Applying machine learning in forecasting hotel booking cancelations

Abstract

Demand forecasting is vital to the tourism and hospitality industry because it helps the hotels' revenue management process to be controlled more effectively. However, Booking Cancellations usually significantly affect the accuracy of demand forecast. Based on data from 119,390 booking observations from July 1, 2015, to August 31, 2017, of a resort hotel and a city hotel in the Algarve and Lisbon, respectively; the aim of this study is to two apply Machine Learning methods-Logistic regression and XGBoost to predict the booking cancelations of these two hotels. The result of the model showed that the XGBoost performed better with an accuracy of 90,96 % compared with that of Logistics Regression is only 86,07 %. Moreover, the model also shows that the most important feature in deciding whether the bookings will be canceled is the Lead Time.