Databricks Machine Learning Professional Question 134
Select 4You are tasked with deploying a machine learning model using batch deployment in Databricks. The model needs to process input data stored in a Delta table, generate predictions, and save the results to another Delta table for later use. Which of the following steps correctly describe the batch deployment process?
- A
Read the input data from the source Delta table using Spark DataFrame.
- B
Deploy the model as a REST API endpoint for generating predictions.
- C
Apply the model to the input data using a UDF or pandas_udf for batch inference.
- D
Write the prediction results back to the target Delta table.
- E
Schedule the batch deployment using a Databricks Job.
Show answer and explanation
Correct answers: A, C, D, E
Explanation
In a batch deployment scenario, the process involves reading input data from a Delta table, applying the model to generate predictions, and writing the output back to a Delta table for later use. To automate this process, it is common to schedule the workflow using Databricks Jobs. REST API endpoints, on the other hand, are used for real-time inference and are not applicable in this context.
- A. Correct.
Correct. Reading the input data from the source Delta table is the first step in the batch deployment process to prepare data for inference.
- B. Incorrect.
Incorrect. REST API endpoints are typically used for real-time inference, not batch deployment.
- C. Correct.
Correct. Applying the model to the input data using a UDF or pandas_udf allows the model to generate predictions in a batch processing manner.
- D. Correct.
Correct. Writing the prediction results back to the target Delta table ensures that the predictions are stored for later use.
- E. Correct.
Correct. Scheduling the batch deployment using a Databricks Job automates the process to run at specified intervals or triggers.