Membership Books Retail Chain Enterprises Based On Data Mining, Customer Relationship Management, Research And Applications

With the development of economy, China's book market t has been expanding constantly. Many large book malls have springed up recently. Currently, due to the sharp competition of the market, in order to seize the market, the retailers implement the membership system. Who have the reader, who can hold the markets. With the popularity of information systems, the companies have a massive sales figure for their members; however these data are not fully utilized. Therefore, how to dig out the high-value information, provide the readers with a more personalized service maintain and develop high-value readers, and develop effective promotional strategies, has significant meanings to the book companies.In this thesis, we use the qualitative and quantitative analysis method, combine theory with practice in data mining technology in the book of retail enterprises management application research. In theory, analyzes the customer management theory and the theory of data mining. In the actual research, Using CRISP - DM commercial mining process standards and enterprises membership sales data were studied by using the tools of mining SPSS Clementine. This research mainly include:First, Using Apriori algorithm based on actual needs establishing the purchase related model, Sales data were pretreated and group then correlation analysis to find the book purchase relationship. Setting appropriate associated support degree and confidence based on sales data characteristics. The analysis result is used to check effectiveness in this application. The rules will be used in member personalized recommendation Etc. The second, Members of the basic features of consumption information cluster analysis model is established by using K-means and Two-Step clustering algorithm. Membership sales data is processed and used by model. Contrastively analysis method is used to find out advantage of two algorithms in actual application. Choose an optimum clustering results and using histogram form based on average consumption amount members to analyze this kind member's consumption. Find high value and the high development potential members based on the results of analysis. With convenient Marketing Department formulate marketing plan. The third, member preferences realism analysis model of library category is established by using Two-Step clustering algorithm. Create the realistic mining data according to the analysis model. Member preferences category real degree is evaluated by using this model. Cluster number assessment by using clustering validity index Sil based on analyzing field attribute. Suitable for the number of clustering cluster results will play a key role and have an important impact on subsequent analysis. The real degree of membership registration information can offers the Marketing Department's marketing technical support. In addition to, the results of all kind of mining model can help enterprises to promote membership value and realizing Enterprise profit maximization.Through this thesis the research will allow enterprises to make full use of existing retail book sales data for members of the management level, improve customer's contribution, to help enterprises to obtain a stable readership and competitive advantage.