README

azure-ml-sdk 1.0.0

A Python SDK for Azure Machine Learning.

Introduction

This package is used to make ML experiments on Azure Machine Learning workspace.

A Python environment is created and pushed to an Azure registry and used mainly in running pipelines in Azure Devops. Also, this package can be installed using pip and used to launch jobs or handle assets individually.

Installation

How can we install it:

  1. Configure the file pip.ini by following the wiki to get access to packages.

  2. install the package:

pip install azure-ml-sdk

Structure

The package is composed of a collection of modules responsible of handling different service in the package:

  • services: The core module of the package which provides services to Azure Machine Learning:

    • connection: A submodule for setting up connection to a workspace

    • asset: A submodule for handling assets (environments, computes, datasets, models)

    • deployment: A submodule for handling deplyments (online or batch)

  • configs: A module for handling configuration files and parsing them

  • utils: An utilitary module (handling environment variables, mlflow models, name validators)

Notebooks

  • tutorial-az-ml-services.ipynb: The main notebook on how to use the package

  • tutorial-parser.ipynb: A notebook about how to parse config files

  • tutorial_mlflow.ipynb: A notebook about how to handle mlflow models

  • tutorial_ml_pipeline.ipynb: An example of machine learning experiment