Minority Report for the Majority

Kenneth Wong
Cadalyst
Oct 16, 2007

Predicting where crime would happen next sounds like a wild idea, a concept possible only in the realm of science fiction, like Phillip K. Dick’s short story “Minority Report.” But lately some technology vendors have been combining crime statistics, weather data, geospatial data, and predictive algorithms to create a magical brew that can forecast when and where the next crime wave will most likely hit.

In Richmond, Virginia, Police Chief Rodney Monroe was once a skeptic. “When my IT people told me we could predict where the crime would occur, I just scratched my head,” he recalled.

In 2005, according to City-Data.com, the Richmond Police Department (PD) registered 755 criminal incidents, which was more than twice the national average of 325. Monroe could have easily predicted the crime rate would remain so — or get a lot worse — if no drastic measure was taken. So he decided to create a Law Enforcement Analytics (LEA) dashboard combining the cumulated 911 reports, police reports, citywide events, and graphical data.

The Birth of LEA
LEA is a joint operation, so to speak. The dashboard is built on the WebFOCUS BI (business intelligence) software platform from Information Builders. Using ESRI ’s ArcGIS software, it displays aggregated crime data, mapped by location and sector. Clementine, a data-mining and analytics software from SPSS, is responsible for predicting possible criminal patterns based on civic events, weather, and historical data.

LEA came online in various phases between 2005 and 2006. According to Intelligent Enterprise magazine, “Richmond PD’s new system integrates and analyzes information every eight hours (at the end of every shift), and it delivers timely insight to top officers responsible for each city sector (with broader deployment to individual patrol cars planned for this year). The dashboard not only shows up-to-date crime statistics mapped by area, it delivers alerts that users can customize based on their role and area of responsibility.”

Kevin Quinn, vice-president of product marketing at Information Builders, remembered the demonstration of LEA. “I was pretty impressed,” he said. “For example, on Halloween, they [Richmond PD] predicted there would be an increase in vandalism-type crimes, but if it rains, it reduces crime in certain areas, so based on that, they were able to deploy their forces efficiently.”

Monroe, who once admitted he felt like “a fish out of water” because he didn’t know “the first thing about technology,” now says, “I’m a believer.” In March, he flew to Chicago, Illinois, to accept the 2007 Gartner BI Excellence Award on behalf of Richmond PD.

GIS to BI: Welcome!
Mergers and acquisitions have become the quickest method of growth in the high-speed business world these days. Information Builders’ Quinn reasoned, “Large organizations now have so many different sources of information from the ERP [enterprise resource planning] systems they have inherited. We have a single platform for tapping into every one of them. And nearly all of them have some kind of location information — ZIP codes and area codes, for example.”

So Information Builders is now pitching its WebFOCUS BI platform as the place where business data (such as inventories, sales, customer locations) can join hands with geospatial data.

Richmond PD’s LEA dashboard uses a combination of technologies — ArcGIS from ESRI, WebFOCUS from Information Builders, and Clementine from SPSS — to predict where crimes will most likely occur.

“ESRI’s GIS and WebFOCUS share a common Java architecture enabling developers to easily add a GIS component to business intelligence applications using a set of Java APIs,” noted Information Builders. “Little or no training is required to use the GIS/mapping functionality of WebFOCUS. End users view the new mapping function as part of their existing applications. Analysts and power users can easily toggle between a map and business intelligence application to suit their needs. In addition to having information displayed on maps, users can select data from a map and easily move it into a report for detailed metrics in user-selectable formats including HTML, PDF, and Excel.”

WebFOCUS software is marketed as server solutions. A single-processor Windows server configuration, for example, starts at $25,000. A larger Unix box or mainframe configuration may cost as much as $130,000. This includes unlimited Web deployment, Quinn said. The WebFOCUS GIS Adapter add-on is priced at approximately $15,000-$20,000. Users will need an ArcGIS license, which is also available from Information Builders.

The Next Phase
Until recently, data-mining and GIS have largely been confined to the government sector, Quinn observed, but the affordability and proliferation of technology will make it possible for the commercial and private sectors to adopt these solutions.

The merging of business and geospatial data represents “a huge opportunity for [GIS developers] to penetrate the commercial sector,” Quinn pointed out. “These applications will help companies study where their customers are, who they are, and how they do business.”

Now that GIS is forging an alliance with BI, what’s next? Apparently, tracking moving objects. “Tracking the real-time locations of trucks, trains, and people via cell phones, GPS, and RFID [radio frequency identification],” Quinn predicted.