Bayesian dynamic operating rules for optimal hotel reservations

During the time prior to rent of rooms in a hotel, decisions have to be made concernin the allocation of the rooms to customers requesting single-bed or double-bed rooms. The paper presents a model for finding the maximum allowable number of bookings of customers requesting single and double-bed rooms at each time prior to rent such that a maximum expected contribution to profit for the rental day is obtained. The model utilizes the time distribution of reservations and cancellations for both room sizes. Consideration is given to the effects of late unbooked arrivals, waitlists, standbys and overbookings. A Bayesian reassessment of probabilities is incorporated in a sequential decision procedure. The model is extendable to handle requests for periods longer than a single night, for group requests of various room types, and for different combinations of them. A numerical example is provided.