Offline policy evaluation for inventory RL

Advanced planning methods are only useful when they improve a real decision under real constraints. Here, offline policy evaluation for inventory rl is framed as a testable system-design choice.

Practitioner lens

Uncertainty assumptions drive the behavior of offline policy evaluation for inventory rl. Document how scenarios are generated, what dependencies they preserve and which low-probability cases deserve special attention.

First experiment

Document assumptions and version the data, code, parameters and scenario set for reproducibility.

This is an original educational brief. It is a starting point for discussion, not a claim that one technique fits every supply chain.