Azure Machine Learning

Getting Started

1. On the machine learning Overview page, click Launch studio.

   

2. Use the drop-down to select vdl subscription and the Machine Learning workspace you want to access, then click Get started.

   

3. Once inside your Machine Learning workspace, you can train, deploy and manage machine learning models, use AutoML, and run pipelines. See Getting started quickly for more information.

   

Using Azure ML Notebook standalone

Requirements

A compute instance in Azure ML. You should see it under Compute --> Compute instances.

Note: If a compute instance has not been created for you, please contact the support team via Slack.

Steps

1. Under Notebooks, create a new notebook in your user directory. You can then enter the code to execute.

   

2. Select the Compute instance assigned to you.

   

3. Click the run all cells button to execute your code.

   

Using Databricks Connect as Remote Compute

Disclaimer: Please note that the Databricks connect configuration shown below is under revision and will likely change in the near future.

Requirements

A compute instance in Azure ML. You should see it under Compute --> Compute instances.

Note: If a compute instance has not been created for you, please contact the support team via Slack.

Steps

1. Under Notebooks, open Terminal.

   

2. Select your Compute instance from the drop-down next to Compute.

3. Execute the code from Databricks Connect Setup in the terminal, while following the prompts to continue as needed. This code installs Python 3.7 and sets up a new kernel for Azure ML notebooks.

   When prompted, enter the following values to configure Databricks connect:

   Host: the URL from the Overview page for your Databricks workspace.

   

   Token: the personal access token generated in your Databricks Workspace User Settings.

   Cluster ID: the value found under Cluster --> Advanced Options --> Tags in your Databricks workspace.

   

   Org ID: the part of the Databricks URL found after .net/?o=

   

   Port: keep the existing value

4. Execute the following code in terminal to test the connectivity to Azure Databricks. databricks-connect test

5. Create a new notebook with Azure ML and select the Python 3 kernel. It should now display Python 3.7.9

   

6. Databricks connect should be setup now! Try the Databricks connect example code in a notebook, replacing public-data/incoming/1test.txt with the path to a file in your data lake container.

Request compute

Please contact the support team through the slack channel to request Azure ML compute. You will receive the following error when creating it yourself: