Automated Machine Learning - Dataset and Compute

This guide offers a step-by-step process for efficiently setting up and utilizing automated machine learning with your datasets and compute resources.

Hafeez Baig
26 steps
3 minutes

Azure

This lab is a continuation of the previous one, Please refer to it before proceeding - <https://scribehow.com/viewer/Getting_Started_with_Azure_Machine_Learning__vJwa1BJjREOJkaRm9iIQ-w>

After the deployment is completed, click on **Go to resource** button

**Overview** page will open, click on **Launch studio** button

**Azure AI** page will open, here you can explore the ML Workspace

Click on the **Compute** option from the left bar

**Compute** page will open, here you can create Compute instances

Click on the **Compute clusters** tab, here you can create compute clusters

**What are compute clusters in Azure AI?** **Compute clusters in Azure AI** are scalable groups of virtual machines used to run machine learning workloads such as training, inference, or data processing. They automatically scale up or down based on job demand, helping optimize cost and performance. Compute clusters are managed within **Azure Machine Learning**, supporting parallel and distributed workloads efficiently.

Scroll the page and click on the **New** button

**Create compute cluster** page will open, select **Dedicated** radio button for Virtual machine tier

Keep the default Machine type - **Standard_DS3_v2** and click on the **Next** buttion

Give the compute name as "**ml-compute-v1**"

Keep the **Minimum number of nodes** as 0

Keep the **Maximum number of nodes** as 1

Set the **Idle seconds before scale down** to **900**

**What is Idle seconds before scale down?** **Idle seconds before scale down** is a setting in Azure compute clusters that defines how long a node must remain idle (not running any job) before it is automatically deallocated. This helps optimize resource usage and reduce costs by shutting down unused compute nodes after a specified period of inactivity.

Scroll the page and click on **Create** button

It will take some time for the compute cluster to succeed

Click on **Data** option from the left bar

**Data** page will open, click on **Create** button

**Create data asset** wizard will open, give the Name as "**my-dataset-1**"

Click and select the **Type** as **Tabular** from the dropdown list

**What is Tabular type in data asset?** A **Tabular type in data asset** in Azure Machine Learning refers to structured data organized in rows and columns, like a table or spreadsheet. This type is commonly used for training machine learning models on datasets such as CSV files, SQL tables, or Excel sheets. It supports tasks like classification, regression, and forecasting.

Click on the **Next** button

**Data source** tab will open, select From **Azure Open Datasets** option and click on the **Next** button

**What is Azure Open Datasets?** **Azure Open Datasets** are curated, publicly available datasets provided by Microsoft to support machine learning and data science projects. These datasets include domains like weather, healthcare, finance, and public safety, and are optimized for use in Azure Machine Learning. They help accelerate model training and experimentation without the need for data collection.

**Azure Open Dataset** tab will open, select the **Sample: Diabetes** option and click on the **Next** button

**Review** tab will open, review the details and click on the **Create** button

**my-dataset-1** page will open, here you can view all the details about your dataset

Click on the **Explore** tab, here you can **Preview** the rows and columns in the dataset

Click on the **Profile** tab under **Explore** page, here you can view the data profile

Congratulations on successfully completing this lab and creating dataset in Azure AI!

Refer to this lab to continue learning - <https://scribehow.com/viewer/Setting_up_Automated_ML_Training__wdW9LoAQQR2-mjwgnmsVVA>

Automated Machine Learning - Dataset and Compute

This guide offers a step-by-step process for efficiently setting up and utilizing automated machine learning with your datasets and compute resources.

Hafeez Baig
26 steps
3 minutes