Autonomous Airline Revenue Management: A Deep Reinforcement Learning Approach to Seat Inventory Control and Overbooking

Syed Arbab Mohd Shihab, Caleb Logemann, Deepak-George Thomas, Peng Wei

Keywords: AI, RL, Revenue Management

Abstract: A revenue management system is a real-time decision making system that maximizes earnings by controlling inventory and pricing according to consumer behavior prediction on micro-market levels. It is the key piece to generate profits in airline, rail, cruise, hotel and rental car industries. In this paper, the seat inventory control and overbooking problem of airline revenue management has been formulated as a Markov Decision Process and then solved by using Deep Reinforcement Learning to find the optimal policy, one that maximizes the revenue for each flight. Multiple fare classes, concurrent continuous arrival of passengers of different fare classes, overbooking and random cancellations that are independent of class have been considered in the model. To generate data for training the agent, a basic air-travel market simulator was developed. The performance of the agent in different simulated market scenarios was compared against theoretically optimal solutions and was found to be nearly close to the expected optimal revenue. This work is among the first few ones to address real-world airline revenue management by using deep reinforcement learning and it has been invited by American Airlines and Airline Group of the International Federation of Operational Research Societies (AGIFORS) for seminar talks and potential collaborations.