Databricks Machine Learning Associate Question 77
Select 3You are working on a machine learning project in Databricks and want to create a feature store table to store processed features for your model. Which of the following steps are necessary to successfully create and populate a feature store table?
- A
Define a Spark DataFrame containing the features you want to store.
- B
Specify a unique primary key for the feature store table.
- C
Write the feature store table using the 'spark.write' method.
- D
Use the 'FeatureStoreClient.create_table()' method to create the feature store table.
- E
Register the feature store table with a model using 'mlflow.log_model()'.
Show answer and explanation
Correct answers: A, B, D
Explanation
To create a feature store table in Databricks, you need to define a Spark DataFrame containing the features, specify a primary key to uniquely identify rows, and use the FeatureStoreClient's 'create_table()' method to create the table. The 'spark.write' method is not applicable for feature store operations, and registering the table with a model is a separate step that is not required for table creation.
- A. Correct.
Correct: A Spark DataFrame is required to define the features to be stored in the feature store.
- B. Correct.
Correct: A unique primary key is required to uniquely identify rows in the feature store table.
- C. Incorrect.
Incorrect: The 'spark.write' method cannot be used to create a feature store table. The Feature Store API must be used.
- D. Correct.
Correct: The FeatureStoreClient's 'create_table()' method is used to define and create the feature store table.
- E. Incorrect.
Incorrect: While registering a feature store table with a model is a common step in MLOps workflows, it is not required for creating the feature store table itself.