Study On Social Network Customer Segmentation Based On Data Mining Usering R

With the development of web2.0 in recent year, social network (also social network service called) as the physical carrier of web2.0 has also experienced tremendous growth. The growth embodies in two aspects:for one thing increasing users of SNS show a differentiated demand; for another growing social network sites open up a smoking battlefield of network. How to identify and provide differentiated services to help enterprises to maintain old customer and develop new customers in the fierce competition is the main topic of this article. The paper researches on customer segmentation deeply and finally gets the practical rules on the basis of social network theory and customer segmentation theory, through clustering and decision tree.This paper is mainly organized on two stages, firstly on theoretical research and then empirical research.In theoretical research, firstly the author draws a conclusion in Chapter 2 on the choice of method and variable of customer segmentation, through research on both social network theory and customer segmentation theory. Then in Chapter 3, the article starts with an introduction to the theory of clustering and decision tree in data mining. Based on that, a comparison in all kinds of arithmetic is made to help us in final data mining model selection.In empirical research, as research tool R is used. Firstly a JAVA crawler is developed to crawl user data from renren; after that, the author calculate 100 random samples using DIANA algorithm and get the number of clusters which can be divided; Then the author go on clustering work with PAM algorithm on the overall sample and finally get 4 cluster; At last, with the conclusion got in last step, the author draws the corresponding trees and rules using C4.5 and CART decision tree algorithm to guide practice.