Prediction And Analysis Of Customers Loss In Telecommunication Based On Combined Forecasting Technique

As the main component of telecommunication market, telecommunicationenterprise faces a new problem that how to make use of effective tool to discover andstudy a customer to be worth well with consume behavior. Since the competitions inthe telecommunication market become more and more fierce, and together with thetopic size effect in this area, the customers’ saving and hunting sharply catches theoperators’ focuses of attention. In this paper, based on the theory of decision tree andthe methodology of information gain, the author analyzes the method to find out therelationship between the customers’ natural qualities and their consumptioncharacteristics.In this paper, the definition of Data Mining and the usual process of Data Miningare described. And the customers’ information of a Telecommunication Enterprise isanalyzed based on Data Mining, by using Data Mining tools of SQL Server. Theanalysis leads to a valuable result and the result verifies that the technique of DataMining is helpful in customer information analysis. Base on the studies of the datamining technology，this technology was applied it to avoid customer churning intelecommunication．The PHS’s history data were used to establish some customerchurning models．After evaluating these customer churning models, the best one wasdetermined．This paper firstly carries on an overview to the current related research,summaries the current applied research of data mining technologies, then analysis theproblems exists in current applications, which determines the research topics in thispaper. Aimed to solve the problem that many factors result in losing of customer,which is difficult to separate the losing customer with one common classificationmode, we choose some key factors and use decision tree and neural network algorithmto obtain the predictions. For more accurate results, a Combined Forecastingalgorithm is provided. And the precision of losing customer gets better promotion.The results of evaluation show that the prediction model is feasible． Theprediction model helps to predict customer-chum behavior in the telecommunicationindustries．...