Calif. town slices and dices its criminal data
- By Trudy Walsh
- May 03, 2002
Police in Westminster, Calif., had plenty of crime data. But the crime database couldn't distinguish between an auto theft and a joy ride, said Lt. Derek Marsh, director of administrative services.
If the database reported that a car had disappeared, officers had no way to determine whether it had been stolen by criminals or borrowed as a prank by teenagers. Lack of data was not the problem. The department had 10 years' worth stored in an Oracle 6 database management system running on an old Digital Equipment Corp. VAX minicomputer. The department hired part-time employees to input data for 85,000 criminal cases per year.
After trying several database analysis tools, department officials decided on C-Insight from MetaEdge Corp. of Sunnyvale, Calif. C-Insight uses a proprietary star schema architecture, which can segment or aggregate multiple data warehouses.
The Westminster police had several disparate data sources, said Gene Garrett, MetaEdge vice president of sales and marketing. One was in Oracle 6 and another in Microsoft SQL Server, plus separate county and state databases. 'C-Insight gives them a single view by aggregating all the data,' he said.
Marsh and his team exported the data from the Oracle database to SQL Server 7 running under Microsoft Windows NT. Using C-Insight, they can now query for elements such as time or geographic location.
'Now we can watch crime trends,' Marsh said.
C-Insight works with geographic information system tools from Environmental Systems Research Institute of Redlands, Calif., MapInfo Corp. of Troy, N.Y., and others.
The first step in crime analysis is to understand where crime happens, Marsh said. For example, many of Westminster's parolees live in halfway houses. 'Say there has been a crime spree at a warehouse that's a hop, skip and a jump from the halfway house. We'll center on it as a likely source of the crime,' he said.Going fishing
If a number of vehicle thefts occur, 'We can look at our map and say, 'Hmm, that section of the city is being hit.' Then we'll throw a bait car into the area and see who will bite,' Marsh said.
C-Insight has a predictive crime modeling tool that helped police in a pilot C-Insight program on the East coast catch a thief who was robbing fast-food restaurants, Marsh said.
'Everyone wants a magic pin map,' he said. 'They want to see the progression of events over time. What color were the stolen cars? Were they near major arteries?'
But for such data to be meaningful, Marsh said, 'it has be meaningfully presented.'
He wants to take the C-Insight analysis to the squad car level. 'This can really help the cop on the street,' he said. 'My mantra is to centralize our data for decentralized access.'
Trudy Walsh is a senior writer for GCN.