The Application Of Data Mining In Customer Churn Prediction Model

Although The Unicom Corporation has been done a lot of work in Stock market, each branch's maintaining activities is still in the status of stage type?practice type?irregularly shaped spots, the effect of maintaining is not obvious. In order to maintain stock work, we need to understand what customers need to maintain, which kind of customers deserve maintain, the resources needed to maintain the most accurate delivery, the most valuable customers, thereby improving the success rate of maintain performance and resource use.Data mining is the way of analysis through the known historical data, preclude the use of mathematical modeling to identify which of the underlying message, provide a reference for business decisions. With the continuous development of this technology data mining is increasingly being recognized by more people. Shandong Unicom operators in the loss of customers to the trend of early warning in the industry is an important application.At first, I introduce transformation mobile internet business transformation, as well as the three operators are facing the problem of customer churn. Secondly, I introduce some important concepts of data mining and data mining standardization process, and I describe the algorithm used herein, related concepts logistic regression and decision trees. Third, I introduce the core content of this article, I make the loss of the whole process of establishing early warning model by using Modeler software for Shandong Unicom target users and I use logistic regression and decision trees to establish the early warning model and compare the two summed advantages and disadvantages of both sides, eventually I use the method of logistic regression model as a last warning to use. Finally, for users to identify high-risk warning put forward the corresponding maintain strategy and a summary of the full text.