Predictive models of hotel booking cancellation: a semi-automated analysis of the literature

Abstract

This study sought to combine data science tools and capabilities with human judgement and interpretation in order to demonstrate how semiautomatic analysis of the literature can contribute to identifying and synthesising research findings and topics about booking cancellation forecasting. The study also focused on recording in detail the analysis’s full experimental procedure to encourage other researchers to conduct automated literature reviews in order to understand more fully the current tendencies in their field of study. The data were obtained through a keyword search in Scopus and Web of Science databases. The methodology presented not only diminishes human bias but also enhances data visualisation and text mining techniques’ ability to facilitate abstraction, expedite analysis and improve literature reviews. The results show that, despite the importance of forecasting booking cancellations to understanding net demand and improving cancellation and overbooking policies, further research on this subject is needed.