While the age of big data is coming, the new strategy of commercial companies has been switched from commodity management to customer management since many companies have huge amounts of customer data. In additon, identification of high-value customers based on customer data is also a common strategy adopted by these companies, with the consideration of increasing profits. Many companies consider exploring new customers the first priority, but the maintenance of old customers is often more important because the cost of exploring a new customer is five times greater than that of maintaining old customers. In this study, the customer data will be divided into two major sections, old and new customers. We then classify old customers into four group and calculate their RFM score. Using association rule learning, decision tree and cluster analysis, we try to find the relationships between old customers and customer values. A real-world data application is also discussed in our study.