The Richmond (Va.) Police Department is installing on Friday a map-based software application that forecasts locations most likely to experience crime in a specified time period based on historic and current criminal statistics.

The application combines SPSS Inc.'s predictive analytics with Information Builders’ enterprise business intelligence software, along with an analytical framework developed by the non-profit organization RTI International. It is designed to enable Richmond Police to use commercially available data mining tools to fight crime. "Law enforcement and intelligence data are ugly as sin itself, and they were never intended to be analyzed," said Colleen McCue, senior research scientist for RTI International, who has worked at the Richmond Police as the supervisor of the crime analysis unit. "It can be a nightmare."

The data to analyze where the local law enforcement agency should best deploy officers is pulled from records management systems and database repositories. These platforms hold information from citizen complaints received by 911 operators to crime reports. The data is processed and imported into the new framework built on SPSS' Clementine predictive analytics tool and Information Builders’ software.

The new software framework McCue designed enables the Richmond Police to pull law enforcement data into commercially available data mining tools and export the information into comprehensive reports.

McCue's challenge was to develop the framework that allowed the Richmond Police to use existing data resources and create a comprehensive method to view the data. It meant being able to import the data into the analysis software from disparate systems and export useable information into comprehensive reports.

The idea was to replicate the process computer models take when developing statistics based on pattern recognition and reasoning, she said.

Data mining software tools have been available to law enforcement in the past, but required clerks to extensively manipulate the information before being able to understand the statistical output after importing the raw data into a commercially available analytical software product such as Clementine.