Train and manage a machine learning model with Azure Machine Learning
Microsoft Learn
Last updated November 11, 2024
A practical, end-to-end path on training and managing a model with Azure Machine Learning. It starts by setting up a workspace and making data available through datastores and data assets, then configuring compute targets and environments to run workloads. From there it covers converting code into a script and running it as a command job, tracking training metrics with MLflow, registering the trained model, and finally deploying it to a managed online endpoint for real-time predictions. The result is a single thread that runs from raw data to a live, queryable model.
What you'll learn
- Setting up an Azure Machine Learning workspace and making data available with datastores and data assets
- Working with compute targets and environments
- Running a training script as a command job
- Tracking training with MLflow and registering the model
- Deploying a model to a managed online endpoint for real-time predictions
Frequently asked questions about Train and manage a machine learning model with Azure Machine Learning
Who is Train and manage a machine learning model with Azure Machine Learning for?
People new to Azure Machine Learning who want to train, track, and deploy a model end to end.
Is Train and manage a machine learning model with Azure Machine Learning free?
Yes — Train and manage a machine learning model with Azure Machine Learning is completely free to take.
What are the prerequisites for Train and manage a machine learning model with Azure Machine Learning?
None stated. As a Develop-track path it is hands-on: expect to work with code and the Azure platform.
Do you need to code for Train and manage a machine learning model with Azure Machine Learning?
Yes — Train and manage a machine learning model with Azure Machine Learning involves hands-on coding.
Why we suggest this course
For someone taking their first run through Azure Machine Learning, this path follows one model the whole way — set up the workspace, make data available, train via a command job, track with MLflow, register, and deploy to an online endpoint — so the pieces connect into a complete workflow rather than isolated features. One thing to know: it is rated Beginner for its track and lists no prerequisites, but as a Develop-track path it is hands-on and expects you to work with code and the Azure platform throughout.