The purpose of this empirical research is to find the effects of sampling method and the rate of binary data in using the several data mining techniques including statistical method. This research also suggests new classifier, ROC(Receiver Operating C...
The purpose of this empirical research is to find the effects of sampling method and the rate of binary data in using the several data mining techniques including statistical method. This research also suggests new classifier, ROC(Receiver Operating Curve), after finding the problems of traditional classifier. The result of research using automobile insurance fraud data shows that the accuracy of data mining using traditional classifier is affected by sampling methods but ROC is very stable for sampling method.