Font Size: a A A

The Study Of China Telecom Customer Churn Prediction Based On Data Mining And Application

Posted on:2019-11-29Degree:MasterType:Thesis
Country:ChinaCandidate:N R ZhangFull Text:PDF
GTID:2428330548495055Subject:Engineering
Abstract/Summary:PDF Full Text Request
With the changes in consumption concept of consumer,market competition situation and enterprise management concept,companies pay more and more attention to customer behavior research.The prediction of the loss of the customer according to the customer's behavior data is beneficial for enterprises respond to that changes timely and implement some retention policies on target.Simltaneously,the timely summary of factors based on the forecast results that could lead to customer outflow may assist companies find their own problems.Therefore,the establishment of such a customer outflow warning mechanism for enterprises has important significance.This paper aims to build a predictive model on the loss of telecom customers.Research process in this paper.First of all,comb the theoretical on management of customer relationship and the knowledge of data mining,which make the theoretical ready for the follow-up model research.The mainly theoretical knowledge is the knowledge of customer outflow management and customer value analysis in customer relationship management.The research model frame is constructed in this paper according to the theoretical knowledge of customer relationship management and data mining,that is,by selecting the Harbin Telecom customer sample as the research object,the preprocessing customer data is used in two ways to establish the loss prediction model: Using the logistic regression model directly is one of those two ways,to another one is to consider influence of the customer data.The RFT model is applied to differentiate customers with different values,after that,each type of customer data can be predicted by logistic model.At the end of this paper,the actual customer data of Harbin Telecom Branch is researched,which indicates that the model introducing customer value for customer outflow is more accurate than the one with logistic regression.By combining theory and practice,this paper establishes a customer outflow prediction model end up with a higher level of prediction in scientific and operable way.
Keywords/Search Tags:data mining, telecommunications, customer outflow, Logistic regression model, RFT model
PDF Full Text Request
Related items