Create Azure Machine Learning Workspace – Machine Learning Fundamental Concepts

Create Azure Machine Learning Workspace

1.

Click “+ Create a resource” and then search for “machine learning” to provision an Azure Machine Learning resource as shown in Figure 3-4.

Figure 3-4Creating a new Azure ML resource

2.

As shown in Figure 3-5, fill in the information on the “Basic” tab, and on the last blade, select Review + Create, and then select Create. You have to provide a subscription, resource group, workspace name, and region. Here, the container registry is none, while storage account, key vault, and application insights are provisioning new. Wait for your workspace to be created.

Figure 3-5Creating a new Azure Machine Learning workspace

3.

Then, navigate to Azure Machine Learning Studio in the new tab, and sign in using your Microsoft account as shown in Figure 3-6.

Figure 3-6Navigate to the newly created resource group and launch Azure ML Studio

4.

As shown in Figure 3-7, your newly provisioned workspace will appear here.

Figure 3-7Azure ML workspace

Create Compute

1.

Now, select the Compute tab (under Manage) and select Compute clusters as shown in Figure 3-8.

Figure 3-8Azure Machine Learning workspace: creating and selecting compute clusters

2.

On the Compute page, select Compute clusters and add a new one to train a machine learning model. As shown in the Figure 3-9 and Figure 3-10, provide the location, virtual machine tier, virtual machine type, the minimum number of nodes, and a maximum number of nodes, and click Create.

Figure 3-9Provisioning a virtual machine in compute clusters

Figure 3-10Advanced setting for compute clusters

Create Pipeline in Designer

1.

Expand the left pane in Azure Machine Learning Studio by clicking the three lines icon. Under Author, select Designer and + to create a pipeline as shown in Figure 3-11.

Figure 3-11Selecting Designer in the Azure Machine Learning Studio

2.	At the top right-hand side of the screen, select Settings. If the Settings pane is not visible, select the wheel icon next to the pipeline name at the top.
3.	As shown in Figure 3-12, in Settings, you must specify a compute target on which to run the pipeline. Under Select compute type, select Compute cluster. Select the previously created Azure ML to compute cluster.

Figure 3-12Configuring compute for the pipeline

4.

In Settings, under Draft Details, change the draft name (Pipeline-Created-on-date) to Auto Price Training, as shown in Figure 3-13. Select the close icon on the top right of the Settings pane to close the pane.

Figure 3-13Configuring settings for the pipeline