Machine Learning Pipelines for Statistics and Machine Learning Toolbox (Beta)

Try new Beta features to create, visualize, and run pipelines for machine learning workflows.

Note: Machine Learning Pipelines are included in the R2026a release. As a result, this Machine Learning Pipelines Beta is no longer supported. Please use the Machine Learning Pipelines functionality in R2026a or later.

Machine Learning Pipelines Beta offers functions to build pipelines from machine learning components, enabling you to manage multi-step workflows including pre-processing, training machine learning models, and making predictions. You can visualize the workflow by viewing the pipeline. Trained pipelines can also be deployed to enterprise applications with MATLAB Production Server, via standalone applications with MATLAB Compiler, and via shared libraries with MATLAB Compiler SDK.

By utilizing pipelines, you can establish standardized workflows for common machine learning tasks, thereby accelerating development time, reducing errors, and easing deployment.

To get started:

1. Download and install this Add-On from File Exchange or the Add-On Manager
2. In the MATLAB command window, type the command: "activate_ml_pipelines" to enable the Beta functionality
3. Open the example as a Live Script using the command "open('BasicPipelineExample.mlx')" in the command window. You can run this script section by section to see pipelines in action. Several interactive pipeline view windows will open.
4. View the documentation using the command: "open('Pipelines_beta_book.pdf')", and open an additional example using: "open('CrossValidationExample.mlx')".