Developing Models in Microsoft Azure Flashcards
What is the purpose of an early termination policy?
To truncate the experiment and save computing resources
To truncate the workspace and save computing resources
To truncate under-performing run and save computing resources
To kill the run instance.
To truncate under-performing run and save computing resources
One of the exit criteria in your automated machine learning experiment is experiment_timeout_minutes. What does this setting allow you to define?
How long the environment associated with the experiment will remain active
How long each run in an experiment can continue to run
How long the compute target associated with the experiment will remain active
How long an experiment can continue in run
How long an experiment can continue in run
Which parameter is not needed to create Estimators?
A reference to computing resources
A script parameter and its location
A reference to workspace
Execution environment and its dependencies
A reference to workspace
Which property of Run objects is needed to get a reference to it?
Run duration property
The Run ID property
Run number property
Run name property
The Run ID property
You are starting a new experiment in which you only need computing resources until the training run, and would like to be decommissioned after the run. What is an ideal candidate?
Run-based managed compute
Cloud-based VM
Attached compute
Persisted managed compute
Run-based managed compute
While creating a Machine Learning workspace, which resource is not created?
Microsoft Key Vault
Microsoft compute target
Microsoft Insight
Microsoft Storage
Microsoft compute target
What Machine Learning environment is not supported by Microsoft Azure?
User-managed
System-managed
Remotely managed
Curated
Remotely managed
You are working on preprocessing data as part of your model creation. You need to remove all the columns from your data except columns date, age, ssn and phone. What method would you use?
data. remove_columns([‘date’,’age’,’ssn’,’phone’])
data. retain_columns([‘date’,’age’,’ssn’,’phone’])
data. extract_columns([‘date’,’age’,’ssn’,’phone’])
data. keep_columns([‘date’,’age’,’ssn’,’phone’])
data.keep_columns([‘date’,’age’,’ssn’,’phone’])
You are working on a dataset and as part of your preprocessing, you need to retain data for employees whose income is less than 100,000. What will achieve this?
data. remove(data[‘income’] > 100000)
data. split(data[‘income’] < 100000)
data. filter(data[‘income’] < 100000)
data. filter(data[‘income’] > 100000)
data.filter(data[‘income’] < 100000)
Which is not a sampling strategy supported by Microsoft Azure?
Random sampling
Stratified sampling
Bayesian sampling
Grid sampling
Stratified sampling
