Predicting and visualization analysis of customer churn in telecommunications leveraging decision tree model

Abstract

Nowadays, how to keep more customers is one of the most crucial problems to be solved facing communication companies in the increasingly competitive market. In these companies, one of their most important business indicators is customer churn for it is significantly less expensive to keep a current client than to find a new one. In order to carry out truly effective methods to minimize customer loss, build a prediction model is a reasonable way to analyze the characteristics and reasons of possible lost customers. In this research, after creating a Decision Tree Model and visualize it, it could be concluded that there was little difference between male and female, but the aged, unmarried and customers who are economic dependent have a great possibility to lose. Moreover, signing duration matters in the prediction. The longer the contract term is, the less likely the users to be lost. According to mentioned and even more analysis below, certain targeted suggestions are introduced to improve the situation. This research would like to provide research thought and analytical method for the prediction of customer churn.