FORECASTING HOTEL BOOKINGS CANCELLATION WITH A MACHINE LEARNING CLASSIFICATION MODEL

Abstract

In the hotel sector, demand management choices are greatly influenced by cancellations of reservations. With the increasing popularity of online travel agencies and the flexibility they offer customers, cancellations have become common in the hotel industry. The ability to generate accurate projections, a crucial tool for successful revenue administration, is constrained by suspensions. Hotels adopt strict cancellations rules and overcrowding techniques to minimise the issues brought on by cancelled reservations, but these actions can also negatively affect income and reputations. To predict the booking cancellations in the United States hotel dataset classification approach in machine learning can be used as the dependent variable is categorical and the model is developed with a good precision score of approximately 75%. This shows it is feasible to forecast if a booking would be cancelled with high accuracy, contrary to what Morales and Wang (2010) assumed. The outcomes enable hotel management to accurately determine net demand, provide better predictions, enhance cancelling rules, establish better overcrowding processes, and employ more effective marketing and allocation of stock strategies.