Effect of Class Imbalanceness in Detecting Automobile Insurance Fraud

This article demonstrates a novel fraud identification methodology in auto insurance sector with the help of an adaptive oversampling technique (ADASYN). Initially, the data imbalanceness is removed from the original claim data set by employing ADASYN on the minority class instances. Three different classifiers, namely, Support Vector Machine, Decision Tree and Multi Layer Perceptron are further used for classifying the anomalous records from the normal ones. The efficacy of the models are tested with the 10-fold cross validation methodology. The performance of the model is demonstrated by several tests carried out on an auto insurance data set. Moreover, the outcomes justify the effectiveness of the proposed algorithm on a balanced data set over an unbalanced one.