Microsoft Certified Azure Data Scientist Associate - DP-100

Get started today

Ultimate access to all questions.

Explanation:

The question requires selecting two sampling methods that can handle both continuous (learning_rate: 0.001 to 0.1) and discrete (batch_size: 16, 32, 64) hyperparameters without an early termination policy. Based on Azure ML documentation and community consensus (57% selected CD with upvoted comments):

C (Bayesian sampling) supports continuous and discrete parameters and does not require early termination policies.
D (Random sampling) supports both parameter types and can be used without early termination.
A (No sampling) is invalid as it would not sample from the defined ranges.
B (Grid sampling) only supports discrete parameters and cannot handle the continuous learning_rate range effectively without discretization, making it unsuitable as a complete solution.

Explanation:

C (Bayesian sampling) supports continuous and discrete parameters and does not require early termination policies.
D (Random sampling) supports both parameter types and can be used without early termination.
A (No sampling) is invalid as it would not sample from the defined ranges.
B (Grid sampling) only supports discrete parameters and cannot handle the continuous learning_rate range effectively without discretization, making it unsuitable as a complete solution.

Comments (0)

No comments yet.

You plan to use Azure Machine Learning's Hyperdrive to find the optimal hyperparameters for training a model. You need to test combinations of the following values without using an early termination policy:

`learning_rate`: any value between 0.001 and 0.1

`batch_size`: 16, 32, or 64

Which two sampling methods should you configure for the Hyperdrive experiment? Each correct answer presents a complete solution.

Exam-Like

Last updated: April 14, 2026 at 14:02

No sampling

0.0%

Grid sampling

25.0%

Bayesian sampling

25.0%

Random sampling

50.0%