Customized dynamic pricing for air cargo network via reinforcement learning

Authors

รศ.ดร.กาญจ์นภา อมรัชกุล

Published

Lecture Notes in Computer Science

Abstract

A customized dynamic pricing problem for air cargo network management is proposed and formulated as a Markov decision process. Our model combines a customized pricing model for a B2B setting and a booking control problem for an air cargo network. Our solution approach employs reinforcement learning. In the numerical example based on historical records in 2016–2019 at one of the largest European carriers, the performance of reinforcement learning is evaluated. The policy from the reinforcement learning clearly outperforms the myopic policy and is within 10% of optimal actions in most cases.

(2562). Customized dynamic pricing for air cargo network via reinforcement learning. Lecture Notes in Computer Science, na(na), 213-224.