Azure Machine Learning Studio Intermediate

Azure Machine Learning Studio is a comprehensive, enterprise-grade platform for building, training, and deploying machine learning models. It provides a visual Designer for no-code ML, AutoML for automated model selection, and full SDK support for code-first workflows.

Creating a Workspace

Terminal 
# Create an Azure ML workspace
az ml workspace create \
  --name my-ml-workspace \
  --resource-group ai-school-rg \
  --location eastus
Python 
from azure.ai.ml import MLClient
from azure.identity import DefaultAzureCredential

# Connect to workspace
ml_client = MLClient(
    credential=DefaultAzureCredential(),
    subscription_id="YOUR_SUBSCRIPTION_ID",
    resource_group_name="ai-school-rg",
    workspace_name="my-ml-workspace"
)

AutoML

AutoML automatically selects the best algorithm and hyperparameters for your dataset. It supports classification, regression, time series forecasting, computer vision, and NLP tasks.

Python 
from azure.ai.ml import automl, Input

# Configure AutoML classification job
classification_job = automl.classification(
    compute="gpu-cluster",
    experiment_name="my-automl-experiment",
    training_data=Input(type="mltable", path="azureml://datastores/training/paths/data/"),
    target_column_name="label",
    primary_metric="accuracy",
    n_cross_validations=5
)

# Set limits
classification_job.set_limits(
    timeout_minutes=60,
    max_trials=20,
    max_concurrent_trials=4
)

# Submit the job
returned_job = ml_client.jobs.create_or_update(classification_job)
print(f"Job URL: {returned_job.studio_url}")

Designer (Visual ML)

The Designer provides a drag-and-drop interface for building ML pipelines without writing code. It includes pre-built modules for data transformation, feature engineering, model training, and evaluation.

Module Category Examples
Data Input/Output Import Data, Export Data, Enter Data Manually
Data Transformation Select Columns, Clean Missing Data, Normalize Data, Join Data
Feature Engineering Feature Hashing, Extract N-Gram Features, One-Hot Encoding
Training Train Model, Tune Model Hyperparameters, Cross Validate
Evaluation Evaluate Model, Score Model, Permutation Feature Importance

Managed Compute

Azure ML provides several compute options for different workloads:

Python 
from azure.ai.ml.entities import AmlCompute

# Create a GPU compute cluster
gpu_cluster = AmlCompute(
    name="gpu-cluster",
    type="amlcompute",
    size="Standard_NC6s_v3",  # V100 GPU
    min_instances=0,
    max_instances=4,
    idle_time_before_scale_down=120  # Scale down after 2 min idle
)

ml_client.compute.begin_create_or_update(gpu_cluster)

Model Deployment

Python 
from azure.ai.ml.entities import ManagedOnlineEndpoint, ManagedOnlineDeployment

# Create an online endpoint
endpoint = ManagedOnlineEndpoint(
    name="my-endpoint",
    auth_mode="key"
)
ml_client.online_endpoints.begin_create_or_update(endpoint)

# Deploy the model
deployment = ManagedOnlineDeployment(
    name="blue",
    endpoint_name="my-endpoint",
    model="azureml:my-model:1",
    instance_type="Standard_DS3_v2",
    instance_count=1
)
ml_client.online_deployments.begin_create_or_update(deployment)
Pro Tip: Use compute instances for interactive development (notebooks) and compute clusters for training jobs. Set min_instances=0 on clusters to automatically scale down and avoid idle costs.

ML Studio Mastered!

You can now build and deploy custom ML models with Azure. Next, explore Azure Cognitive Services for pre-built AI capabilities.

Next: Cognitive Services →
← Azure OpenAI Cognitive Services →