Question 1
Domain 1: Databricks Machine LearningA machine learning engineer wants to build a training dataset using managed features rather than manually assembling raw columns. Which Databricks capability is used to train a model with features from a feature store table?
Correct answer: B
Explanation
When training with managed features in Databricks, the feature store is the capability used to supply feature store table features to the model training workflow. — official.txt
Why each option is right or wrong
A. Use Delta Live Tables to replace the feature store during model training.
Delta Live Tables is not identified here as the capability for training with feature store table features.
B. Use the feature store to train the model with features from a feature store table.
The source material explicitly states that a model is trained with features from a feature store table, so the feature store is the Databricks capability used in this training workflow.
C. Use model serving to create the training features before fitting the model.
Model serving is not described here as the mechanism for training with feature store table features.
D. Use Auto Loader to attach feature store values directly during model fitting.
Auto Loader is not named here as the capability for training a model with features from a feature store table.