深度生存分析在电信客户流失预测中的应用

Application of Deep Survival Analysis in Telecom Customer Churn Prediction

目前电信行业市场已逐渐趋于饱满,增加新客户极其不易,控制客户流失成为了目前三大运营商最需要解决的问题之一。文章主要基于生存分析与深度学习理论,证明深度学习在生存分析中的应用在预测风险方面表现优于或优于其他生存方法。通过Deep Surv模型,使该模型与基本Cox回归模型在预测后的C指数值进行比较,结果表明加入了深度学习的生存分析相比传统的生存分析模型对电信领域的预测效果更好。

At present,the telecom industry market has gradually become full,and it is not easy to increase new customers.Controlling customer loss has become one of the most important problems to be solved by the three major operators.Based on survival analysis and deep learning theory,this paper proves that the application of deep learning in survival analysis performs better than other survival methods in predicting risk.The Deep Surv model was used to compare the predicted C-index values between the model and the basic Cox regression model.The results show that the survival analysis with Deep learning has a better prediction effect than the traditional survival analysis model in the telecommunications field.