Research On Technology Of Predicting Customer-Churn Behavior In The Wireless Telecommunications Industry

Along with the reformation and development of the telecommunication system,China had entered WTO and telecom market was open to the world. The big telecom companies of abroad participate telecom work of our country constantly, the telecom trade compete-tion of our country will gradually vigorous. Therefore, establish the valid marketing and service strategy science and reasonably to reduce biggest the customer churn rate the biggest, especially the excellent and important customer become particularly important. Therefore predicting the customer-churn tendency and reason in future and providing corresponding services will become the key to reduce churn rate of customer.In our country, the telecom companies develop, strengthen continuously through many years, which have owned many mature databases application system and produce a flood of business handles data., and possessed the substance base to analyze the relation of the customer. But facing a great deal of data, how to analyze and predict customer churn tendency, the characteristic of churn customer and how to service for customer to detain them .The traditional the method of analyzing market and customers' behavior that based on manual management looks helpless. it must make use of the advanced data mining technique and study technique of machine, to heighten the ability of forecast. The technology of data mining, the ability of the mature data mining arithmetic to deal with lots of data and the development of machine study technique of AI research domain, which provide the necessary support of technique and theory for the character analysis of customer-churn behavior in the wireless telecommunication industry.This paper's purpose is to improve the efficiency of data mining, it has advanced data mining method based on rough set and artificial neural network. It through heredity reduction arithmetic selects the import space and training data, adopts artificial neuralnetwork excavation data that was reduced. This method exerts the ability of rough set's reduction knowledge and the high precision characteristic of Artificial Neural Network, it gains very good result to apply this method to characteristic analysis of motion communication's customer churn.Based on above, this paper puts forward to set up the forecast system basic frame of mobile communications joining together the oneself characteristics of the forecast system, we establishing the forecast model of customer-churn behavior and putting forward the solution scheme of mobile communications customer churn, and application actual customer's data to evaluate capability of forecast model. The result of evaluation indicated that the predicting model is scientific and accordant with the fact. The predicting model is significative to solve the question of predicting customer-churn behavior in the wireless telecommunications industry.