What is Azure Machine Learning designer?
Azure Machine Learning designer is a drag-and-drop interface used to train and deploy models in Azure Machine Learning. This article describes the tasks you can do in the designer.
- To get started with the designer, see Tutorial: Train a no-code regression model.
- To learn about the components available in the designer, see the Algorithm and component reference.
The designer uses your Azure Machine Learning workspace to organize shared resources such as:
Model training and deployment
Use a visual canvas to build an end-to-end machine learning workflow. Train, test, and deploy models all in the designer:
- Drag-and-drop datasets and modules onto the canvas.
- Connect the modules to create a pipeline draft.
- Submit a pipeline run using the compute resources in your Azure Machine Learning workspace.
- Convert your training pipelines to inference pipelines.
- Publish your pipelines to a REST pipeline endpoint to submit a new pipeline that runs with different parameters and datasets.
- Publish a training pipeline to reuse a single pipeline to train multiple models while changing parameters and datasets.
- Publish a batch inference pipeline to make predictions on new data by using a previously trained model.
- Deploy a real-time inference pipeline to a real-time endpoint to make predictions on new data in real time.
A pipeline consists of datasets and algorithmic modules, which you connect. Pipelines have many uses: you can make a pipeline that trains a single model, or one that trains multiple models. You can create a pipeline that makes predictions in real time or in batch, or make a pipeline that only cleans data. Pipelines let you reuse your work and organize your projects.
As you edit a pipeline in the designer, your progress is saved as a pipeline draft. You can edit a pipeline draft at any point by adding or removing modules, configuring compute targets, creating parameters, and so on.
A valid pipeline has these characteristics:
- Datasets can only connect to modules.
- Modules can only connect to either datasets or other modules.
- All input ports for modules must have some connection to the data flow.
- All required parameters for each module must be set.
When you're ready to run your pipeline draft, you submit a pipeline run.
Each time you run a pipeline, the configuration of the pipeline and its results are stored in your workspace as a pipeline run. You can go back to any pipeline run to inspect it for troubleshooting or auditing. Clone a pipeline run to create a new pipeline draft for you to edit.
Pipeline runs are grouped into experiments to organize run history. You can set the experiment for every pipeline run.
A machine learning dataset makes it easy to access and work with your data. Several sample datasets are included in the designer for you to experiment with. You can register more datasets as you need them.
A module is an algorithm that you can perform on your data. The designer has several modules ranging from data ingress functions to training, scoring, and validation processes.
A module may have a set of parameters that you can use to configure the module's internal algorithms. When you select a module on the canvas, the module's parameters are displayed in the Properties pane to the right of the canvas. You can modify the parameters in that pane to tune your pipeline. You can set the compute resources for individual modules in the designer.
For some help navigating through the library of machine learning algorithms available, see Algorithm & module reference overview. For help with choosing an algorithm, see the Azure Machine Learning Algorithm Cheat Sheet.
Use compute resources from your workspace to run your pipeline and host your deployed models as real-time endpoints or pipeline endpoints (for batch inference). The supported compute targets are:
|Azure Machine Learning compute||✓|
|Azure Kubernetes Service||✓|
To perform real-time inferencing, you must deploy a pipeline as a real-time endpoint. The real-time endpoint creates an interface between an external application and your scoring model. A call to a real-time endpoint returns prediction results to the application in real time. To make a call to a real-time endpoint, you pass the API key that was created when you deployed the endpoint. The endpoint is based on REST, a popular architecture choice for web programming projects.
Real-time endpoints must be deployed to an Azure Kubernetes Service cluster.
To learn how to deploy your model, see Tutorial: Deploy a machine learning model with the designer.
Azure Machine Learning Endpoints (preview) provide an improved, simpler deployment experience. Endpoints support both real-time and batch inference scenarios. Endpoints provide a unified interface to invoke and manage model deployments across compute types. See What are Azure Machine Learning endpoints (preview)?.
You can also publish a pipeline to a pipeline endpoint. Similar to a real-time endpoint, a pipeline endpoint lets you submit new pipeline runs from external applications using REST calls. However, you cannot send or receive data in real time using a pipeline endpoint.
Published pipelines are flexible, they can be used to train or retrain models, perform batch inferencing, process new data, and much more. You can publish multiple pipelines to a single pipeline endpoint and specify which pipeline version to run.
A published pipeline runs on the compute resources you define in the pipeline draft for each module.
The designer creates the same PublishedPipeline object as the SDK.
- Learn the fundamentals of predictive analytics and machine learning with Tutorial: Predict automobile price with the designer
- Learn how to modify existing designer samples to adapt them to your needs.