Applying the CG-logistic Regression Method to Predict the Customer Churn Problem

The aim of this paper is to study a new prediction method for the churn problem for airlines. For this end, a conjugate gradient logistic regression (CG-logistic regression) model is built, which integrates a conjugate gradient method with a logistic regression model. First, we classify churn and non-churn customers utilize the logistic regression model to, and then the conjugate gradient iterative method is applied as the optimum solution using the logistic regression model. At last, we present the outcomes of a simulative assessment and prove that the presented method is conducive to analyzing the churn problem for airlines.

I. Introduction

Large data are broadly applied in miscellaneous domains including communication, finance and education. For many years, the growth of the Internet has ushered big data into the explosive growth stage. Big data are data with scale, diversity and timeliness, and new analytical methods are required to effectively extract commercial value from these new resources.