Implementation of Ensemble Machine Learning Techniques on Hotel Reservation System

Abstract

The hotel industry effectively supports the wider national economy. The sector supports the nation's efforts to promote cultural events and improve visitor amenities. The primary issue this sector is currently dealing with is a hotel booking cancellation. Because of the impact this issue has on the hotel industry's resource allocation, labour needs, customers’ satisfaction, and overall decision-making process. In result, the hotel's reputation, business processes, and financial performance may also suffer. The major goal of this research is to create a model that will help the hotel industry make wise decisions. We completed the necessary data preprocessing and transformation procedures using the Kaggle hotel bookings dataset. Furthermore, we employed a number of machine learning methods to predict the cancellation requests in the future. Additionally, this study used a number of ensemble techniques, including voting, stacking, and bagging, to improve the model's accuracy. The findings showed that the stacking strategy outperformed all other models and had an accuracy rate of 86.76%. The proposed model and analysis discussed in this paper may help the hotel sector forecast the kinds of requests that will likely be cancelled in the future. Keywords—Hotel Booking Cancellation, Machine Learning, Ensemble Machine Learning, Classification