jhajduk-microsoft/dp-100-exam

Study artifacts from Hands On Lab experience

Hands-On Lab: Microsoft DP-100 Exam Preparation

Topic 1: Design a Machine Learning Solution and Determine Compute Specifications

Objective: In this lab, you will design a machine learning solution and determine the appropriate compute specifications for a training workload using Azure Machine Learning.

1. Create an Azure Machine Learning Workspace:

   • Open the Azure Portal and create a new Azure Machine Learning Workspace. • Creation of container registry for images • High business impact workspace for sensitive data. No telemetry to MSFT

2. Prepare the Data:

   • Download a sample dataset for training, such as the Iris dataset, and save it to Azure Storage.

3. Set Up Git Integration:

   • Enable Git integration for version control of your machine learning project in the Azure Machine Learning Workspace. • Use compute instances to integrate Git • Compute will be created in the same location as the workspace

4. Define the Machine Learning Solution:

   • Define the problem statement and objectives of the machine learning project.
   • Determine the appropriate algorithm for the chosen problem.

5. Create Compute Targets:

   • Set up various compute targets, such as CPU, GPU, and distributed clusters, for different training workloads.

6. Select an Environment:

   • Choose a pre-built environment or create a custom environment suitable for the selected algorithm.

7. Configure Attached Compute Resources:

   • Set up Apache Spark pools for distributed data processing if required.

8. Determine Compute Specifications:

   • Use Azure Machine Learning compute monitoring tools to evaluate the compute utilization and decide the appropriate compute specifications for training workloads.

Topic 2: Model Deployment Requirements and Development Approach Selection

Objective: In this lab, you will explore model deployment requirements and choose the appropriate development approach for building or training a model.

1. Prepare the Model:

   • Train a machine learning model on the prepared dataset using the chosen compute target and environment.

2. Evaluate Model Performance:

   • Use various evaluation metrics to assess the model's performance.

3. Define Model Deployment Requirements:

   • Determine the necessary deployment resources, such as memory, processing power, and concurrency.

4. Choose the Development Approach:

   • Decide whether to use automated machine learning, custom code development, or a combination of both for model development.

5. Deploy the Model:

   • Deploy the trained model as a web service or container in Azure.

6. Set Up Scaling and Monitoring:

   • Configure scaling options and monitoring for the deployed model.

Topic 3: Manage Azure Machine Learning Workspace and Data

Objective: In this lab, you will learn how to manage an Azure Machine Learning Workspace and handle data-related tasks.

1. Register Datastores:

   • Register the Azure Storage resources you plan to use as datastores in the Azure Machine Learning Workspace.

2. Create and Manage Data Assets:

   • Create data assets in the Azure Machine Learning Workspace for easy access during model training.

3. Monitor Data Usage:

   • Use Azure Machine Learning tools to monitor data usage and data distribution across different compute targets.

Topic 4: Manage Compute for Experiments in Azure Machine Learning

Objective: In this lab, you will manage the compute resources for experiments in Azure Machine Learning.

1. Create Compute Targets:

   • Create different compute targets like CPU clusters, GPU clusters, and AKS clusters in the Azure Machine Learning Workspace.

2. Select an Environment:

   • Choose an appropriate environment for a specific machine learning use case (e.g., TensorFlow, PyTorch, scikit-learn).

3. Configure Attached Compute Resources:

   • Configure attached compute resources, such as data storage and network settings, for the chosen compute targets.

4. Monitor Compute Utilization:

   • Use Azure Machine Learning's monitoring tools to track compute utilization and identify bottlenecks.

The above hands-on labs will help you gain practical experience in the key topics covered in the Microsoft DP-100 exam. Remember to review the official Microsoft documentation for Azure Machine Learning and practice using the Azure Portal and Azure CLI to reinforce your understanding. Good luck with your exam preparation!