The Application Of Data Mining In Retail Industry

Previous business model has been outdated with an extending and increasing customer demand in this age of breakneck economic growth. Along with growing requirements of service by chain retailers, it is significant to figure out how to satisfy customers real requirements of service. Meanwhile, there are more retailers with more customer data, however, it is quite urgent to seek out productive information from large amount of data. Data mining was born under such background, which refers to searching productive information from large databases, of which, Apriori algorithm of association analysis was first applied to supermarket marketing, then sales data and information resources of supermarket had been widely used. In China, however, even if many retailers had accumulated a certain amount of customer information, sales data, they didn’t process relevant analysis according to these data, also can’t make a significant contribution to business decisions using data mining. In this thesis, application research of convenience store(CVS) marketing decisions has been applied by association and clustering of real deal data, which come from 10 Zhijia CVS.In this thesis, firstly, it is introduced of brief knowledge of CVS and relevant theory of data mining, secondly, based on full understanding of data mining, excel and SAS are used to data pretreatment and simply description analysis, to learn about the whole picture of marketing of CVS. Thirdly, association analysis of CVS commodity purchasing is done with the EM module of SAS by using the pretreated data sets.In association analysis of commodity purchasing, in the first place, association rules between commodity category analysis was done, which use the traditional method of support and confidence, and all kinds of goods produced by cross-selling profit to evaluate value of association rules, accordingly work out shelf placement of commodities. Then, association analysis of commodities was done, to optimize commodities distribution, design sales promotion, and accelerate commodities recommendation. And clustering is used to research different commodities sales variation with time. The method proposed is reasonable and practical in marketing, which is supposed to call in reinforcements.