Stability Of Telecom User Identification And Application Research

The world is heading for a digital information age.The advancements in science and technology has inspired the development of telecommunication industry.Over the past few years,the telecommunications industry become the fastest growing industry in the global economy,among which China mobile,C hina unicom and C hina telecom,the three major domestic telecommunications giant were listed abroad,which makes the telecommunications industry become the main pillar of the national economy development.But in recent years,with the reform and restructuring of C hina's telecom industry,expanding application of telecom popularization,the telecommunications market becomes increasingly competitive,basic services tend to be saturated,Average Revenue Per User represent a significant decline and customers' loyalty has been reduced.In the current market situation,in order to prevent the losing of old customers,promoting the transformatio n of new users to stable users,confirming the new users with potential stability,identifying the stable high-value users and developing personalized service according to features of the users for enhancing the user experience and promoting revenue growth,become the focus in telecom industry.Based on the data from users' bills,this article adopts the method of classification and clustering to build the model of user stability identification and precision of the user based on marketing,the main research mainly includes four parts:1.Cleaning and pretreatment of data,which includes the unification processing of data,dealing with the noise,missing value processing,etc2.According to the theory of user cycle,exploring the changing rule of the length of the network and the speed of the loss.The stability of the users is defined by the length of the network.The methods of random forest,adaboost,support vector machine(SVM)are used to establish the model of user stability identification.After the model comparison,the random forest is selected as the model of user stability identification.3.According to the stability characteristics of the user and user behavior attribute data,Hybrid clustering analysis was carried out on the users.The users were get into six classes,depicting the user portrait of each type of group.4.Marketing strategy and the suggestion.According to the each type of group users' stability and significant behavior characteristic,adopting the personalized service to achieve the precision marketing and enhance the users' experiences,so as to create value for operators.