Research And Implementation On User Churn Warning System For Telecom Package

Since the loss of users is a very serious problem in telecom companies,telecom operators in order to be able to get more market share,continue to launch a new package mode,hope to attract more users,but a variety of package mode,make the user dazzling,do not know exactly what kind of package economical and practical,which caused the instability of users in a large extent,resulting in the loss of user often occurred.The combination of telecom operators user package history,using data mining technology to find the characteristic factors of users package stability,and mining customer chum model based on these factors,for without loss of user pre sentence type analysis,according to the various types of users to push customized packages,retain users.The main works are as follows:(1)A large number of users of telecom operators accumulated use package based on the historical record,combining mining customer churn model,determine the target data source extraction method for a communication user of a plurality of continuous numerical attribute reduction fitting characteristics.The original data is pre processed,the data availability is taken into account,and the scale of data is compressed;(2)According to the problem of mining and application requirements of this paper involves the proposed composite strategy for mining mining association pattern mining and pattern classification,reduce the risk of overfitting is easy to produce classification mining,mining may cause interference free rules too much to predict subsequent loss mitigation relevance model;(3)Combined with the construction of association rules and classification rules,this paper designs a user churn prediction method based on K nearest neighbors,and realizes the prediction of user chum.On the basis of this,a prototype system based on data mining is proposed to predict the loss of users.At present,the system has been deployed,running applications.