Research And Application Of Data Mining In Telecom Customer Churn Prediction

Along with the rapid development of the telecommunication technology and further opening up of the telecommunication market, competition in China telecom industry is increasingly intense, and the customer contest is more and more fierce. The reorganization of the telecom in2008promoted the competition among telecom operators. Today, with the gradually mature marketing, every enterprise exist customer churn. Our customers are still a very unstable group. How to enhance the customer loyalty is a problem which every modern enterprise marketers has been researching all the time. For the communication market, the remaining customer market is gradually narrowing. From a cost point of view, to develop a new customer costs much more than to keep an old customer. Reducing customer churn means cost savings, which has been accepted for all telecom operators.However, the operation data is voluminous, but time is limited. How can we get the guiding and significant data analysis results and discover customer churn law in a reasonable time? Can we have any better method which can maintain and retain these hard-safeguarding customers, and not let them leave from the net easily?Data mining is a new technology appearing with the development of artificial intelligence and database technology. Its core function is to obtain useful information from the enormous data warehouse, which is provided for business analysis. For the telecom operators'perspective, data mining helps to find out the trend of business, to uncover the known facts, to predict the unknown results, and to analysis key factor which can complete the task. In recent years, data mining technology, with its powerful data analysis function, is widely applied in telecom customer churn prediction.The purpose of this paper is to research and realize a practical customer churn prediction model and a churn customer segmentation model, based on a deep understanding of the telecom customer churn knowledge and some common data mining algorithms, according to data mining modeling process. Meanwhile, the paper combines the sample data in2011of a provincial capital city's telecom operator Business Support System. The analysis results are evaluated and verified.On the basis of the evaluation and analysis results in customer churn prediction model and churn customer segmentation model, the paper obtains some effective customer churn rule sets, and gives out the forecast accuracy. At the same time, for churn customer segmentation result, the paper puts forward some opinions and suggestions on customer retention strategies.At the end, the paper summaries the algorithm study and experiment work, the problems encountered and the next step work.