A team is building a monitoring solution for a model that serves predictions via a REST API. They need to track both model performance (e.g., accuracy) and endpoint health (e.g., latency). They have access to an inference table that logs inputs, predictions, and labels, and they also have access to API logs with latency and error rates. What is the most efficient way to set up monitoring?

Use a snapshot table for both performance and endpoint health.

Create a single Lakehouse Monitor on the inference table and include latency metrics as custom metrics.

Create a single Lakehouse Monitor on the API logs and compute accuracy from the logs.

Create a Lakehouse Monitor on the inference table for performance and a separate monitor on the API logs for endpoint health.

In Lakehouse Monitoring, what is required when defining a custom metric?

A Python UDF that runs on each row

A pre-trained machine learning model

A separate Delta table for storing the metric

A SQL expression that computes the metric

A data science team monitors a churn-prediction model. They want to be notified via email when the PSI (Population Stability Index) for the 'age' feature exceeds 0.25. Which configuration steps are required in Lakehouse Monitoring?

Use the built-in KS test and set an alert for p-value below 0.05.

Create a dashboard and manually check the PSI value daily.

Set a custom metric for PSI and configure an alert rule with a threshold of 0.25, then add an email notification.

Set the baseline to the current data and enable the default alert for all metrics.

Which of the following is a required step when constructing a monitor for an inference table?

Selecting the prediction column

DatabricksCertified Machine Learning Professional

Domain 2Objective 5

Drift Detection and Lakehouse Monitoring MACHINE-LEARNING-PROFESSIONAL Practice Questions (Page 6)

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 26–30

26expert · hard
A team is building a monitoring solution for a model that serves predictions via a REST API. They need to track both model performance (e.g., accuracy) and endpoint health (e.g., latency). They have access to an inference table that logs inputs, predictions, and labels, and they also have access to API logs with latency and error rates. What is the most efficient way to set up monitoring?
Select an answer first
27foundation · easy
In Lakehouse Monitoring, what is required when defining a custom metric?
Select an answer first
28foundation · easy
Which statistical method is commonly used to detect data drift by comparing the distribution of a numerical feature between a baseline and current data?
Select an answer first
29application · medium
A data science team monitors a churn-prediction model. They want to be notified via email when the PSI (Population Stability Index) for the 'age' feature exceeds 0.25. Which configuration steps are required in Lakehouse Monitoring?
Select an answer first
30foundation · easy
Which of the following is a required step when constructing a monitor for an inference table?
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 6)

Questions 26–30

In Lakehouse Monitoring, what is required when defining a custom metric?

Which statistical method is commonly used to detect data drift by comparing the distribution of a numerical feature between a baseline and current data?

A data science team monitors a churn-prediction model. They want to be notified via email when the PSI (Population Stability Index) for the 'age' feature exceeds 0.25. Which configuration steps are required in Lakehouse Monitoring?

Which of the following is a required step when constructing a monitor for an inference table?

Keep studying — ML Ops