Data Mining Techniques in Fraud Detection
 

Rekha Bhowmik
University of Texas at Dallas
rekha.bhowmik@utdallas.edu
 

ABSTRACT

The paper presents application of data mining techniques to fraud analysis. We present some classification and prediction data mining techniques which we consider important to handle fraud detection. There exist a number of data mining algorithms and we present statistics-based algorithm, decision tree-based algorithm and rule-based algorithm. We present Bayesian classification model to detect fraud in automobile insurance. Naïve Bayesian visualization is selected to analyze and interpret the classifier predictions. We illustrate how ROC curves can be deployed for model assessment in order to provide a more intuitive analysis of the models.
 

Keywords: Data Mining, Decision Tree, Bayesian Network, ROC Curve, Confusion Matrix