Fraud detection and prevention is important to Provident Financial’s Home Credit Division, a leading lender in the United Kingdom’s home credit industry, where agents make and collect cash loan payments in person.
Because cash changes hands in thousands of small, door-to-door transactions every day, Provident needed an effective system to detect and prevent fraud.
With PASW Statistics Base and PASW Modeler, Provident Financial’s investigators accurately identify and prosecute fraud, leading to a reduction in losses due to fraud each year.
Interested in Provident Financial? Download the PDF
Fraud is a general term that describes many types of white-collar crime, including theft of loan funds, manipulation of collections to increase bonuses, and other deceptive practices. Fraud is also a harsh fact of life.
In the home credit industry, the difficulty of tracking cash exchanges and the opportunity to commit fraud make detection difficult. Not only is fraud hard to detect, but strict U.K. consumer protection laws and rules of evidence also make it difficult to prosecute fraud once it is uncovered. For Provident Financial, a major source of personal loans in the U.K. and Central Europe, with an agent force of more than 25,000 serving roughly 1.5 million customers, that’s where PASW Statistics Base and PASW Modeler help.
Nearly 80 percent of the fraud discovered is identified using PASW Modeler.
To help reduce fraud and target offenders, Provident uses PASW Statistics Base and PASW Modeler. With these comprehensive data analysis and data mining products, Provident Financial narrowly focuses its investigation efforts and identifies agents who should receive “targeted visitations” from fraud investigators.
According to Paul Wilkinson-Smith, a fraud analyst in the Field Security Department within Provident Personal Credit (PPC), a subsidiary of Provident Financial, most losses are the result of agent fraud; however, in some cases, agents, management, and customers can conspire to steal funds. “With PASW Statistics Base, we were able to create a very effective statistical profile that tracked fraudulent activities back to our agents,” Wilkinson-Smith said.
PPC used a model built with PASW Statistics Base—using an analytical technique called logistic regression—to evaluate about 10 million customer records in its Sybase® database each week. The model runs on the same HP UNIX™-based server as the Sybase system, and generates visitation forms that are distributed to 60 area security managers for follow up. Security managers use these leads to conduct audits of identified agents and customers.
“The combination of a professional investigative field force and our PASW Statistics Base and PASW Modeler systems definitely has reduced the incidence and magnitude of our theft problem, as we detect and/or deter it earlier,” Wilkinson-Smith stressed. “In fact, nearly 80 percent of the fraud we identify is found using PASW Statistics Base and PASW Modeler.”
Wilkinson-Smith said PPC uses a combination of its own business knowledge, PASW Modeler neural networks, and rule-induction models to further refine the data it extracts using its PASW Statistics Base logistic regression model. “When we first brought in PASW Modeler, we already had many wins using our PASW Statistics Base models to profile fraud activity,” he explained. “We wanted to use PASW Modeler to help us profile fraudulent customers. Initially, PASW Modeler consultants helped us develop models that would work for us. Today, using PASW Modeler and PASW Statistics Base together, we have the two systems complementing each other very nicely.”
“We split our test data in half,” Wilkinson-Smith continued. “We use half to train the PASW Modeler neural network and the other half to test the network once we believe it is ready. We also test the network using raw data that has not been analyzed previously. We look at the neural network results and then manually evaluate that data to determine how well the network performed. We can tweak the neural network to some extent by adjusting thresholds, maximum performance levels, and other factors. While the neural net is self-contained, it can be adjusted and tuned for optimal performance.”
PASW Modeler further enhances the company’s fraud-detection capability. “We can look at the rules that PASW Modeler generates and test them against known data,” Wilkinson-Smith said. “This combination of modeling tools provides a very powerful data mining system that enables us to target our investigations and save our field agents an unbelievable amount of time.”
“Fraud profiles change very frequently,” he added. “PASW Statistics Base and PASW Modeler enable us to change our models as often as we need to. We can retrain the PASW Modeler neural network and rule-induction engine whenever we detect a shift in fraud patterns and profiles. I am a big PASW Modeler fan. It provides a superb visual development environment that saves us time and money. It’s brilliant!”
Worldwide Offices | Communications | Site Map | Contact Us | Terms and Conditions | Privacy Policy
© 2010 SPSS Inc., an IBM Company All rights reserved.
SPSS Inc., an IBM Company Headquarters, 233 S. Wacker Drive, 11th floor
