Nearest Neighbour and Statistics Method based for Detecting Fraud in Auto Insurance

Fraud is an actions that can cause harm to individuals or organizations. It can be used anomaly detection algorithm to detect the occurrence of a fraud. This study develops prediction modeling in the field of anomaly detection to detect the occurrence of a fraud using Nearest Neighbor based Method (distance based and density based) and Statistics Methods (interquartile range). In this study, we use open fraud dataset which has been used to demonstrate fraud detection capabilities. The dataset was a benchmarking dataset in the form of a minority report open dataset that is German car insurance data. The results of performance measurement are then compared with the results obtained by previous researchers using the same dataset. From the experiment results, the performance measurement obtained in the method used in this study is superior in some cases.