A Randomized Linear Programming Method for Network Revenue Management with Product-Specific No-Shows

• DOI: 10.1287/trsc.1110.0386
• Publication Date: 2011-11-05T02:15:04Z
• Authors: Sumit Kunnumkal; Kalyan Talluri; Huseyin Topaloglu
• Keywords: Total revenue

Revenue management practices often include overbooking capacity to account for customers who make reservations but do not show up. In this paper, we consider the network revenue problem with no-shows and overbooking, where show-up probabilities are specific each product. No-show rates differ significantly by product (for instance, itinerary fare combination an airline) as sale restrictions demand characteristics vary However, models that no-show individual difficult handle because state-space in dynamic programming formulations (or variable space approximations) increases significantly. propose a randomized linear program jointly control decisions product-specific no-shows. We establish our formulation gives upper bound on optimal expected total profit, is tighter than deterministic appears existing literature. Furthermore, asymptotically tight regime leg capacities scaled linearly same rate. also describe how can be used obtain bid price policy. Computational experiments indicate approach quite fast, able scale industrial problems, provide significant improvements over standard benchmarks.