In recent days, Microsoft’s improvements to Power BI include the release of the October update for On-premises data gateway, the introduction of new contact lists for reports and dashboards, and plenty more. Earlier this year, the Redmond firm revealed the public preview of Automated Machine Learning (AutoML) for Dataflows in Power BI.
Today, AutoML has reached general availability in all public cloud regions that offer Power BI Premium and Embedded services. A bunch of new capabilities have been added to the service ever since its preview version became available in April.
For those unaware, AutoML allows business analysts to easily develop machine learning (ML) models. Automatic ‘guardrails’ to aid this process include class balancing, training-test data split, cross-validation, and more. Essentially, this means that much of the background knowledge required for training models – something that data scientists, as opposed to business analysts, would have – is not required.
The following features were offered in the public preview:
- Train a machine learning model to perform Binary Prediction, General Classification, and Regression
- View the model training report
- Apply the ML model to their data, and view predictions and explanations
In condensed form, further capabilities that have been introduced with general availability include:
- Binary prediction support for non-Boolean outcomes
Improved Feature Recommendation
Controlling training time
Improved training reports
You can learn more about each of these new features in more detail by reading the original blog post. A tutorial on how to build machine learning models using AutoML can be found here, while the service’s full documentation is available here.