Dynamic pricing and inventory control with nonparametric demand learning

This paper studies joint dynamic pricing and inventory planning with demand learning. Demand is assumed to be a function of price with an uncertain price-sensitivity parameter. We introduce a nonparametric functional-coefficient autoregressive (FAR) state-space model without assumptions on the parametric structure and apply a Bayesian method using Markov chain Monte Carlo (MCMC) algorithms to estimate model parameters. We develop an optimal control model and obtain optimal pricing and inventory plan based on the estimated parameters. We use numerical computations with single and dynamic replenishment policies to evaluate the proposed demand learning algorithm and optimal control based methods and demonstrate the importance of dynamic pricing, inventory control, and demand learning.