🔍

Azure Machine Learning API Reference

Explore the comprehensive REST API for Azure Machine Learning, enabling programmatic control and integration of your machine learning workflows.

Workspace Management

GET /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.MachineLearningServices/workspaces/{workspaceName}

Get Workspace Details

Retrieves detailed information about a specific Azure Machine Learning workspace.

Parameters:

Name Type Description Required
subscriptionId string Your Azure subscription ID. Yes
resourceGroupName string The name of the resource group containing the workspace. Yes
workspaceName string The name of the Azure ML workspace. Yes

Responses:

200 OK

{
  "name": "my-ml-workspace",
  "id": "/subscriptions/...",
  "type": "Microsoft.MachineLearningServices/workspaces",
  "location": "East US",
  "properties": {
    "friendlyName": "My ML Workspace",
    "creationTime": "2023-10-27T10:00:00Z",
    "description": "Workspace for managing ML experiments and models.",
    "storageAccount": "/subscriptions/.../resourceGroups/.../providers/Microsoft.Storage/storageAccounts/mystorage",
    "keyVault": "/subscriptions/.../resourceGroups/.../providers/Microsoft.KeyVault/vaults/mykeyvault",
    "applicationInsights": "/subscriptions/.../resourceGroups/.../providers/Microsoft.Insights/components/myappinsights"
  }
}
POST /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.MachineLearningServices/workspaces/{workspaceName}/create

Create Workspace

Creates a new Azure Machine Learning workspace. This is a simplified example; actual creation involves many more properties.

Parameters:

Name Type Description Required
subscriptionId string Your Azure subscription ID. Yes
resourceGroupName string The name of the resource group to create the workspace in. Yes
workspaceName string The desired name for the new Azure ML workspace. Yes

Request Body:

{
  "location": "West US 2",
  "tags": {
    "environment": "production"
  },
  "sku": {
    "name": "Basic"
  }
}

Responses:

201 Created

{
  "id": "/subscriptions/...",
  "name": "my-new-workspace",
  "type": "Microsoft.MachineLearningServices/workspaces",
  "location": "West US 2",
  "properties": {
    "provisioningState": "Creating",
    "creationTime": "..."
  }
}

Compute Management

GET /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.MachineLearningServices/workspaces/{workspaceName}/computes/{computeName}

Get Compute Resource

Retrieves information about a specific compute resource (e.g., Compute Cluster, AKS, Managed Endpoint).

Parameters:

Name Type Description Required
subscriptionId string Your Azure subscription ID. Yes
resourceGroupName string The name of the resource group. Yes
workspaceName string The name of the Azure ML workspace. Yes
computeName string The name of the compute resource. Yes
POST /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.MachineLearningServices/workspaces/{workspaceName}/computes/create

Create Compute Resource

Creates a new compute resource for training or deployment within the workspace.

Parameters:

Name Type Description Required
subscriptionId string Your Azure subscription ID. Yes
resourceGroupName string The name of the resource group. Yes
workspaceName string The name of the Azure ML workspace. Yes

Request Body:

{
  "computeType": "AmlCompute",
  "properties": {
    "vmSize": "STANDARD_DS3_V2",
    "scaleSettings": {
      "maxNodes": 5,
      "minNodes": 0,
      "targetNodeUtilizationPercentage": 60
    }
  }
}

Experiments and Runs

GET /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.MachineLearningServices/workspaces/{workspaceName}/experiments/{experimentName}/runs

List Runs

Retrieves a list of runs for a given experiment.

Parameters:

Name Type Description Required
subscriptionId string Your Azure subscription ID. Yes
resourceGroupName string The name of the resource group. Yes
workspaceName string The name of the Azure ML workspace. Yes
experimentName string The name of the experiment. Yes
top integer Maximum number of runs to return. No
skip integer Number of runs to skip for pagination. No

Common Authentication

All Azure Machine Learning API requests must be authenticated. Typically, this is done using Azure Active Directory (Azure AD) service principals or managed identities. You will need to include an Authorization header with a valid OAuth 2.0 Bearer token in your requests.

Authorization: Bearer <YOUR_ACCESS_TOKEN>

Rate Limiting

Azure Machine Learning APIs are subject to rate limits. Exceeding these limits may result in temporary throttling. Consult the Azure documentation for specific limits.

Error Handling

API responses will include standard HTTP status codes. Error responses typically contain a JSON body with details about the error, including an error code and a message.

{
  "error": {
    "code": "InvalidRequest",
    "message": "The specified resource was not found."
  }
}