The objective of this study is to apply data mining techniques on membership & transaction database for customer value. The study uses specialty chain stores 36 month’s transaction data as an empirical case. The study was applying AHP and expert method to decide the weight of the RFM variables and further evaluate Customer Lifetime Value, CLV, through the weighed variables. Use these above variables as index of segmenting the customers by the Expectation-Maximization algorithm, divided into 5 major groups. Later through Association Rules (Apriori) and Sequence Cluster Analysis with filters and the retail characteristics, set the customizing cross selling and upselling strategy, in order to provide suggestion to the managerial marketing decisions.