Abstract
The development and application of computational data mining techniques in financial fraud detection and business failure prediction has become a popular cross-disciplinary research area in recent times involving financial economists, forensic accountants and computational modellers. Some of the computational techniques popularly used in the context of financial fraud detection and business failure prediction can also be effectively applied in the detection of fraudulent insurance claims and therefore, can be of immense practical value to the insurance industry. We provide a comparative analysis of prediction performance of a battery of data mining techniques using real-life automotive insurance fraud data. While the data we have used in our paper is US-based, the computational techniques we have tested can be adapted and generally applied to detect similar insurance frauds in other countries as well where an organized automotive insurance industry exists.
Highlights
The annual cost of settlements from fraudulent insurance claims in Australia was estimated at $1.4 billion dollars in 1997, which added $70 to the annual premium of each insurance policy (Baldock, 1997)
logit analysis (LA) had slightly superior classification accuracy, but all four models performed comparably with near 70% overall accuracy. These results support further testing of Cox and See5 models as automotive insurance classifiers, as Decision trees (DTs) and Survival analysis (SA) models are known for being sensitive to training datasets
Analysis on a much larger dataset is desirable. This would allow for hold-out sample tests as indicated by Wilson (2009). These hold-out sample tests should have more realistic proportions of fraudulent/legitimate claims rather than the synthetic even split in the data used here
Summary
The annual cost of settlements from fraudulent insurance claims in Australia was estimated at $1.4 billion dollars in 1997, which added $70 to the annual premium of each insurance policy (Baldock, 1997). These figures are likely to be much higher today as fraud is a growing problem (Morley et al, 2006). Automated statistical techniques for insurance fraud are designed to assist detection of fraudulent claims in a time efficient manner If successful, this would reduce the costs of fraud outlined in the previous paragraph. After that, concluding remarks on the problem as well as the methods are noted to round off the paper
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
