Fraud detection in the insurance using decision tree, naïve Bayesian and support vector machine data mining algorithms (Case study- automobile's body insurance)

One of the most significant problems with which the insurance companies are faced is to confront with the suspicious-of-fraud claims. With aim of acquisition of financial profit, these claims are taken place. Today, data mining techniques can assist in correctness and incorrectness of those who are insured. In this paper, three data mining methods of decision tree, naive Bayesian and support vector machine have been used to identify the frauds in automobile's body insurance. Results show that decision tree method with accuracy of 92.50% has a better efficiency compared to naïve Bayesian method with accuracy of 90.28% and support vector machine method with accuracy of 30.28%.