Question 10
Domain 1: Databricks Machine LearningWhat is a practical benefit of `write_table()` using primary keys when updating a feature table?
Correct answer: A
Explanation
`write_table()` can use primary keys so updates target existing entities instead of rewriting the whole table. That matches the feature-store idea of writing data in an “entity-aware” way, so you avoid manually rebuilding the workflow for each update.
Why each option is right or wrong
A. It lets feature rows be written in an entity-aware way instead of recreating the whole workflow manually
In Databricks feature tables, `write_table()` can use the table’s primary key to match incoming records to existing entities, so the update is applied at the row/entity level rather than forcing a full table rebuild. That aligns with the feature-store workflow described in the exam guide for writing data to a feature store table in Unity Catalog, where the practical advantage is avoiding manual re-creation of the pipeline each time the feature values change.
B. It converts the feature table into a model endpoint
Model endpoints serve deployed models for inference, not feature-table storage.
C. It removes the need for feature definitions
Feature definitions still describe features; writing data does not eliminate them.
D. It forces all models to retrain immediately
Model retraining is separate from feature-table updates and is not forced immediately.