Dynamic pricing policies for interdependent perishable products or services using reinforcement learning

• Dynamic prices maximize the expected revenue of interdependent products.
• Reinforcement learning optimizes the pricing of interdependent products.
• Interdependent pricing enhances learning.

Many businesses offer multiple products or services that are interdependent, in which the demand for one is often affected by the prices of others. This article considers a revenue management problem of multiple interdependent products, in which dynamically adjusted over a finite sales horizon to maximize expected revenue, given an initial inventory for each product. The main contribution of this article is to use reinforcement learning to model the optimal pricing of perishable interdependent products when demand is stochastic and its functional form unknown. We show that reinforcement learning can be used to price interdependent products. Moreover, we analyze the performance of the Q-learning with eligibility traces algorithm under different conditions. We illustrate our analysis with the pricing of services.