Artificial Intelligence Applied to Address Tourism Challenges: Predicting Hotel Room Cancellations

The contemporary era has witnessed a substantial surge in the application of artificial intelligence across diverse domains of research. Notably, the integration of machine learning techniques has garnered considerable attention in the realm of forecasting economic phenomena. Of particular interest in recent times is the predictive modeling of hotel room cancellations, an issue for which conventional research methodologies have proven inadequate. The intricacies of this problem demand sophisticated predictive capabilities that are best addressed through the deployment of artificial intelligence.This study is centered on the utilization of continuous learning algorithms with the primary objective of harnessing existing datasets while accommodating evolving prediction requirements. The methodology is designed to adapt progressively to the distinct characteristics of Vietnamese data, thereby ensuring robust predictive performance. The core innovation in this research lies in the amalgamation of the sliding window approach within the framework of continuous learning, coupled with the selection of classical machine learning algorithms in artificial intelligence. This amalgamation transforms the selected classical algorithms into continuous learning models, custom-tailored to the specific demands of this economic phenomena. The study's findings are systematically juxtaposed to discern the optimal algorithm for forecasting hotel room cancellations—an issue of substantial economic significance, especially within the context of Vietnam. To facilitate both economic researchers' experimentation and the practical implementation of this methodology within hotel establishments, a dedicated website has been established. This platform serves as a valuable resource for evaluating the real-world utility of the proposed approach. Research purpose: The significance of this matter within the Vietnamese context is profound. Tourism plays a pivotal role in Vietnam's economic landscape, contributing significantly to the nation's Gross Domestic Product (GDP). Addressing the challenge of booking cancellations is of paramount importance, as it offers numerous advantages for both travelers and hoteliers. Effectively addressing this issue fosters trust among visitors and ensures the sustainable growth of the tourism sector in the long term, particularly in Vietnam. Research motivation: The issue of hotel booking cancellations presents a formidable challenge with far-reaching implications for both the global hospitality industry and travelers. Beyond its financial ramifications for hotels, cancellations significantly impact the quality of travelers' experiences. This challenge has been exacerbated during the Covid-19 pandemic and also after that. due to the heightened uncertainty stemming from the unpredictable nature of the outbreak and the subsequent implementation of social distancing policies. In the specific context of Vietnam, booking cancellations have emerged as a prominent issue. Research design, approach, and method: This study is centered on an emerging research area that aligns with an innovative algorithmic framework. Its primary aim is to contribute to solving the challenge of hotel room cancellations. To achieve this objective, the study relies on a meticulously curated database and is focused on identifying a suitable method within this dataset to serve as the foundational algorithm for a classification system. The ultimate goal is to provide organizations with the necessary tools, research insights, and practical outcomes to effectively address the issue of room cancellations through the utilization of an artificial intelligence system. This, in turn, will enhance the reliability of room reservations for both domestic and international establishments, not only in Vietnam but also in other regions. The research endeavor seeks to leverage artificial intelligence systems to predict customer room booking needs. This approach aims to improve businesses' understanding of tourist requirements, ultimately leading to increased revenue and fostering economic progress, both within Vietnam and on a global scale. Main findings: The primary objective of this investigation is to conduct a comparative analysis of significant machine learning classifiers within the framework of evolving learning methodologies, contrasting them with conventional static machine learning approaches observed in prior research. This decision arises from the imperative need to embrace a continuous machine learning framework as opposed to a static model. This necessity arises due to the sustained utilization of the hotel cancellations dataset for future research endeavors pertaining to hotel cancellations in Vietnam. We embark on an exploration of a diverse set of six distinct machine learning algorithms in artificial intelligence (MLPClassifier, KneighborsClassifier, Linear Discriminant Analysis, BernoulliNB Classifier, Decision Tree Classifier and GaussianNB Classifier), enhanced by the incorporation of progressive strategies inspired by the Klinkenberg concept, to dynamically determine the optimal window size. In summary, the DecisionTreeClassifier algorithm appears to be highly suitable for the hotel room reservation prediction problem and practical applications. Practical/managerial implications: The research introduces a demonstration system designed to familiarize researchers with the practical application of the developed algorithms and systems. This topic is poised for further development in the future, which may involve updating the dataset through survey methods and incorporating specialized knowledge to acquire the most standardized and closely aligned dataset with reality. The system's functionalities have been thoroughly implemented, effectively meeting the initial requirements. It is designed with separate sections for general users and developers, offering a user-friendly interface for predicting hotel reservations while providing developers with additional pages for system enhancement and development.