問題文
A prescriptive story suggests changes, but some suggestions involve variables the business cannot alter (such as a customer's age). How should the model be configured to keep recommendations useful?
選択肢
- Rely on the security predicate to hide the unchangeable suggestions.
- Delete all rows containing unchangeable variables so they cannot be recommended. The story would be trained on a narrower population than the one it is meant to score.
- Increase the number of sensitive variables that the business cannot alter so the model ignores this customer's age.
- Mark only the variables that can actually be changed as actionable so recommendations focus on changeable levers.