forces. The predictive policing-enabled mobile officer therefore may be the best alternative to ensure that the American people continue to be protected from current and future crimes.
Rodrigo (Roddy) Moscoso
Rodrigo (Roddy) Moscoso is the executive director of the Capital Wireless Information Net (CapWIN) Program at the University of Maryland, which provides software and mission-critical data access services to first responders in and across dozens of jurisdictions, disciplines, and levels of government. Formerly with IBM Business Consulting Services, he has more than 20 years of experience supporting large-scale implementation projects for information technology, and extensive experience in several related fields such as change management, business process reengineering, human resources, and communications.
