A variant bagging forecasting framework for customer churn in airline

Abstract

The goal of this study is to forecast customer churn and analyze the influence of quality of service on customer churn in airline industry. Following a multifactor approach, a variant Bagging forecasting framework is proposed to mine the inner patterns of customer churn. A probabilistic sampling approach is embedded in the developed model simulating the customer churn probabilities. The airline customer feedback data by airline carriers in the U.S. was used to train the prediction model. The results indicate the accuracy for predicting customer churn is 96 %, and the most important factors for customer churn are in-flight entertainment, seat comfort, and type of travel. We also investigated the effects of service quality (both hedonistic and utilitarian factors) on various customer groups and discovered that improving hedonistic service quality can effectively reduce customer churn.