Research On Customer Segmentation Of Retail Based On Transaction Data

Under the increasingly fierce competition environment, development of China’s prosperous consumer market gradually slow down. Tracking customers’change, understanding customers’needs, meeting customers’expectation and identifying customers’segmentation plays a significant role in the process of retail marketing, when the markets services their consumers. On the other hand, under the influence of social factors and the tremendous change in the market, the development of information technology provides the marketing opportunities and challenges. The development of database marketing helps the market collecting original customer data, segmenting customer, and guiding enterprises to carry out differentiation management to the customers, which has been the key point to the enterprises.On the basis of reviewing the related theory and research about customer segmentation and data mining, the paper analyzed customer segmentation model of the department store, and combined with the actual case--YS Company to analyze the customers. Then summing up the corresponding customer groups’characteristics and combining customer value, the paper came up with targeted marketing strategies for the enterprise.In this paper, combining with the instance of YS company the research’s main conclusion are as follows:Due to the behavior variable, the dynamic behavior segmentation model was put forward. According to the RFM analysis, the paper combined with the actual case, introduced the attribute--customer purchases and adopted the method of clustering to segment customers.Due to the commodity the customer purchased, static category segmentation model was put forward. Retails provides diversified commodity for customers to meet customer expectations, but characteristics of customers are dispersive. Based on the categories of customer purchasing, two new indicators are introduced as SC_N and SC_M. Then, combining CHAID decision tree, identifying the importance of each category, and obtaining the effective characteristics of the segment is done.