Research Of Customer Segmentation In Telecom Industry

Currently,the competition in the telecom industry is becoming moreand more fierce.With the development of smartphones and applications of Internet,instant messaging applications like WeChat and QQ have impacted on the traditional telecom business such as SMS business and voice business.So,customer segmentation and potential valuable customer identification are especially important.This paper expands index system,instead of the traditional customer segmentation index system which was made of the single index,for example,customer value.Base on the changes by the application and development of Internet,this paper establishes an index system,which is suitable for the telecom industry.During the study,this article integrates the idea of data mining into the index system.Based on the available amounts of raw data and data pre-processing,combined with industry characteristics of telecom,this paper establishes customer segmentation index system of online customers and offline customers.For the offline customer,the index system mainly includes the traditional characteristics such as voice,consumption.For the online customer,the index system mainly includes Internet behavior such as traffic and browsing behavior of APP.In addition,this paper analyses comprehensively the customer characteristics of telecom industry from many perspectives.In the analysis process,this paper analyses the characteristics of online customers and offline customers separately,not only includes the traditional statistics analysis,like age,sex,plan,terminal,voice and so on,but also analyses online customers APP use behavior creactly,from many aspects,and shows the result visualy.In the aspect of model improvement,in order to overcome the Kohonen SOM algorithm and K-Means algorithm shortcomings,this paper combines Kohonen SOM algorithm and K-Means algorithm and creates the Kohonen SOM + K-Means clustering analysis model.At first,Kohonen SOM makes an initial cluster,makes sure the quantity of K,the results are used as initial input for K-Means clustering.Finally,this paper divides the online customers into 'ordinary people','the best networker','reading fans','life fans','shopaholic',and divides offline customers into'inactive customer group','long-distance and night active customer group''low active costomer group','high call duration customers of the family network','higher consumption local customer group',and 'high consumption roaming customer group'.According to characteristics of different customer group,it proposes stragies and suggestion about precision marketing of telecom operation company.This paper also uses association mining to analyze the APP behavior of high-value customers based on the clustering results.The APP of customers will be recommend through the association mining combined with the telecommunications business rules screened 101 association rules.At the end of this paper,the main work of the full paper is summarized.The next stage research is prospected according to the shortage of this paper.