Back to Gap Map Infrastructure / Classic ML & computer vision

NGBoost

Stanford ML Group
open source / Overall score: 2.6

NGBoost implements natural gradient boosting, a boosting method that predicts a full probability distribution for each example rather than a single value. Built on scikit-learn, it lets the user choose the distribution, the proper scoring rule and the base learner, and returns predictive distributions alongside point predictions. It comes from the Stanford ML Group.

Openness

5 high confidence
5.0
license
Apache-2.0 (OSI)
source
public (github.com/stanfordmlgroup/ngboost)
core features withheld
no — Stanford ML Group research library; nothing sold

Apache-2.0 and built from the public repository. It is a research library from Stanford's ML Group, and nothing is sold beside it.

Adoption

3 high confidence
3.0

Measured on monthly PyPI downloads of the ngboost package.

Capability

2 high confidence
2.0

NGBoost is one boosting method, engineered to output a predictive distribution, much as XGBoost is one gradient-boosted tree method engineered for speed and scale. Neither offers a catalog of other learners to choose from.

Verified 2026-11-12