dmlc/xgboost
The KDD paper that became infrastructure
It exists to make gradient boosting fast and portable enough to run on a laptop or a billion-row cluster without changing code.
Not currently ranked — collecting fresh signals.
star history
What it does
XGBoost trains ensembles of decision trees using gradient boosting. The project wraps a C++ core in bindings for Python, R, Java, Scala and more, targeting fast, accurate solutions to tabular data problems. According to the README, the same codebase scales from a single machine to distributed environments including Kubernetes, Hadoop, SGE, Dask, Spark and PySpark.
The interesting bit
Portability is treated as a first-class feature rather than an afterthought: the library originated as a University of Washington research project and now ships as infrastructure that runs the same code from a laptop to a cluster handling billions of examples. That academic-to-production trajectory is rare; most research projects do not end up with hardware vendors on their sponsor page.
Key highlights
- C++ core with language wrappers for Python, R, Java, Scala and others
- Distributed backends include Kubernetes, Hadoop, SGE, Dask, Spark and PySpark
- Handles datasets with billions of examples, per the README
- Apache-2 licensed; development funded partly by an open-source collective
- Canonical reference is the KDD 2016 paper by Chen and Guestrin
Verdict
If you work with tabular data and need a model that scales from a notebook to a Kubernetes cluster, this is a sensible default. If your problem involves images, text, or non-structured data, look elsewhere.
Frequently asked
- What is dmlc/xgboost?
- It exists to make gradient boosting fast and portable enough to run on a laptop or a billion-row cluster without changing code.
- Is xgboost open source?
- Yes — dmlc/xgboost is open source, released under the Apache-2.0 license.
- What language is xgboost written in?
- dmlc/xgboost is primarily written in C++.
- How popular is xgboost?
- dmlc/xgboost has 28.6k stars on GitHub.
- Where can I find xgboost?
- dmlc/xgboost is on GitHub at https://github.com/dmlc/xgboost.