Sparse feature format:
- Sparse feature format allows easy handling of missing values, and improve computation efficiency.
Push the limit on single machine:
- Efficient implementation that optimizes memory and computation.
Layout of gradient boosting algorithm to support generic tasks, see project wiki.

Planned key components

Gradient boosting models:
- regression tree (GBRT)
- linear model/lasso
Objectives to support tasks:
- regression
- classification
- ranking
- matrix factorization
- structured prediction (3) OpenMP implementation

File extension convention: (1) .h are interface, utils and data structures, with detailed comment; (2) .cpp are implementations that will be compiled, with less comment; (3) .hpp are implementations that will be included by .cpp, with less comment

Languages

C++ 45.5%

Python 20.3%

Cuda 15.2%

R 6.8%

Scala 6.4%

Other 5.6%