Application Of Data Mining Technology In Customer Churn Prediction

Data mining means to take out the potential unknown information mode and trend from the data, in order to improve the ability of market-decision and abnormity-test mode, to predict the tend in future on the base of experience in the past. It devotes to digital analysis and understanding, to find out potential technology in the data. It has become major object on applying of information and technology.In this dissertation churn prediction the Decision Tree Algorithm of data mining is researched which is applied the customer churn prediction. Firstly, the basic concept of Data mining is introduced, and several classification methods of decision tree are elaborated in detail, including ID3 algorithmic method which means dividing nodes on entropy attribute, C4.5 algorithmic method which could deal with the continuous attribute and absent value. Then an improved method of the C4.5 is proposed and its characteristics is researched. Customer churn prediction of the PHS (Personal Handy-Phone System), resulting of the serious competition in the telecom field, is an expression problem with practical meaning. The decision tree is applied to analyze the customer churn of the PHS in the field of telecommunication in this dissertation, including building mode, issue, compare, analysis, and reasoning conclusion with SAS. It has been carried on a deep exploration and attempt in this dissertation on the commercialized application of the theoretical knowledge.