Robust Dynamic Pricing of Perishable Products

We are going to study dynamic pricing for perishable products when demand is uncertain and the underlying probabilities are not known precisely. In our study, we consider a linear price-response function with additive uncertainty. With this additive uncertainty assumption, we assign distributions to both market size and price elasticity with some means and standard deviations, and explore the optimal pricing for the products under different levels of uncertainty embedded in demands with the above parameters. To solve such a semi-infinite optimization problem, we use RISKOptimizer to deploy a genetic algorithm. Furthermore, since the pricing in real practice usually is bound to some constraints regarding nominal values while the optima from RISKOptimizer may slightly differ from the nominal values, we also use @RISK to simulate the resulting objectives corresponding to the nominal pricing values and analyze the impact on decision making. In addition to the discussion of the above robust optimization modeling, we also survey the control parameters of the genetic algorithm to be tuned for efficiency.