Introduction

  • 2 minutes

Going through trial and error to find the best performing model can be time-consuming. Instead of manually having to test and evaluate various configurations to train a machine learning model, you can automate it with automated machine learning or AutoML.

AutoML allows you to try multiple preprocessing transformations and algorithms with your data to find the best machine learning model.

Diagram illustrating how multiple models are trained and evaluated, after the model with the best performing metrics is selected.

Imagine you want to find the best performing classification model. You can create an AutoML experiment using the visual interface of Azure Machine Learning studio, the Azure command-line interface (CLI), or the Python software development kit (SDK).

Note

You can use AutoML for other tasks such as regression, forecasting, image classification, and natural language processing. Learn more about when you can use AutoML.

As a data scientist, you may prefer to configure your AutoML experiment with the Python SDK.

Learning objectives

In this module, you'll learn how to:

  • Prepare your data to use AutoML for classification.
  • Configure and run an AutoML experiment.
  • Evaluate and compare models.