Study On The Customer Stability Of Network Shops Based On Decision Tree

Nowadays online shopping has been matured gradually. The scale of users continues to grow, and the online mall and merchants have been developed correspondingly. Based on the concept of data mining, using the database technology, this paper discuss the issue of customer stability and loyalty to the online stores in online e-commerce platform.Firstly this paper determines more than 30 factors that describe status and sales volume of an online store on an online mall as the development evaluation index of the store. Then using a trisection method to process the origin date of each attribute and complete the standardization and unification of attribute data. Later on using the decision tree ID3 algorithm of data mining to find the factor and attribute that affect the online store stability. Finally 4 factors that directly affect customer stability to a store are determined: popularity of collection, logistics service quality, the seller's credit of the store and comments of the previous customers.In the process of mining though ID3 algorithm, in order to remove the attribute numbers' effect on the amount of information gain and also to standardize data of each attribute that collected from online stores on an online mall, this paper uses the method difference average on the value of each attribute, trisects the difference between the minimum and maximum value of each attribute, and divides all attribute to 3 values: high, medium and low. This way decrease the dependence of ID3 algorithm on amount of attribute value, allows the data with different kinds of specification has a unified expression and satisfies the requirement of ID3 algorithm. Also using the trisection method has made the complexity of attribute value simple and decreases the amount of computation. When dealing with more complex problem, it can greatly increase the computational efficiency.At the same time though the mining of attribute information of online stores on an online mall, this paper has finally determined a few simply attribute index to better reflect the customer loyalty index of online stores. According to the relationship between customer stability index and online store attribute, the mining results effectively and quantitatively offers an objective evaluation angle for the costumer that cannot direct access to the online store to conduct a comprehensive evaluation. Also it provides concrete reference for the management of the shop on the platform of online mall, the development of online stores and customers' evaluation on the online stores.