Research Of Optimized K-means And Decision Tree Algorithm For Telecom Customer Predict

With the continuous development of information technology,the scale of the telecommunications industry is more and more big.Customer service has become the most important competitiveness for a enterprise,and this is especially true in the telecommunications industry.The requirement of telecom users is becoming more and more high.How to maintain customer satisfaction and reduce customer retention,it has became a new subject for each telecommunication enterprises.In this paper,based on data mining and analysis related algorithms,according to the characteristics of customers in telecom industry,the model for customer churn analysis is established.The model could find the reasons of the loss of customers,help telecom enterprise making market strategy,and maintain customer loyalty.Based on the study of customer churn prediction algorithm in domestic and foreign,analyzing the reason of the telecom enterprise customer churn and countermeasures.Studied how to through the clustering algorithm to the customer group,adopted a modified k-average algorithm to get the user groups with similar characteristics,through customer clustering analysis is helpful to find customer churn tendency of customer behavior characteristics,which helps to retain measures according to different customer group characteristics make customers.Also studied how to use the decision tree classification customer churn prediction,on the basis of the basic algorithm of decision tree is studied,the decision tree attribute selection metric,decision tree pruning methods such as technology,classification model is established.Then based on the improved K-average algorithm and decision tree classification algorithm telecom customer churn prediction model is established,and the use of real data on the effectiveness of the model test,the accuracy of the model is as high as 80%.Based on the J2 EE technology,customer churn prediction system of telecommunications enterprise was designed and developed.This system can help telecom enterprise management personnel to carry out the customer churn prediction,the targeted to develop customer care and support the marketing campaign.