The objective of this research is to apply the data mining theories on membership/transaction databases for marketing knowledge discovering. The research uses a fast-growing coffee chain shop: 85℃ Coffee as an empirical case. The researcher proposes a RFMD model (based on the RFM model) to cluster the customers of 85℃ Coffee chain shops in Taipei. The study apply the C4.5 algorithm on the result of clustering and socioeconomic variables. As a result, there are four classification rules are created by this research. These rules can be applied in database marketing plan. In contrast with traditional market segmentation method (which employs customers benefit variables), this research discovers high value customer precisely via the classification rules. Hence, the results can easily achieve the target markets for marketing projects.