Abstract— In most industries where switching costs are prevalent, the landscape of activities is painted of Customers attrition or churn (Clients who want to switch or change their suppliers for various reasons). This phenomenon is ubiquitous in the telecommunication industry and every aspect related to it, leads to believe that it’s steeply growing. As the market is fiercely competitive, and the number of prepaid customers is increasing, it is vital that companies proactively tackle the defection of their customers by determining behaviors that might ultimately create churn. In this paper we proposed a hybrid learning model to predict churn in mobile telecommunication networks. Experiments were carried out using WEKA a Machine Learning tool; along with a real dataset from an Asian mobile operator to evaluate the performance of the model. The results show that the new hybrid model is more accurate than single methods.

Index Terms— Churn prediction, logistic regression, telecommunications, voted perceptron.

G. D. O. Olle and S. Q. Cai are with the Department of Management Sciences and Information Management, Huazhong University of Science and Technology, 430074-Wuhan, PR–China (e-mail: georgesdelortolleolle@gmail.com, caishuqin@sina.com.cn).

Cite: Georges D. Olle Olle and Shuqin Cai, " A Hybrid Churn Prediction Model in Mobile Telecommunication Industry," International Journal of e-Education, e-Business, e-Management and e-Learning vol. 4, no. 1, pp. 55-62, 2014.