I don’t think so.
Unit tests don’t tell you if a problem constraint is achieving the outcome you wanted or how the plan is likely to impact your bottom line. They don’t help with stakeholder management.
While I do believe treating decisions as code is important, decision models are unique in other ways that necessitate specific types of testing.
This has certainly been part of conversations I’ve had with algo devs. Are others encountering this as well? Do you see this the same way? How do you work with dev teams to establish better intuition around model performance?