Microsoft Certified: Azure Data Scientist Associate Practice Exam: Test Your Knowledge 2025

You are designing a machine learning solution for a retail company that needs to predict customer churn. The data science team has limited experience with infrastructure management and wants to focus on model development. Which Azure service should you recommend?

A data scientist is training a classification model in Azure Machine Learning. During training, they notice the model achieves 98% accuracy on the training set but only 65% accuracy on the validation set. What problem is occurring and what should be done?

You need to deploy a trained machine learning model to Azure that will handle real-time predictions with low latency requirements (under 100ms). The model receives individual prediction requests from a web application. Which deployment option should you choose?

You are preparing a machine learning model for deployment. You need to ensure that the model can be versioned, tracked, and easily rolled back if issues occur in production. Which Azure Machine Learning feature should you use?

A company wants to establish responsible AI practices for their machine learning projects in Azure. They need to understand which features in a model contribute most to predictions. Which Azure Machine Learning capability should they implement?

You are building a training pipeline in Azure Machine Learning. The pipeline includes data preprocessing, feature engineering, model training, and evaluation steps. You need to ensure that intermediate results are cached and reused when only the model training step changes. What should you configure?

A data science team is working with sensitive healthcare data. They need to train models in Azure Machine Learning while ensuring that data scientists cannot download or export the raw data from the workspace. Which combination of features should be implemented?

You are training a deep learning model using Azure Machine Learning compute clusters. The training job frequently fails due to node failures in the cluster. You need to ensure training can resume from the last checkpoint without losing progress. What should you implement?

You have deployed a model to an Azure Machine Learning real-time endpoint. You notice that prediction latency increases significantly during peak hours. The endpoint currently uses a single instance. What should you do to improve performance during peak times while minimizing costs?

You need to convert a trained scikit-learn model to ONNX format before deployment to improve inference performance across different platforms. Which Azure Machine Learning component should you use?

A retail company wants to use Azure Machine Learning to forecast sales. They have historical sales data stored in Azure SQL Database with millions of records. The data needs to be incrementally loaded for model retraining. Which approach should you use?

You have deployed a classification model to production. After several weeks, you notice the model's accuracy has degraded from 85% to 70%. What is the most likely cause and what should you monitor?

You need to perform hyperparameter tuning for a neural network model in Azure Machine Learning. The search space is large, and you want to find optimal parameters quickly while managing compute costs. Which sampling method should you use?

You are preparing to deploy a model that requires specific Python packages and dependencies. You need to ensure the deployment environment exactly matches the training environment. What should you create and register?

A financial services company needs to audit all model predictions for compliance purposes. They must track which model version made each prediction along with the input features and timestamp. What should you implement?

You are implementing a machine learning pipeline that includes parallel processing of multiple datasets for training ensemble models. Each dataset requires different preprocessing steps. The pipeline must optimize execution time and resource utilization. How should you configure the pipeline?

Your organization uses Azure Machine Learning workspaces across multiple Azure subscriptions for different business units. You need to implement a governance strategy that ensures consistent security policies, naming conventions, and resource tagging while allowing business units autonomy in model development. What Azure feature should you implement?

You have trained a computer vision model using Azure Machine Learning and need to deploy it to edge devices with limited connectivity and processing power. The model must make predictions locally when offline. Which deployment approach should you use?

You are deploying a high-stakes medical diagnosis model that requires human review before predictions are acted upon. The model must integrate with an existing approval workflow system via webhooks. You need to implement A/B testing between the current model and a new version while maintaining the approval workflow. What deployment strategy should you implement?

You are building a recommendation system that requires training on a massive dataset that doesn't fit in memory. The training process involves multiple iterations over the data with different feature transformations. You need to optimize the data loading strategy to minimize training time. What approach should you implement?