Question 16
Domain 1: Databricks Machine LearningWhat is a practical advantage of model aliases such as `Champion` and `Challenger`?
Correct answer: A
Explanation
Aliases provide a stable name for a model role while the underlying registered version can be swapped after validation. The exam guide explicitly says, "Promote a challenger model to a champion model using aliases," which supports using labels like Champion and Challenger without changing client references.
Why each option is right or wrong
A. They let clients reference a stable label while the underlying version can change after validation
In the Databricks MLflow/Unity Catalog model registry flow, aliases are the mechanism used to point a human-readable name at a registered model version, so clients can keep calling the same label even when the backing version is updated after evaluation. The exam guide explicitly includes the objective “Promote a challenger model to a champion model using aliases,” which implies the alias remains stable while the promoted version changes underneath it.
B. They remove the need to version models at all
Model aliases do not remove versioning; they provide a stable label over versioned models.
C. They guarantee the newest model is always the best
Aliases do not judge model quality; validation still determines whether a challenger becomes champion.
D. They automatically create feature tables
Feature tables are created through feature store workflows, not model aliases.