A team monitors a model that predicts customer lifetime value. They use a Lakehouse Monitor with a baseline set to the training data from January. In July, the drift metrics table shows a PSI of 0.3 for the 'purchase_frequency' feature, which exceeds the alert threshold of 0.25. However, the team knows that a major marketing campaign in June changed customer behavior. What is the most appropriate action?

Change the baseline to the June data to eliminate the alert.

Ignore the alert because the drift is expected.

Retrain the model with recent data that reflects the new customer behavior.

Increase the alert threshold to 0.4 to avoid future alerts.

A team monitors a model with a Lakehouse Monitor. They want to alert when either the PSI for 'age' exceeds 0.3 OR the endpoint error rate exceeds 5%. They also want to notify different teams: the data science team for drift and the DevOps team for endpoint issues. What is the best way to configure this?

Create a single alert rule for the PSI drift and rely on manual checks for the endpoint error rate.

Create a single alert rule for the endpoint error rate and rely on manual checks for the PSI drift.

Create two separate alert rules: one for the PSI drift metric and one for the endpoint error rate, each with its own notification channel.

Create a single alert rule that combines both conditions and notifies both teams.

An ML engineer needs to create a monitor on a time series table that tracks daily sales data. They want to include a custom metric that computes the 7-day rolling average of sales. In the Lakehouse Monitoring UI, where should they define this metric?

In the 'Baseline' section of the monitor configuration

In the 'Metrics' section of the monitor configuration

In the 'Schedules' section of the monitor configuration

In the 'Alerts' section of the monitor configuration

DatabricksCertified Machine Learning Professional

Domain 2Objective 5

Drift Detection and Lakehouse Monitoring MACHINE-LEARNING-PROFESSIONAL Practice Questions (Page 3)

Part of the ML Ops domain, which accounts for 44% of the MACHINE-LEARNING-PROFESSIONAL exam. Databricks does not publish an official question count, but from its 120-minute exam (~50–80 total, ~22–35 in this domain), expect 4–7 from this objective — we provide 34 practice questions to prepare you well beyond it. (estimate)

34questions here

7free pages

10concepts

44%of the exam

Questions 11–15

11expert · hard
A team monitors a model that predicts customer lifetime value. They use a Lakehouse Monitor with a baseline set to the training data from January. In July, the drift metrics table shows a PSI of 0.3 for the 'purchase_frequency' feature, which exceeds the alert threshold of 0.25. However, the team knows that a major marketing campaign in June changed customer behavior. What is the most appropriate action?
Select an answer first
12foundation · easy
In a typical ML monitoring pipeline, which component is responsible for capturing model inputs, predictions, and ground truth for later analysis?
Select an answer first
13foundation · easy
Using an inference table, which metric would you compute to evaluate the proportion of correct predictions among all predictions?
Select an answer first
14expert · hard
A team monitors a model with a Lakehouse Monitor. They want to alert when either the PSI for 'age' exceeds 0.3 OR the endpoint error rate exceeds 5%. They also want to notify different teams: the data science team for drift and the DevOps team for endpoint issues. What is the best way to configure this?
Select an answer first
15application · medium
An ML engineer needs to create a monitor on a time series table that tracks daily sales data. They want to include a custom metric that computes the 7-day rolling average of sales. In the Lakehouse Monitoring UI, where should they define this metric?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to MACHINE-LEARNING-PROFESSIONAL

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Databricks. “MACHINE-LEARNING-PROFESSIONAL” is a trademark of its owner, used for identification only.

Drift Detection and Lakehouse Monitoring MACHINE-LEARNING-PROFESSIONAL Practice Questions (Page 3)

Questions 11–15

In a typical ML monitoring pipeline, which component is responsible for capturing model inputs, predictions, and ground truth for later analysis?

Using an inference table, which metric would you compute to evaluate the proportion of correct predictions among all predictions?

An ML engineer needs to create a monitor on a time series table that tracks daily sales data. They want to include a custom metric that computes the 7-day rolling average of sales. In the Lakehouse Monitoring UI, where should they define this metric?

Keep studying — ML Ops