Fraud Detection with SAS ® Data

Fraud is a significant problem in many industries, such as banking, insurance, telecommunication, and public service. Detecting and preventing fraud is difficult, because fraudsters develop new schemes all the time, and the schemes grow more and more sophisticated to elude easy detection. Many organizations have implemented fraud detection and prevention systems based on SAS data mining to help them stay ahead of the fraudsters and avoid losing money. This presentation will provide an overview of different data mining techniques that have proven successful in detecting different types of fraud. Using case studies, successful implementations in different industries will be described. INTRODUCTION Organizations are exposed to fraud in many different ways. Fraud can be described as a group of intentional acts made for personal gain by causing significant losses to organizations. Besides the losses, their reputation can be affected, as well as their customers’ loyalty. Using insurance as an example, Figure 1 shows that there is a wide spectrum of fraud, ranging from opportunistic to premeditative. Opportunistic fraud is committed when the opportunity arises without much planning going into it (for example, when a car accident occurs, the claimant decides to inflate the claim to receive more money than he or she is entitled to). Usually, with this type of fraud, only one person or a very small group of people is involved. On the other side of the spectrum, there are potentially large groups of offenders who invent schemes to defraud insurance organizations. For example, they stage accidents to make large — and illegal — claims for fake injuries and car damage against an auto insurance company. Figure 1: Spectrum of Fraudster Types in Insurance Fraud is a very apparent problem across all industries. It ranges from external fraud committed by offenders outside the organization, either customers or not, to internal fraud, where employees of the organization are part of the scheme. A proactive approach to fraud detection helps to reduce the effect of fraud on organizations. Through the use of advanced analytics, such as data mining, organizations are able to detect ongoing fraud earlier and take preventive measures to minimize losses. ANALYTICAL APPROACHES TO FRAUD DETECTION