Summary

In this chapter, we covered two important classes of problems in supply chain: Inventory optimization and vehicle routing. These are both very complex problems, and reinforcement learning has recently emerged as a competitive tool to address them. In the chapter, for the former problem, we provided you with a detailed discussion on how the create the environment and solve the corresponding reinforcement learning problem. The challenge in this problem was the high variance across episodes, which we mitigated through a careful hyperparameter tuning procedure. For the latter problem, we described a realistic case of a gig driver who delivers meal orders that dynamically arrive from customers. We discussed how the model can be made more flexible to work with a varying size of nodes via pointer networks.

In the next chapter, we will discuss yet another exciting set of applications around personalization, marketing, and finance. See you there!