Research On Prediction Model Of Customer Churn In Mobile E-commerce Platform

In recent years the rapid development of Internet and the popularity of mobile devices,which has brought great changes to people's lives,but also brought great business opportunities for people,more and more companies began to try a new mode of operation,the mobile electronic commerce arises at the historic moment.Mobile electronic commerce development has matured at present,as a new era of the current characteristics,but with an increasingly competitive market,also brought a lot of challenges for the enterprise,how to obtain the biggest profit,win in the competition in the industry,become a mobile business enterprise issues of common concern.In numerous profit model,online advertising is the first Internet company which is the most widely used way of cash,mobile electronic commerce as a member of the emerging Internet industry,with a lot of Internet users,the congenital conditions for the development and promotion of its online advertising provides the basis.If we can grasp the profit model of online advertising,mobile e-commerce enterprises will be able to make a lot of profits.In order to promote the development of online advertising,it is necessary to establish a good relationship with online advertising customers,reduce customer churn,from this perspective,this article is for the mobile advertisements on a business enterprise to establish customer churn prediction model.In the research process of this paper,first of all,the problem of the loss of advertisers on the mobile business platform is confirmed.By reviewing the relevant literatures of customer relationship management theory,this paper sorted out the causes of customer loss and mastered the methods of customer loss management.On this basis,combining with the characteristics of mobile e-commerce platform of the growing business of online advertising,mobile e-commerce platform advertising is given the definition and classification of customer churn,which has been clear about the mobile e-commerce platform advertising customer churn problem.In this paper,the data mining theories and methods are systematically sorted out and summarized,so as to make theoretical preparation for the subsequent research.The second mobile e-commerce platform advertising customer churn prediction modeling framework design,combined with logistic regression model based on decision tree method of mobile e-commerce platform advertising customer churn prediction model,using WEKA platform for the decision tree analysis,using SPSS mining platform and logistic regression modeling,defines the prediction variables,this article and proposes the corresponding research hypothesis.Then,model verification was carried out,data collection and data preparation were completed,and two prediction models were established.Second,the logistic model of the original variables was directly adopted.To set up two models were analyzed the results and model evaluation,found that after attribute reduction model has higher prediction accuracy,so after attribute reduction model is chosen as the final loss prediction model of this article,and determines the ability to predict the biggest variable,customer churn research hypothesis is proved.Finally the prospect of research of customer churn prediction model are discussed in this paper,on a mobile e-commerce platform advertising customer churn probability was calculated,combining the theory of customer segmentation,RFM model is used to analyse the advertising high customer classification,developed for different types of clients save strategy,help mobile e-commerce enterprises to carry out the online advertising business better,improve the quality of customer service.