Presenting Bank Service Recommendation for Bon Card Customers : (Case Study: In the Iranian Private Sector Banking Market)

Data analysis and knowledge retrieval play a prominent role for bank industry. These organizations employ different transaction services such as Internet, Mobile Applications, Point of Sale (POS), and Automated Teller Machine (ATM). This paper considers the customers who use the POS located in different places for purchase through the Bon Card. In this study, we propose to recommend a banking POS service which have not been considered by each customer previously. Authors have prepared dataset that consist of triples of <;customer, product, rating> in order to recommend POS to customer. To determine the rating, an interesting concept is used that is also a marketing analysis tool. Therefore, the concept of Recency, Frequency and Monetary (RFM) are used as a method to determine the rating that each customer has given to the product. There are customers who did not rate some products in this dataset. Our goal is to offer POS to customers, so the matrix factorization technique is chosen.Matrix Factorization technique is a method for modeling the relationship between customers and products. Matrix factorization method helps us to predict unknown ratings. Since the rating value is based on RFM, so this prediction has two meanings: 1) It is determined what rating each customer can give for each product, 2) whatever the predicted value is larger number means that the algorithm recommends the best service for the best customer. In this paper, we present an architecture in banking area for recommending the specific POS for available customers with the help of a powerful approach called singular value decomposition. The proposed method has been validated using Root Means Square Error (RMSE).