Evaluation of Destructive e-Banking Users’ Loyalty by Applying RFM

Abstract

World economy has encountered money laundering phenomenon and its destructive effects on economy of countries over the last decades. Money laundry detection is one of the fields in which data mining tools can be very helpful in detecting it. Nowadays, recognizing credible customers to present banking facilities to them is of high importance. On the other hand, by increasing cheating in banking, detecting the fraud customers is also important. In this study, the decision tree was being trained by providing a fuzzy decision tree and users’ behavioral vectors. The output of fuzzy tree can signify users’ risking behaviors. Some features of customers’ accounts need to be extracted in order to identify the people with high risks. For instance, the variance of money transaction and money transfer can be done. In another part of the study, RFM (Recency, Frequency, Monetary) features and MLP (Multi-Layer Perception) classifications were used to identify the loyal customers. The three-fold features of refreshment, frequency, and shopping amount was completely discussed for every customer, whereby the customers’ scores were established as well. This classification aimed to categorize the credible users of neural networks. The results of current research indicate that the represented techniques possess high precision values, in comparison with previous techniques.