Exploring Call Detail Records for Churn Analysis in a Telecommunication Company

• Main Authors: Kun-Hong Xue, 薛焜鴻
• Other Authors: Wei-Guang Teng
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/62813143660033398193

碩士 === 國立成功大學 === 工程科學系專班 === 96 === As the market of mobile communication gradually gets saturated in recent years, the competition among telecommunication operators also becomes severe. While it is difficult to effectively attract new subscribers, to reduce customer churn has become a critical issue. In this work, we utilize techniques of social network analysis and graph theory to explore the call detail records so as to understand more about user behavior in their communities. Moreover, we propose in this work an algorithm to discover core members in their respective communities. It can be easily noticed that core members are with significant influence power and network values. More customer churns may happen as core members transfer to another telecommunication operator. Thus, the possibility of customer churn can be reduced by satisfying core members. Through empirical studies, our approach is not only of solid theoretical basis but also feasible in real telecommunication environment. Consequently, our approach can provide telecommunication operators a valuable reference to identify and to reduce customer churn in early stages.