Home / ML Associate practice test / Databricks Machine Learning / Question

Databricks Machine Learning Associate · free practice question

A batch inference job uses a model packaged with two feature lookups: account_creation_date and num_lifetime_purchases. The scoring DataFrame provides…

Exam-style question on Databricks Machine Learning from the free ML Associate practice test, with the answer and a full explanation below.

Updated November 10, 2026

Databricks Machine Learning

A batch inference job uses a model packaged with two feature lookups: account_creation_date and num_lifetime_purchases. The scoring DataFrame provides account_creation_date explicitly for a what-if analysis. How does Databricks handle this during score_batch?

  1. ADatabricks ignores all provided columns that overlap with packaged features and always looks up both features from the feature table so scoring results remain identical to training-time retrieval logic
  2. BDatabricks fails the scoring request because packaged feature models require either all feature values to be supplied manually or none of them to be supplied in the scoring DataFrame
  3. CDatabricks uses the provided account_creation_date values from the scoring DataFrame and looks up only the remaining packaged feature values that were not supplied
  4. DDatabricks rewrites the packaged model metadata so future runs stop retrieving account_creation_date from the feature store permanently after the first custom scoring request is completed
  5. EDatabricks converts the custom provided feature column into a model tag and still returns predictions based only on the original feature table values, because custom feature overrides are not supported
Answer & explanation

Correct answer: C

WHY C: Databricks supports custom feature values during score_batch; supplied columns override lookup for those features, while missing packaged features are still retrieved. WHY NOT A: Overlapping supplied feature columns are not ignored. WHY NOT B: Partial overrides are supported. WHY NOT D: One scoring request does not rewrite model metadata. WHY NOT E: The provided values are used for inference, not merely logged as tags.

Take the full ML Associate practice test free →

Would you pass today?

This question is one of 45 in the free, timed Databricks Machine Learning Associate practice test. Take it and get a section-by-section score.

Start the free test →