A systematic review of machine learning applications in hotel occupancy forecasting

Abstract For Revenue Management (RM) in hotels, demand forecasting is considered to be one of the most important aspects in making operational and strategic decisions. For years, the basis for demand forecasting, hotel occupancy prediction, has depended on traditional techniques. However, the emergence of artificial intelligence (AI) and machine learning (ML) has changed the landscape, improving the methodological rigour of forecasting techniques. Focusing on bibliometric trends, study characteristics, and methodological approaches, this study systematically reviews the extant literature specific to the applications of AI and ML for hotel occupancy forecasting. Our analyses reveal that in recent years, the number of papers using these techniques for hotel occupancy forecasting has increased, largely due to increased prediction accuracy yielded by ML models. This work also assesses the quality of the scientific research in the area, giving recommendations for future forecasting studies, particularly with respect to forecasting methodology and ML model training and development. Following the PRISMA guidelines, this research fills the gap in the existing literature by assessing the quality of the current state of the literature, proposing a checklist of requirements for high-quality research, and giving recommendations for future studies and the development of standards for ML forecasting research in hospitality literature.