Azure AI Machine Learning Pipeline Reference

Overview

This page provides a comprehensive guide to the Azure AI Machine Learning Pipeline. It allows users to construct and manage different stages of the pipeline, enhancing the capabilities of your AI models.

The pipeline enables seamless deployment and scaling of machine learning applications.

The Pipeline Stages

The pipeline is divided into several stages, each representing a distinct step in the ML lifecycle:

• Data Ingestion: Receive data from various sources.
• Data Transformation: Clean, transform, and prepare data.
• Feature Engineering: Create and select relevant features.
• Model Training: Train and optimize machine learning models.
• Model Evaluation: Assess model performance.
• Model Deployment: Deploy the model to a production environment.
• Model Monitoring: Track model performance & relevance.

The Pipeline Flow

The pipeline is initiated with a data source and then progresses through each stage, using the Azure Machine Learning Studio, Azure Machine Learning Designer, or the Azure CLI. Each stage requires a specific configuration and potentially, custom scripts.

Example: Data Transformation

Let's consider a simple example: Extracting a 'city' column from a CSV file.

Our pipeline will ingest the CSV data, transform the 'city' column using a function, and store it in a new table.