An engineer wants to register a trained model version in Unity Catalog using MLflow APIs from a notebook. The workspace is Unity Catalog enabled. Which step is required when the client is not already defaulting to Unity Catalog?
Answer & explanation
Correct answer: B
WHY B: When UC is not already the default, MLflow must be configured to use databricks-uc, and the registered model name must use the full <catalog>.<schema>.<model> format.
WHY NOT A: A local backend is not required; the tracking URI is not what selects the Unity Catalog registry.
WHY NOT C: S3-only artifact storage is not the registration prerequisite; runs with artifacts in MLflow-managed storage or Unity Catalog volumes can be registered.
WHY NOT D: The workspace registry is not a required intermediate step, and renaming does not move a model into Unity Catalog.
WHY NOT E: score_batch is unrelated to registration prerequisites.