A DLT pipeline performs streaming inference on IoT sensor data. The model requires a feature that is the average temperature per device over the last 5 minutes. The pipeline is using a sliding window with a 1-minute slide. What is the correct way to define this in DLT?

Use a materialized view to compute the average.

Use a groupBy on the device ID without a window function.

Use a filter to select only the last 5 minutes of data before applying the model.

Use a groupBy with a window function on the event time column.

Which DLT feature is essential for stateful streaming inference scenarios that require aggregations or windowing over time?

Databricks SQL's materialized views

Structured Streaming's built-in support for stateful operations like windowed aggregations

A DLT pipeline performs streaming inference on financial transactions. The model requires a feature that is the total transaction amount per user over a 24-hour sliding window. The pipeline is running with a 24-hour window and a 1-hour slide. The team notices that the pipeline is running out of memory. What is the most effective way to reduce memory usage without losing accuracy?

Enable state TTL (time-to-live) to expire old state entries that are no longer needed.

Use a tumbling window instead of a sliding window to reduce state size.

Increase the slide interval to 2 hours to reduce the number of overlapping windows.

Reduce the window duration to 12 hours to halve the state size.

A DLT pipeline for streaming inference is running slower than expected. The team suspects that the model inference is the bottleneck. How can they confirm this?

Look at the cluster's CPU utilization to see if it is maxed out.

Review the model's training time in MLflow.

Check the streaming query progress metrics to see the processing time per batch.

Check the Kafka topic's partition count.

A DLT pipeline applies a model to streaming data that requires a feature computed from a 30-minute tumbling window. The pipeline is running on a cluster with limited memory. The team wants to reduce memory usage while maintaining the exact same feature semantics. Which approach is most appropriate?

Use a materialized view instead of a streaming table to avoid state management.

Enable state TTL with a duration slightly longer than the window duration.

Use a sliding window with a 1-minute slide to approximate the tumbling window.

Reduce the window duration to 15 minutes to halve the state size.

DatabricksCertified Machine Learning Associate

Domain 4Objective 4

Identify How Streaming Inference Is Performed with Delta Live Tables MACHINE-LEARNING-ASSOCIATE Practice Questions (Page 1)

Part of the Section 4: Model Deployment domain, which makes up ~30% of our current practice bank. Databricks does not publish an official question count, but from its 90-minute exam (~35–60 total, ~11–18 in this domain), expect 2–3 from this objective — we provide 20 practice questions to prepare you well beyond it. (estimate)

20questions here

4free pages

5concepts

Questions 1–5

1application · medium
A DLT pipeline performs streaming inference on IoT sensor data. The model requires a feature that is the average temperature per device over the last 5 minutes. The pipeline is using a sliding window with a 1-minute slide. What is the correct way to define this in DLT?
Select an answer first
2foundation · easy
Which DLT feature is essential for stateful streaming inference scenarios that require aggregations or windowing over time?
Select an answer first
3expert · hard
A DLT pipeline performs streaming inference on financial transactions. The model requires a feature that is the total transaction amount per user over a 24-hour sliding window. The pipeline is running with a 24-hour window and a 1-hour slide. The team notices that the pipeline is running out of memory. What is the most effective way to reduce memory usage without losing accuracy?
Select an answer first
4application · medium
A DLT pipeline for streaming inference is running slower than expected. The team suspects that the model inference is the bottleneck. How can they confirm this?
Select an answer first
5expert · hard
A DLT pipeline applies a model to streaming data that requires a feature computed from a 30-minute tumbling window. The pipeline is running on a cluster with limited memory. The team wants to reduce memory usage while maintaining the exact same feature semantics. Which approach is most appropriate?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to MACHINE-LEARNING-ASSOCIATE

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-ASSOCIATE” is a trademark of its owner, used for identification only.