Purposeful underestimation of demands for the airline seat allocation with incomplete information

We study stochastic programming formulations for the origin destination model in airline seat allocation under uncertainty. In particular, we focus on solving the stability issues of the traditional probabilistic model by purposefully underestimating the demands. The stochastic seat allocation models assume at least the possession of the distributional information, which is usually difficult to satisfy in a constantly changing environment. We propose a heuristic that consists of dynamically incorporating available information by solving a sequence of stochastic programming models. We show that the proposed method, named 'seat reservation (SR)', can ease most negative effects of incomplete distributional information and under some restrictive conditions, the SR will yield optimal revenue. The seat reservation method suggests that a revenue management company must (1) obtain timely results using adequately up–to–date computational facilities; (2) be conservative when allocating resources and (3) actively and continually revise previous estimations.