Based On Data Mining Telecom Churn Prediction

With the constant deepening of the reform of the telecommunications industry, telecommunications market split, customer selection of telecommunications products and room operators is growing. Telecommunications industry competition, escalating customer turnover rate, in order to maintain market share, operators use a variety of promotion to attract new customers. However, according to surveys, telecom operators to attract a new customer the cost of the flower is to maintain an existing customer of 5-10 times that of enterprises, the long-term than short-term customer loyalty to the client to obtain more profitable. Therefore improve the rate of customer loss of prediction accuracy will be directly related to the telecommunications enterprises survival and development, from the historical data mining in order to effectively use information to commercial decision-making has become an important application of statistical direction. This article is mainly in light of the actual projects on the telecommunications industry the loss of customers to conduct data mining, operator for a historical data has been lost through the client and the client Net data statistical analysis, excavation, loss of customers set up the model prediction. Model analysis resulted in the loss of customers the main factor in the prediction Net client at a certain period of time the possibility of loss.Through this research work, full use of the existing telecom operators data, from the perspective of many, many levels of analysis of the loss of telecommunications customers. Through neural networks, decision tree classification model, such as for off-grid customer loss prediction, the prediction model for the loss of performance analysis, evaluation of data mining model of commercial value. In this paper, a proposed telecommunications suited to their customer loss prediction model, experiments show that, based on the technology derived from the loss of telecommunications customers model, in the overall prediction accuracy rate guaranteed under the premise, so the loss of prediction accuracy rate of customers and coverage rates have been significantly improved. This is essential for promoting the field of telecommunications and correctly grasp the characteristics of the loss of clients, guiding Telecom customer retention strategy for a reference.