Train a model with AutoML

AutoML is a feature of Azure Databricks that tries multiple algorithms and parameters with your data to train an optimal machine learning model.

This exercise should take approximately 30 minutes to complete.

Before you start

You’ll need an Azure subscription in which you have administrative-level access.

Provision an Azure Databricks workspace

An Azure Databricks workspace provides a central cloud-based environment for all your data science activities in Azure Databricks. You can create a workspace interactively in the Azure portal, use one you already have, or follow the steps below to use a script to provision a Premium workspace for this exercise.

Note: For this exercise, you need a Premium Azure Databricks workspace in a region where model serving is supported. See Azure Databricks regions for details of regional Azure Databricks capabilities. A Premium workspace is not required to use AutoML to train a model, but it is required to deploy a model. If you already have a Premium or Trial Azure Databricks workspace, you can skip this procedure and use your existing workspace.

  1. In a web browser, sign into the Azure portal at https://portal.azure.com.

  2. Use the [>_] button to the right of the search bar at the top of the page to create a new Cloud Shell in the Azure portal, selecting a PowerShell environment and creating storage if prompted. The cloud shell provides a command line interface in a pane at the bottom of the Azure portal, as shown here:

    Azure portal with a cloud shell pane

    Note: If you have previously created a cloud shell that uses a Bash environment, use the the drop-down menu at the top left of the cloud shell pane to change it to PowerShell.

  3. Note that you can resize the cloud shell by dragging the separator bar at the top of the pane, or by using the —, ◻, and X icons at the top right of the pane to minimize, maximize, and close the pane. For more information about using the Azure Cloud Shell, see the Azure Cloud Shell documentation.

  4. In the PowerShell pane, enter the following commands to clone this repo:

     rm -r mslearn-databricks-ml -f
 git clone https://github.com/MicrosoftLearning/mslearn-databricks-ml

  5. After the repo has been cloned, enter the following commands to change to the folder for this lab and run the setup.ps1 script it contains:

     cd mslearn-databricks-ml/Allfiles/labs/04
 ./setup.ps1

  6. If prompted, choose which subscription you want to use (this will only happen if you have access to multiple Azure subscriptions).

  7. Wait for the script to complete - this typically takes around 5 minutes, but in some cases may take longer. While you are waiting, review the What is AutoML? article in the Azure Databricks documentation.

Create a cluster

Azure Databricks is a distributed processing platform that uses Apache Spark clusters to process data in parallel on multiple nodes. Each cluster consists of a driver node to coordinate the work, and worker nodes to perform processing tasks. In this exercise, you’ll create a single-node cluster to minimize the compute resources used in the lab environment (in which resources may be constrained). In a production environment, you’d typically create a cluster with multiple worker nodes.

Tip: If you already have a cluster with a 13.3 LTS ML or higher runtime version in your Azure Databricks workspace, you can use it to complete this exercise and skip this procedure.

  1. In the Azure portal, browse to the msl-xxxxxxx resource group that was created by the script (or the resource group containing your existing Azure Databricks workspace)
  2. Select your Azure Databricks Service resource (named databricksxxxxxxx if you used the setup script to create it).

  3. In the Overview page for your workspace, use the Launch Workspace button to open your Azure Databricks workspace in a new browser tab; signing in if prompted.

    Tip: As you use the Databricks Workspace portal, various tips and notifications may be displayed. Dismiss these and follow the instructions provided to complete the tasks in this exercise.

  4. In the sidebar on the left, select the (+) New task, and then select Cluster.
  5. In the New Cluster page, create a new cluster with the following settings:
    • Cluster name: User Name’s cluster (the default cluster name)
    • Policy: Unrestricted
    • Cluster mode: Single Node
    • Access mode: Single user (with your user account selected)
    • Databricks runtime version: Select the ML edition of the latest non-beta version of the runtime (Not a Standard runtime version) that:
      • Does not use a GPU
      • Includes Scala > 2.11
      • Includes Spark > 3.0
    • Use Photon Acceleration: unselected
    • Node type: Standard_DS3_v2
    • Terminate after 30 minutes of inactivity
  6. Wait for the cluster to be created. It may take a minute or two.

Note: If your cluster fails to start, your subscription may have insufficient quota in the region where your Azure Databricks workspace is provisioned. See CPU core limit prevents cluster creation for details. If this happens, you can try deleting your workspace and creating a new one in a different region. You can specify a region as a parameter for the setup script like this: ./setup.ps1 eastus

Upload training data to a SQL Warehouse

To train a machine learning model using AutoML, you need to upload the training data. In this exercise, you’ll train a model to classify a penguin as one of three species based on observations including its location and body measurements. You’ll load training data that includes the species label into a table in an Azure Databricks data warehouse.

  1. In the Azure Databricks portal for your workspace, in the sidebar, under SQL, select SQL Warehouses.
  2. Observe that the workspace already includes a SQL Warehouse named Starter Warehouse.
  3. In the Actions (⁝) menu for the SQL Warehouse, select Edit. Then set the Cluster size property to 2X-Small and save your changes.
  4. Use the Start button to start the SQL Warehouse (which may take a minute or two).

Note: If your SQL Warehouse fails to start, your subscription may have insufficient quota in the region where your Azure Databricks workspace is provisioned. See Required Azure vCPU quota for details. If this happens, you can try requesting for a quota increase as detailed in the error message when the warehouse fails to start. Alternatively, you can try deleting your workspace and creating a new one in a different region. You can specify a region as a parameter for the setup script like this: ./setup.ps1 eastus

  1. Download the penguins.csv file to your local computer, saving it as penguins.csv.
  2. In the Azure Databricks workspace portal, in the sidebar, select (+) New and then select File Upload and upload the penguins.csv file you downloaded to your computer.
  3. In the Upload data page, select the default schema and set the table name to penguins. Then select Create table on the bottom left corner of the page.
  4. When the table has been created, review its details.

Create an AutoML experiment

Now that you have some data, you can use it with AutoML to train a model.

  1. In the sidebar on the left, select Experiments.
  2. On the Experiments page, select Create AutoML experiment.
  3. Configure the AutoML experiment wuth the following settings:
    • Cluster: Select your cluster
    • ML problem type: Classification
    • Input training dataset: Browse to the default database and select the penguins table
    • Prediction target: Species
    • Experiment name: Penguin-classification
    • Advanced configuration:
      • Evaluation metric: Precision
      • Training frameworks: lightgbm, sklearn, xgboost
      • Timeout: 5
      • Time column for training/validation/testing split: Leave blank
      • Positive label: Leave blank
      • Intermediate data storage location: MLflow Artifact
  4. Use the Start AutoML button to start the experiment. Close any information dialogs that are displayed.
  5. Wait for the experiment to complete. You can use the Refresh button on the right to view details of the runs that are generated.
  6. After five minutes, the experiment will end. Refreshing the runs will show the run that resulted in the best performing model (based on the precision metric you selected) at the top of the list.

Deploy the best performing model

Having run an AutoML experiment, you can explore the best performing model that it generated.

  1. In the Penguin-classification experiment page, select View notebook for best model to open the notebook used to rain the model in a new browser tab.
  2. Scroll through the cells in the notebook, noting the code that was used to train the model.
  3. Close the browser tab containing the notebook to return to the Penguin-classification experiment page.
  4. In the list of runs, select the name of the first run (which produced the best model) to open it.
  5. In the Artifacts section, note that the model has been saved as an MLflow artifact. Then use the Register model button to register the model as a new model named Penguin-Classifier.

  6. In the sidebar on the left, switch to the Models page. Then select the Penguin-Classifier model you just registered.

    Note: The remainder of this exercise requires a Premium Azure Databricks workspace.

  7. On the Penguin-Classifier page, use the Use model for inference button to create a new real-time endpoint with the following settings:
    • Model: Penguin-Classifier
    • Model version: 1
    • Endpoint: classify-penguin
    • Compute size: Small

    The serving endpoint is hosted in a new cluster, which it may take several minutes to create.

  8. When the endpoint has been created, use the Query endpoint button at the top right to open an interface from which you can test the endpoint. Then in the test interface, on the Browser tab, enter the following JSON request and use the Send Request button to call the endpoint and generate a prediction.

     {
   "dataframe_records": [
   {
      "Island": "Biscoe",
      "CulmenLength": 48.7,
      "CulmenDepth": 14.1,
      "FlipperLength": 210,
      "BodyMass": 4450
   }
   ]
 }
  9. Experiment with a few different values for the penguin features and observe the results that are returned. Then, close the test interface.

Delete the endpoint

When the endpoint is not longer required, you should delete it to avoid unnecessary costs.

In the classify-penguin endpoint page, in the ⁝ menu, select Delete.

Clean-up

If you’re finished working with Azure Databricks for now, in Azure Databricks workspace, on the Compute page, select your cluster and select ■ Terminate to shut it down. Then, on the SQL Warehouses page, stop your SQL warehouse.

If you don’t plan to use your workspace again, close the Azure Databricks web portal, return to the Azure portal, and delete the msl-xxxxxxx resource group containing your Azure Databricks workspace.