The Design And Implementation On Customer Retention Framework With Telecommunications Full Service

As long as the telecom market opening-up step by step, the competition among the telecommunication companies is getting critical. Customer Retention gradually become the main influence for enterprise benefit. Telecommunication companies possess one-up data warehouse system at home and abroad, which offer conditions for customer churn and detainment analysis based on the data mining. The prediction of churn plays a major role in the analysis and operation system, which has built lots of forecast models and a series of warning indicators and templates. But the vacancy of complete detainment system and the disjunction of churn forecast and customer retention activity caused the forecast model cannot achieve the desired effects.Retention framework to maintain customer relationship management as an important component of the need for a set of theories and systems to support. In this paper, the current relevance of this issue for the theory and practice of application, and Rob Mattison in the framework proposed by WAR, based on customer life-cycle theory, based on customer value and customer segmentation theory to build a complete and effective based on the WAR/F telecommunications data mining framework for customer retention framework to maintain. For the existing full-service telecom industry background, and to maintain retention in the process of marketing decision support models and the lack of an effective means of supporting the status quo, focusing on the customer to identify the model of family relations.Identification of family relationships for modeling the data in the non-equilibrium problem, a good explanation in favor of a price-sensitive application of the promotion of CS-C4.5 decision tree algorithm. Family relations is different from the single dimension of data of data, modeling the lack of adequate and effective variable affecting performance modeling. In this paper, an analysis of the variables through the communication loop has two new variables: social circle distance coefficient, similar acts of communication loop coefficient; and the realization of the process in the model through data exploration and the use of CS-C4.5 algorithm to solve the problem of non-equilibrium data. Assessment of the final model show that the model was a better prediction performance, designed to achieve the desired objectives.