Study On Customer Churn Prediction Based On Artificial Neural Network

The neural network technique is a kind of widely used data mining technique, it is usually applied to the field of pattern recognition, artificial intelligence etc. In this paper the neural network technology is used in the telecommunication trade. The result shows that it can help telecommunication enterprises to recognize the customer churning accurately and high-efficiently.Customers who will churn soon usually demonstrate the behavior different from other customers with high loyalty. For example: Telephone call will turn into a certain number regularly, or the incoming calls and outing calls are not in balance etc. After that, we can discern which is user churn incline characteristic, we can summarize these characteristics, and to predict a user's churn behavior.But these characteristic various, and interrelated, and judge success rate is very low. These characteristics are complicated, various, difficult to deal with. But the neural network can be discerned conveniently. That why we use neural network to predict customer churn.The data mining techniques used for discerning usually be known as decision tree, neural network and rough set. We use neural network because of its simpleness and accurateness.After we find the churn user, we need some informatics management tools, help customer manager to care for the churn user. In the end of the article, it describe neural network optimize.After adopting this kind of prediction method, obviously improved the churn rate of enterprise's customer of telecommunication and retrieved enormous losses for telecommunication enterprises.