关键词: 贝叶斯分类； 电信数据, 增量学习； 客户流失； 预测

Abstract: With the increasing competition of telecom market, customer churn became one of the focused problems. Because the telecommunication data is huge and has the characteristic of time series, this paper proposes an improved Bayesian classification to study the customer churn problem. The improved Bayesian classification model is designed to make up for the shortcomings of the former Bayes which assumed that each attribute has the same effect on the classification results. Furthermore, by coping with the increasing data, this paper explores the incremental learning method to improve the accuracy of the classifier. The experimental results show that the proposed method has higher accuracy.

Key words: Bayesian classification； telecommunication data； incremental learning method； customer churn, prediction