TransmogrifAI is a popular AutoML library coded in Scala programming language. The library/framework runs on Apache Spark. It’s originally focused on accelerating the productivity of machine learning developers by facilitating them with machine learning automation.

The library also provides access to an API that enforces compile-time type-safety, reusability, and modularity. Users of this framework can achieve a high level of accuracy with an almost 100x reduction in time.

TransmogrifAI is an idea choice for engineers that want to develop production-ready ML applications in a few hours. They can create modules, strongly types ML workflows, and these are reusable modules.

Model insights use stored feature metadata to help with debugging. It also provides valuable insights to end-users, making ML models less of a black box.

Project Background

• • Library: AutoML library
  • Author: Kevin Moore, Kin Fai Kan, Leah McGuire, Matthew Tovbin, Max Ovsiankin, Michael Loh, Michael Weil, Shubha Nabar, Vitaly Gordon, Vlad Patryshev
  • Initial Release: 2018
  • Type:  Accelerates ML developer productivity
  • License: BSD-3-Clause License
  • Contains: Data objects
  • Language: Scala
  • GitHub: salesforce/TransmogrifAI
  • Runs On: Apache Spark
  • GitHub Discussions: /salesforce/TransmogrifAI/discussions
  • Stackflow: questions/tagged/transmogrifai

Applications

• Multi-class Classification
• Regression
• Aggregation and joins
• Conditional aggregation