A study on a predictive model of customer defection in a hotel reservation website

Abstract

This paper examines a hotel reservation website's customer defection. Applying statistic and data mining technology including logistic regression and random forests, we examine customer database to identify the attributes that affect customer attrition and develop a model of customer defection in the hotel reservation website. The empirical evaluation results showed the model has 78.9% accuracy, which suggest that the proposed churn prediction technique exhibits satisfactory predictive effectiveness.