Yggdrasil Decision Forests (YDF) is a production grade collection of algorithms, developed and used by Google, for the training, serving, and interpretation of Decision Forest models. YDF is open-source and is available in C++, CLI (command-line-interface), TensorFlow (under the name TensorFlow Decision Forests;TF-DF), Javascript, and Go (inference only).

See the Features page for the full list of features of YDF. Read the Yggdrasil Decision Forests paper for details about YDF’s design.

TensorFlow Decision Forests (TF-DF) is the official port of YDF to TensorFlow using the Keras API. TF-DF makes it easy to train and use decision forests in the TensorFlow ecosystem.

Content

• Why decision forests?

CLI API

• Quick start
• Installation
• Commands
• User manual

C++ API

• Example

TF-DF / Python API

• Quick Start
• Installation
• Tutorials

Reference

• Training APIs
• How to improve models
• Hyper-Parameters
• Early Stopping
• Evaluation metrics
• What are decision forests
• Features
• Long-time-support

Deploy a model

• Serving APIs
• C++
• Go
• JavaScript/Wasm
• TF Serving
• Python
• Benchmark
• Model formats

Resources

• Changelog
• YDF on Github
• TF-DF on Github
• TF-DF on tf.org
• Contributing
• Credits