Describe Azure Machine Learning Capabilities AI-900 Practice Questions (Page 2)

Questions 6–10

1. 6expert · hard

   A company needs to train a deep learning model on a large dataset. The training job is long-running and can be interrupted. They want to minimize costs while maintaining the ability to complete the job. Which compute strategy should they use?

   • AUse a compute instance and keep it running until the job completes.
   • BUse Azure Databricks with a job cluster that terminates after the job.
   • CUse a dedicated compute cluster with a high VM size and no checkpointing.
   • DUse a low-priority compute cluster with checkpointing enabled.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A company has multiple models deployed to different endpoints. They want to track which model version is used by each endpoint and monitor for performance degradation. What should they implement?

   • AUse pipelines to monitor endpoints
   • BUse run history to track deployments
   • CUse the model registry to tag versions and Application Insights for monitoring
   • DUse datastores to store model metadata
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   A model deployed to an AKS endpoint is showing high latency. The team suspects the model is not handling concurrent requests efficiently. They need to diagnose the issue and improve performance. Which approach should they take?

   • ARestart the endpoint to clear the cache.
   • BUse Application Insights to analyze request latency and then adjust the deployment's concurrency settings or scale replicas.
   • CUse Azure Monitor to check CPU usage and then disable autoscaling.
   • DDeploy the model to a larger VM size without analyzing the issue.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A retail company wants to build a churn prediction model but has no data science team. They have historical customer data stored in Azure Blob Storage and need to quickly compare several algorithms to find the best one. Which Azure service should they use?

   • AAzure Machine Learning automated ML
   • BAzure Cognitive Services
   • CAzure Synapse Analytics
   • DAzure Databricks
   Reveal Solution Discussion Select an answer first
6. 10expert · hard

   A team needs to build a model for a binary classification problem. They have a highly imbalanced dataset and want to use Automated ML. They also need to ensure that the model's performance is evaluated using a metric that accounts for the imbalance. What should they do?

   • AUse a single algorithm with class weights and no Automated ML.
   • BUse accuracy as the primary metric and disable featurization.
   • CConfigure Automated ML to use AUC_weighted as the primary metric and enable data featurization.
   • DManually balance the dataset and then run Automated ML with accuracy as the metric.
   Reveal Solution Discussion Select an answer first