Databricks Machine Learning Associate Question 79
Select 3You are working on a machine learning project in Databricks and want to create a feature store table to store and serve features for a customer churn prediction model. Which of the following steps are necessary to create a feature store table in Databricks?
- A
Define a DataFrame containing the features you want to store.
- B
Use the FeatureStoreClient to create or update the feature table.
- C
Register the feature table with a Databricks Unity Catalog before using it.
- D
Specify a primary key when creating the feature table.
- E
Manually write the feature data into a Delta table outside the feature store.
Show answer and explanation
Correct answers: A, B, D
Explanation
Creating a feature store table in Databricks involves defining a DataFrame with the desired features, using the FeatureStoreClient to create or update the table, and ensuring a primary key is specified for identifying records. These steps are essential for leveraging the feature store's functionality, such as versioning, lineage tracking, and serving features for machine learning models.
- A. Correct.
This is correct because creating a feature store table starts with defining a DataFrame that contains the features you want to store.
- B. Correct.
This is correct because the FeatureStoreClient is required to create or update feature store tables in Databricks.
- C. Incorrect.
This is incorrect because registering a feature table with Unity Catalog is not a mandatory step for creating a feature store table in Databricks. However, Unity Catalog can be used for governance purposes.
- D. Correct.
This is correct because specifying a primary key is essential to uniquely identify records in the feature store table.
- E. Incorrect.
This is incorrect because writing feature data into a Delta table outside the feature store bypasses the feature store's capabilities and does not create a proper feature store table.