Describe Azure Machine Learning Capabilities AI-900 Practice Questions (Page 6)

Questions 26–30

1. 26application · medium

   A manufacturing company wants to predict equipment failure using sensor data. They have a large dataset with many features and limited data science expertise. They need to run automated machine learning to find the best model, but they also need to ensure the experiment is reproducible and can be audited later. What should they do?

   • AUse a pre-built Azure Cognitive Services anomaly detector to identify failures.
   • BUse Automated ML in Azure Machine Learning and log all runs with parameters and metrics.
   • CUse Azure Databricks to run a custom PySpark script that trains a single model.
   • DUse Azure Machine Learning Designer to manually build a model pipeline.
   Reveal Solution Discussion Select an answer first
2. 27expert · hard

   A team has deployed a model to an AKS endpoint in Azure Machine Learning. They notice that the endpoint occasionally returns errors during traffic spikes. They need to improve reliability without over-provisioning resources. Which combination of actions should they take?

   • AEnable autoscaling on the AKS deployment and configure Azure Monitor alerts for error rates.
   • BUse Azure DevOps to manually restart the endpoint during spikes.
   • CIncrease the replica count to a fixed high number and disable autoscaling.
   • DDeploy the model to Azure Container Instances instead.
   Reveal Solution Discussion Select an answer first
4. 28expert · hard

   A team needs to build a model to predict customer churn. They have a large dataset with many missing values and categorical features. They have limited time and want to use Automated ML, but they also need to ensure the model is explainable for regulatory purposes. Which approach should they take?

   • ARun Automated ML with data preprocessing and enable model explainability.
   • BUse Azure Machine Learning Designer to build a pipeline and then deploy it.
   • CManually clean the data and then run Automated ML without explainability.
   • DUse a single algorithm with default hyperparameters and use SHAP to explain it.
   Reveal Solution Discussion Select an answer first
5. 29foundation · easy

   Which Azure service is commonly used to store and manage tabular data for machine learning workflows?

   • AAzure Cognitive Search
   • BAzure DNS
   • CAzure Blob Storage
   • DAzure Monitor
   Reveal Solution Discussion Select an answer first
6. 30foundation · easy

   What is the purpose of model versioning in Azure Machine Learning?

   • ATo deploy the model to multiple endpoints simultaneously.
   • BTo encrypt the model file for secure storage.
   • CTo track different iterations of a model so you can compare and roll back to previous versions.
   • DTo automatically retrain the model when new data arrives.
   Reveal Solution Discussion Select an answer first