A Comparative Analysis of Various Clustering Techniques for Sales Fraud Detection

Abstract

Nowadays, sales fraud increasingly becomes common in our society. In this regard, sales fraud detection must be required to prevent such schemes in every organization. This paper deals with the implementation of various clustering techniques such as k-means, k-modes, hierarchical clustering, partitioning around medoids (PAM), and also the self-organizing map (SOM) techniques to efficiently analyze and detect the presence of sales fraud. The results of the comparative analysis state that the self-organizing maps can be used efficiently to detect sales fraud as compared with the other considered clustering techniques.