Databricks Certified Machine Learning Associate Practice Questions

Databricks Certified Machine Learning Associate exam preparation

Databricks Certified Machine Learning Associate practice questions in English: 170 questions with explanations. The mock exam has 48 questions and 90 minutes.

Databricks Certified Machine Learning Associate free question 2 / 10 · Domain 1: Databricks Machine Learning

A team wants to add automatic retraining to a model that already runs in production. Which pair of statements correctly describes where the Databricks recommended MLOps workflow says to start and how triggered retraining is set up?

  1. Start with scheduled retraining and move to triggered retraining when needed, using a SQL alert with an email destination whose message itself starts the training workflow
  2. Start with scheduled retraining and keep it as the sole mechanism, because the architecture does not allow alerts to start the training workflow
  3. Start with scheduled retraining and move to triggered retraining when needed, using a SQL alert with a webhook destination that starts the training workflow
  4. Start with triggered retraining based on monitoring thresholds, and switch to scheduled retraining once new data arrives on a regular basis
Show the answer and explanation

Answer: C

Why this is correct
The workflow supports automatic retraining that reuses the model training pipeline. Databricks recommends beginning with scheduled, periodic retraining and moving to triggered retraining when needed. Scheduled retraining applies the training code to the latest data through a scheduled job. Triggered retraining is built from a SQL alert that checks whether a metric, such as drift or model quality, is anomalous against a threshold; the alert is configured with a webhook destination, which can then trigger the training workflow.

Why the other options are wrong
"Start with triggered retraining based on monitoring thresholds, and switch to scheduled retraining once new data arrives on a regular basis" … The order is reversed. The page recommends beginning with scheduled, periodic retraining and moving to triggered retraining when needed, not the other way around. Starting with thresholds means automatic retraining runs before the metrics and thresholds have been worked out.
"Start with scheduled retraining and move to triggered retraining when needed, using a SQL alert with an email destination whose message itself starts the training workflow" … The start is right but the mechanism is not. The page names a webhook destination as the way an alert triggers the training workflow. An email message is a notification to a person and does not start a workflow.
"Start with scheduled retraining and keep it as the sole mechanism, because the architecture does not allow alerts to start the training workflow" … The page says to move to triggered retraining when needed, and describes an alert-based trigger for it. Scheduled retraining is where to begin, not the only option the architecture supports.

Key point
Scheduled retraining is the starting point because it needs no monitoring signal. Triggered retraining adds a SQL alert on a drift or quality metric whose webhook destination starts the training workflow.

Go to the Databricks Certified Machine Learning Associate practice app (10 free questions; purchase unlocks all questions)

See the full practice set