A study of enhancing marketing effectiveness by using data mining techniques-A case study of a hypermarket in central Taiwan

• Main Authors: Horng-wen Jang, 張弘文
• Other Authors: Chien-I Lee
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/n5v8dy

碩士 === 國立臺南大學 === 數位學習科技學系碩士班 === 101 === Today, more and more abundant historical transaction data is stored in Enterprise database. Without the maturity of data mining techniques, enterprise cannot easily discover potentially useful information about consumer shopping tendency that is previously unknown during the process of data analysis, thus developing a further marketing strategies to meet consumer expectations and enhancing enterprise''s competitive edge. In this thesis, Recency, Frequency and Monetary models are adopted as a clustering benchmark. Data mining techniques for association rules is first applied to the analysis of historical transaction data collected from each member group, and the combination of the usage of mining rules derived from the mining process and direct mails with Product Ads (DM) file help Marketing Planners to generate suitable recommended products for distinct member groups. In addition, rules derived from mining of non-member group is also used to develop a viable marketing strategy for member recruitment in this hypermarket. Finally, some valuable attributes derived from historical transaction information collected from members are used as the input variables of decision tree analysis, and help produce precious features from the shopping behaviors among highly valuable members. Cases of highly valuable members can be also used in comparison to those of non-member group in this hypermarket. The results show that potential and valuable association rules can be discovered from consumer behaviors of member groups in hypermarket, and these rules are quite suitable for applications with combination of a variety of marketing techniques; Moreover, through decision tree analysis, some derived attributes’ values of member groups can be extracted from historical transaction data of members and the features of shopping behaviors among highly valuable members could be thus discovered. The department in charge of exploiting potential members can adopt these useful information to contact those customers with the same features in the member recruiting , further promoting the profits and competitiveness of corporation.