MINING IMPLICIT PATTERNS OF CUSTOMER PURCHASING BEHAVIOR BASED ON THE CONSIDERATION OF RFM MODEL

Association rules have been developed for years and applied successfully for market basket analysis and cross selling among other business applications. One of the most used approaches in association rules is the Apriori algorithm. However the Apriori algorithm, has long known for its weaknesses that generate enormous amount of rules and alreadyknown facts. In this study, we integrate the RFM attributes with the classical association rule mining, Apriori. Based on RFM model, two indicators, RF score and Sale ratio, are used as measure of interestingness. We propose two algorithms, DWRF and DWRFE, to mine for implicit pattern. In our experimental evaluation, the performance of Apriori, DWRF and DWRFE are compared. The result of our algorithms offers an effective measurement of interesting patterns. Moreover, the DWRF algorithm that uses the RF score as a measure of interestingness seems to be able to promptly reflect the fast-changing customer’s purchase patterns.