A Multi-Classifier Ensemble Algorithm for Predicting Travelers Repurchases Based on Evidence Theory

Repurchase predictions represent a multi-attribute decision problem and an important aspect of marketing strategy. Owing to the high loss and broad base of consumers, this process generates various types of uncertainty, such as unclear, incomplete, and imprecise data, especially in the airline industry. However, existing approaches cannot effectively address these uncertainties. This work applies evidence theory and common machine learning algorithms to propose a multi-classifier ensemble algorithm for predicting consumer repurchase behavior. The framework was trained and validated using 29 selected features from the 2019 to predict travelers repurchase behavior in a marketing activity of one airline in China. Besides, the public datasets were also used for an illustrative example to validate the effectiveness of the developed framework. The results suggested that the ensemble framework described in this report outperforms traditional prediction models in terms of overall predictive performance.