Efficient Revenue Management: Classification Model for Hotel Booking Cancellation Prediction

Abstract

One of the major problems in hotel business management is related to the problem of cancellation of hotel bookings, which implies causing losses of considerable revenues and disruptions of operational activities. Minimizing cancellation prediction errors can assist hotels in setting proper price models and effectively utilizing resources. The purpose of this work is to improve the current classification model for the identification of hotel booking cancellations. Failure to forecast the number of cancellations is another problem because the hotel has to lay down more inventory and loses money in the process. The data collected from kaggle involved variables like booking lead time, customer characteristics, or booking tendencies. Those include data pre-processing functions, in particular min-max normalization algorithms. In this study, the Linear Discriminant Analysis (LDA) method was used in feature extraction to classify booking cancellations. The performance of the Osprey Optimization Fine-Tuned Random Forest (O-FRF) model was assessed using different statistical measures like accuracy (92.12%), precision (88.54%), recall (91.36%), and F1-score (90.67%). The developed classification model could be used by hotel revenues as a valuable tool since it gives accurate probabilities of booking cancellation