A fast, distributed, high performance gradient boosting (GBT, GBDT, GBRT, GBM or MART) framework based on decision tree algorithms, used for ranking, classification and many other machine learning tasks.

data-mining decision-trees distributed gbdt gbm gbrt gradient-boosting kaggle lightgbm machine-learning microsoft parallel python r

Перейти к файлу

Guolin Ke c2ba086ce4 fix bug in sparse bin iterator		2017-01-10 23:09:02 +08:00
.github	adding issues template	2016-12-23 17:03:49 +08:00
R-package	change init_score to double type	2017-01-10 20:52:42 +08:00
docs	solve libpath issue (#187 )	2017-01-10 14:28:46 +08:00
examples	python code style with pep8 (#161 )	2017-01-04 15:19:12 +08:00
include/LightGBM	change init_score to double type	2017-01-10 20:52:42 +08:00
pmml	refine pmml.py (#179 )	2017-01-10 00:27:52 +08:00
python-package	fix bug in sparse bin iterator	2017-01-10 23:09:02 +08:00
src	fix bug in sparse bin iterator	2017-01-10 23:09:02 +08:00
tests	refine pmml.py (#179 )	2017-01-10 00:27:52 +08:00
windows	add voting based parallel	2016-12-13 15:53:01 +08:00
.gitignore	re-define callback order (#114 )	2016-12-07 15:14:54 +08:00
.travis.yml	python code style with pep8 (#161 )	2017-01-04 15:19:12 +08:00
CMakeLists.txt	refine Dataset class (#113 )	2016-12-08 21:36:11 +08:00
LICENSE	Add license.	2016-10-11 15:42:22 +08:00
README.md	[R-package] update examples	2017-01-08 10:09:36 +08:00

README.md

LightGBM, Light Gradient Boosting Machine

LightGBM is a gradient boosting framework that uses tree based learning algorithms. It is designed to be distributed and efficient with the following advantages:

Faster training speed and higher efficiency
Lower memory usage
Better accuracy
Parallel learning supported
Capable of handling large-scale data

For more details, please refer to Features.

Experiments on public datasets show that LightGBM can outperform other existing boosting framework on both efficiency and accuracy, with significant lower memory consumption. What's more, the experiments show that LightGBM can achieve a linear speed-up by using multiple machines for training in specific settings.

News

01/08/2017 : Release R-package beta version, welcome to have a try and provide feedback.

12/05/2016 : Categorical Features as input directly(without one-hot coding). Experiment on Expo data shows about 8x speed-up with same accuracy compared with one-hot coding (refer to categorical log and one-hot log). For the setting details, please refer to IO Parameters.

12/02/2016 : Release python-package beta version, welcome to have a try and provide feedback.

Get Started And Documents

To get started, please follow the Installation Guide and Quick Start.

How to Contribute

LightGBM has been developed and used by many active community members. Your help is very valuable to make it better for everyone.

Check out call for contributions to see what can be improved, or open an issue if you want something.
Contribute to the tests to make it more reliable.
Contribute to the documents to make it clearly for everyone.
Contribute to the examples to share your experience with other users.
Check out Development Guide.
Open issue if you met problems during development.

Microsoft Open Source Code of Conduct

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.