Describe Azure Machine Learning Capabilities AI-900 Practice Questions (Page 7)

Questions 31–35

1. 31application · medium

   A model deployed to an endpoint is receiving requests, but the team wants to see the number of requests and response times over time. What should they use?

   • APipeline
   • BDatastore
   • CAzure Monitor and Application Insights
   • DModel registry
   Reveal Solution Discussion Select an answer first
2. 32expert · hard

   A company has data in an on-premises SQL Server and wants to use Azure ML to train a model. They have strict data residency requirements that prevent copying data to the cloud. What should they do?

   • ATrain the model on-premises using Azure Arc
   • BCopy the data to Azure Blob Storage in a specific region
   • CUse Azure Data Factory to move data to Azure SQL Database
   • DUse Azure ML with a datastore that references the on-premises SQL Server via a gateway
   Reveal Solution Discussion Select an answer first
4. 33expert · hard

   A team is setting up an Automated ML experiment. They have data in an Azure Data Lake Gen2 storage account. They need to ensure that the experiment can access the data securely without exposing credentials, and they want to deploy the best model to a real-time endpoint. Which steps should they take?

   • AMount the Data Lake Gen2 as a local drive on a compute instance and run Automated ML there.
   • BCreate a datastore that uses managed identity to access Data Lake Gen2, create a dataset from the datastore, run Automated ML, and deploy to AKS.
   • CUse a shared access signature (SAS) token in the datastore and deploy to ACI.
   • DCopy the data to a public blob container and use the URL directly in the experiment.
   Reveal Solution Discussion Select an answer first
5. 34foundation · easy

   Which Azure Machine Learning feature allows you to record metrics, parameters, and artifacts for each training run?

   • AModel registry
   • BExperiments and runs
   • CDatastores
   • DPipelines
   Reveal Solution Discussion Select an answer first
6. 35foundation · easy

   What is the purpose of monitoring a deployed endpoint in Azure Machine Learning?

   • ATo automatically retrain the model when accuracy drops.
   • BTo change the model version without redeploying.
   • CTo secure the endpoint with authentication keys.
   • DTo track the health and performance of the endpoint, such as request latency and error rates.
   Reveal Solution Discussion Select an answer first