Modeling of Potential Customers Identification Based on Correlation Analysis and Decision Tree

Abstract

The precise identification for potential customers who will subscribe mobile data services steadily is becoming the attention of mobile communications corporations in near year. With the application of data mine techniques, this paper introduces a model which will help mobile communications corporations to find potential customers who will subscribe Multimedia Messaging Service (MMS, as one of mobile data services, is a standard way to send messages that include multimedia content to and from mobile phones.) .

In this paper, the modeling process will be displayed. Firstly, with the using of histogram technique, data preprocessing which is an important part of modeling is described in detail. Then, the application of correlation analysis and decision tree to find rules for identifying potential and stable customers will help us to generate the final model. In the conclusion, a suggestion on future developments to efficiently improve the model will be given.

The result of validation for this model show a high accuracy and achieves the standard for practical business application. Based on targets mined by the model, a mobile communications corporation has launched a series of precise marketing and achieve significant economic benefits in 2010.

Project Supported by Natural Science Foundation of China (NSFC, NO. 60863005, NO. 61011130038), the Government Foundation of Guizhou Province of China (NO.200802).