Applications of minimax regret to inventory and revenue management

Guillaume Roels
University of California, Los Angeles (UCLA)

The minimax regret objective overcomes the conservativeness of the maximin criterion, while relying on the same level of information. In this talk we apply the minimax regret criterion to two celebrated problems of operations management, namely the newsvendor model and network revenue management. Our approach helps evaluate the value of additional demand information (mean, mode, variance, etc.) in inventory management and leads to the development of high-performing heuristics in revenue management.

Back to Workshop IV: Robust Optimization