Databricks Machine Learning Associate Question 428
Select 4You are building a machine learning pipeline in Databricks and need to create a feature store table in Unity Catalog to manage features across multiple teams. Which of the following steps are necessary to successfully create a feature store table in Unity Catalog?
- A
Enable Unity Catalog for your workspace and configure the appropriate metastore.
- B
Use the 'create_table' method from the Feature Store Python API, specifying Unity Catalog as the catalog.
- C
Use the 'fs.create_table()' method and set the 'path' parameter to a location in DBFS.
- D
Ensure that the feature store table is created in a Delta table format.
- E
Assign the required permissions to users or groups at the catalog, schema, or table level in Unity Catalog.
Show answer and explanation
Correct answers: A, B, D, E
Explanation
To create a feature store table in Unity Catalog, you must ensure Unity Catalog is enabled and configure the metastore. The Feature Store Python API allows you to create tables in Unity Catalog without relying on file paths, as Unity Catalog manages metadata and permissions. Feature store tables are stored as Delta tables to ensure reliability and scalability. Proper permissions must also be assigned to control access to the tables.
- A. Correct.
Correct: Unity Catalog must be enabled and properly configured for your workspace to use its features, including the feature store.
- B. Correct.
Correct: The 'create_table' method from the Feature Store Python API facilitates creating feature tables and supports Unity Catalog integration.
- C. Incorrect.
Incorrect: The 'fs.create_table()' method does not use the 'path' parameter for Unity Catalog. Instead, tables are managed directly in the catalog, not as raw files in DBFS.
- D. Correct.
Correct: Feature store tables in Databricks are stored as Delta tables, which provide support for versioning and ACID transactions.
- E. Correct.
Correct: Unity Catalog requires proper permissions to be assigned at the catalog, schema, or table level to manage access to feature store tables.