< Terug naar vorige pagina

Publicatie

Gaussian processes for daily demand prediction in the hospitality industry

Boekbijdrage - Boekabstract Conferentiebijdrage

In tourism forecasting literature, the majority of the studies relies on econometric models. While econometric models are effective for monthly or yearly data, many machine learning techniques are yet to be explored in the demand literature and it can be beneficial to do so on daily data. This study fills the research gap by predicting daily hotel occupancy rates on a city level in Brussels using booking.com data. A forecasting methodology is proposed with feature extraction and selection, processing steps that are missing in most approaches. Special attention is given to Gaussian Processes to make predictions with credible intervals, an uncertainty measure that is helpful for managers during decision-making. The novel approach uses online data to create a proxy measure for occupancy rate. The new methodology is verified with internal occupancy data from hotels in Brussels and can be extrapolated to cities for which no internal hotel data is available.
Boek: EURO 2018 : meeting abstracts
Aantal pagina's: 1
Jaar van publicatie:2018
Toegankelijkheid:Closed