Detecting Valuable Customers Using the Trade Patterns of Financial Transactions Applying Integrated RFM and OLAP

Abstract

One of the challenges that banks are faced with is recognition and differentiation of customers and providing customized services to them. Recognizing valuable customers based on their field of business is one of the key objectives and competitive advantages of banks. To determine guild patterns of the valuable customers based on their transactions and value of each guild for the bank, the banking tools on which the customer’s transactions take place need to be surveyed. Using deeper insights into the value of each guild, banks can provide customized services to ensure satisfaction and loyalty of their customers. Study population was comprised of the holders of point of sale (POS) devices in different guilds and the transactions done through the devices in an 18-months period. Datamining methods were employed on the set of data and the results were analyzed. Data preparation and analysis were done though online analytical processing (OLAP) method and to find guild patterns of the bank customers, value of each customer was determined using recency, frequency, monetary (RFM) method and clustered based on K-means algorithm. Finally, specifications of customers in the most valuable cluster were analyzed based on their guilds and the rules were extracted from the model developed using C5 decision tree algorithm.