July 21, 2010 -- Memphis police are singing the praises of predictive analytics used in their Operation Blue CRUSH project, which they say has helped to lower crime in their city by 31 percent since 2006.
The Blue CRUSH (Criminal Reduction Using Statistical History) project grew out of Project Safe Neighborhood and a partnership with the University of Memphis Criminology and Research department and expanded to include collaboration and data exchange with the local and regional authorities including the Shelby County Sheriff's office, the U.S. Marshall's office and the U.S. Attorney General's office.
The project involves Esri geographic information mapping, RTIS, Crystal Reports and Microsoft infrastructure but the analysis focus is IBM's Predictive Analytics Portfolio. The initial project leaned heavily on SPSS, which was later acquired and became part of the IBM analytics platform.
The first use of SPSS, in 2006, was to address violent crimes and deadly weapons. "We were targeting where crimes were spotted in Memphis, we could create frequencies on gun crimes, where, what time, what date and when we pulled that data we were able to do a spatial join to our map and run queries and crank out a lot of high-quality information for the officers of our organized crime unit." says John Williams, a civilian supervisor at the Memphis Police Department. "We moved resources to those areas and saw immediate results in the number of arrests for gun-related crimes."
Data entry includes tables loaded automatically via the Memphis PD's record management system into tables, joined with information from other agencies and, now, global-positioning system data. Tactical special operations, DUI, canine, mounted and other units can be assigned to hot spots Williams says are remarkable well identified through analytics.
Besides the overall 31 percent crime reduction, which Williams attributes largely to the ability to pre-deploy idle resources, MPD claims violent crimes are also down 15.4 percent since 2006.
One indicator that there was a connection between Blue CRUSH and crime rate statistics was the observation that crime statistics actually rose in neighboring jurisdictions because criminals had displaced their activities because of better policing.
Williams says some suburbs of Memphis don't share information but he's open to the transparency that would bring. "We understand crime has no boundaries and we'd love to work with them because we know no one is isolated or immune."
That would also extend to an invitation to share information with the Memphis School District, which would likely raise privacy concerns. At the same time, a lack of collaboration and information sharing between police and other U.S. agencies was cited as a cause of poor awareness of events leading to terrorist attacks including 9/11.
The ability of Memphis to summon federal, state, local and educational contributions to the challenge successfully has drawn interest calls from a variety of police agencies, Williams says. The program is being modeled and so is a new real-time crime center that has centralized the police analysts who used to work in separate precincts with separate supervision.
"Everybody is trained, cross-trained and gets remedial training and hands-on help from the university's crime resource center, which is located here also," Williams says. "Once the analysts were in one place we wanted to hand more responsibility to each of them because they're the ones best suited to know even more about their precinct than the criminology department."
With tools and resources centralized, Williams expects faster deployment of resources and a bigger impact on crime statistics he says has been "an incredible improvement" to watch.
Criminal behavior appears especially well suited for predictive analysis through business and operational rules and tools to rapidly detect fraud, privacy attacks and financial abuse as well as overt acts of crime. IBM has worked with the city of Edmonton, Alberta on a program similar to the Memphis project; other vendors, including Information Builders and Microsoft, offer law enforcement-friendly analytics. In another example, the District of Columbia uses tools and software from SAS to identify and manage parole violations and other law enforcement issues.