关键词: 电信资费预演, 客户量预测, 线性回归分析, 指数平滑法, 决策树算法

Abstract:

With the use of theories and methods of statistics and data mining,this paper researched the prediction method of the amount of the new customers（which take this new network service packages the first time）and the transfer customers（which have ever chosen the original network service packages）.The amount of new customers could be predicted according to time series method based the historical data of a similar package,and the amount of transfer customers could be predicted with the finding to the rules of customer change service packages through data mining。Finally,a case study of a telecom operator was carried out in the paper.Linear regression analysis and exponential smoothing method were used and compared in the establishment of prediction model of amount of the new customers,and decision tree algorithm was used to find the rules of customer change service packages.

Key words: telecom services tariff preview, amount prediction of customers, linear regression analysis, exponential smoothing method, decision tree algorithm