R-package

LightGBM R Package

Installation

Windows users may need to run with administrator rights (either R or the command prompt, depending on the way you are installing this package). Rtools must be installed for Windows. Linux users might require the appropriate user write permissions for packages.

You can use a command prompt to install via command line:

cd R-package
R CMD INSTALL --build  .

You can also install directly from R using the repository with devtools:

devtools::install_github("Microsoft/LightGBM", subdir = "R-package")

For the devtools install scenario, you can safely ignore this message:

Warning message:
GitHub repo contains submodules, may not function as expected! 

If you want to build the self-contained R package, you can run unix_build_package.sh(for UNIX) or win_build_package.cmd (for Windows). Then use R CMD INSTALL lightgbm_0.1.tar.gz to install.

When your package installation is done, you can check quickly if your LightGBM R package is working by running the following:

library(lightgbm)
data(agaricus.train, package='lightgbm')
train <- agaricus.train
dtrain <- lgb.Dataset(train$data, label=train$label)
params <- list(objective="regression", metric="l2")
model <- lgb.cv(params, dtrain, 10, nfold=5, min_data=1, learning_rate=1, early_stopping_rounds=10)

OSX installation

The default installation cannot successfully complete in OSX because clang doesn't support OpenMP.

You can use the following script to change default compiler to gcc, then compile LightGBM R package:

brew install gcc --without-multilib
mkdir -p ~/.R
touch ~/.R/Makevars
cat <<EOF >>~/.R/Makevars
C=gcc-6
CXX=g++-6
CXX1X=g++-6
LDFLAGS=-L/usr/local/Cellar/gcc/6.3.0/lib
CPPFLAGS=-I/usr/local/Cellar/gcc/6.3.0/include
SHLIB_OPENMP_CFLAGS = -fopenmp
SHLIB_OPENMP_CXXFLAGS = -fopenmp
SHLIB_OPENMP_FCFLAGS = -fopenmp
SHLIB_OPENMP_FFLAGS = -fopenmp
EOF

Note:

• For LDFLAGS=-L/usr/local/Cellar/gcc/6.3.0/lib and CPPFLAGS=-I/usr/local/Cellar/gcc/6.3.0/include, you may need to change 6.3.0 to your gcc version.
• For gcc-6 and g++-6, you may need to change to your gcc version (like gcc-7 and g++7 if using gcc with version 7).
• For CXX1X, if you are using R 3.4 or a more recent version, you must change it to CXX11.

To check your LightGBM installation, the test is identical to Linux/Windows versions (check the test provided just before OSX Installation part)

Performance note

With gcc, it is recommended to use -O3 -mtune=native instead of the default -O2 -mtune=core2 by modifying the appropriate file (Makeconf or Makevars) if you want to achieve maximum speed.

Benchmark example using Intel Ivy Bridge CPU on 1M x 1K dataset:

Compilation Flag	Performance Index
-O2 -mtune=core2	100.00%
-O2 -mtune=native	100.90%
-O3 -mtune=native	102.78%
-O3 -ffast-math -mtune=native	100.64%

Examples

Please visit demo:

• Basic walkthrough of wrappers
• Boosting from existing prediction
• Early Stopping
• Cross Validation
• Multiclass Training/Prediction
• Leaf (in)Stability
• Weight-Parameter Adjustment Relationship