Revenue Management under Customer Choice Behaviour with Cancellations and Overbooking

Revenue management is the practice of pricing perishable goods to optimise revenue. A realistic revenue management model allows overbooking and incorporates customer buying behaviour and cancellations. The latter is motivated by our research using real data, which shows that for a hotel a large proportion of the reservations are cancelled. However, existing revenue management models do not take both cancellations and customer choice behaviour into acount. We propose a revenue management model which takes cancellations into account in addition to both overbooking and buying behaviour of customers. This model is based on the customer choice model introduced by Talluri and Van Ryzin. We model the customer choice cancellation model as a Markov decision process and propose a dynamic programming formulation to solve the problem. The state space and action space of the problem are too large, such that the problem is intractable to solve for all practical purposes. Therefore we propose two heuristics, each appropriate in a dierent setting. Numerical results show that that the heuristics perform well and the customer choice model without cancellations does not. Moreover, we provide a parameter estimation method based on Newman et alii. This estimation method is fast and provides good parameter estimates, such that the model can well be applied in practice.