Azure Machine Learning (AzureML)

Deploy Azure Machine Learning (AzureML) Service

Note

You will only deploy the service once.

Sign in to the Azure Portal as described in the sign in page.
In the search bar at the top, type azure machine learning
Click on Azure Machine Learning entry under Services
From the Azure Machine Learning service page (which will be empty), Click on + Create
Select New workspace from the dropdown
Select your subscription from the Subscription drow-down
Select the resource group you crated from the dropdown (most likely dsan6000)
Enter aml as the Name. You will notice that Storage account, Key vault, and Application insights values will auto-populate, and the Container registry value will be “None”. Do not change anything in those four fields.
Make sure the region is East US 2
Click Review + create
Click Create once validation passes

Your AzureML workspace will begin to be deployed. You’ll see a notification when it’s done.

Once the workspace is deployed you can access it by a direct URL (bookmark this): https://ml.azure.com/.

If you are not logged into Azure you will be asked to log in.

Important

You will start with creating (and using) a single compute instance. To create additional compute instances, the process is the same.

Log into AzureML Studio
Select your workspace
Click on Compute on the left-hand navigation panel (towards the bottom)
Click + New
Enter a Compute name (you can edit the default value.) This value has to be globally unique, so we recommend calling the compute instance dsan6000-<NETID>-ci01 (change to your own.)
Select CPU
Select the Standard_E4ds_v4 instance size
Click Next
Make sure Auto shut down is enabled and change the value to 30
Click Review + Create
Verify all settings
Click Create

When the compute instance gets created it will be running. It should shut-down after 30 minutes of inactivity.

You can start and stop the compute instance from within the AzureML Studio or