32-bit Python is not supported. Please install 64-bit version. If you have a strong need to install with 32-bit Python, refer to Build 32-bit Version with 32-bit Python section.

setuptools is needed.

pip install lightgbm

You may need to install wheel via pip install wheel first.

Compiled library that is included in the wheel file supports both GPU and CPU versions out of the box. This feature is experimental and available only for Windows and Linux currently. To use GPU version you only need to install OpenCL Runtime libraries. For NVIDIA and AMD GPU they are included in the ordinary drivers for your graphics card, so no action is required. If you would like your AMD or Intel CPU to act like a GPU (for testing and debugging) you can install AMD APP SDK on Windows and PoCL on Linux. Many modern Linux distributions provide packages for PoCL, look for pocl-opencl-icd on Debian-based distributions and pocl on RedHat-based distributions.

For Windows users, VC runtime is needed if Visual Studio (2015 or newer) is not installed.

For Linux users, glibc >= 2.14 is required for LightGBM <=3.3.3 and glibc >= 2.28 is required for newer versions. Also, in some rare cases, when you hit OSError: libgomp.so.1: cannot open shared object file: No such file or directory error during importing LightGBM, you need to install OpenMP runtime library separately (use your package manager and search for lib[g|i]omp for doing this).

For macOS (we provide wheels for 3 newest macOS versions) users:

• Starting from version 2.2.1, the library file in distribution wheels is built by the Apple Clang (Xcode_8.3.3 for versions 2.2.1 - 2.3.1, Xcode_9.4.1 for versions 2.3.2 - 3.3.2 and Xcode_11.7 from version 4.0.0) compiler. This means that you don't need to install the gcc compiler anymore. Instead of that you need to install the OpenMP library, which is required for running LightGBM on the system with the Apple Clang compiler. You can install the OpenMP library by the following command: brew install libomp.
• For version smaller than 2.2.1 and not smaller than 2.1.2, gcc-8 with OpenMP support must be installed first. Refer to Installation Guide for installation of gcc-8 with OpenMP support.
• For version smaller than 2.1.2, gcc-7 with OpenMP is required.

pip install --no-binary :all: lightgbm

For Linux and macOS users, installation from sources requires installed CMake.

For Linux users, glibc >= 2.28 is required. Also, in some rare cases you may need to install OpenMP runtime library separately (use your package manager and search for lib[g|i]omp for doing this).

For macOS users, you can perform installation either with Apple Clang or gcc.

• In case you prefer Apple Clang, you should install OpenMP (details for installation can be found in Installation Guide) first and CMake version 3.16 or higher is required.
• In case you prefer gcc, you need to install it (details for installation can be found in Installation Guide) and specify compilers by running export CXX=g++-7 CC=gcc-7 (replace "7" with version of gcc installed on your machine) first.

For Windows users, Visual Studio (or VS Build Tools) is needed. If you get any errors during installation, you may need to install CMake (version 3.8 or higher).

pip install lightgbm --install-option=--nomp

All requirements, except the OpenMP requirement, from Build from Sources section apply for this installation option as well.

It is strongly not recommended to use this version of LightGBM!

pip install lightgbm --install-option=--mpi

All requirements from Build from Sources section apply for this installation option as well.

For Windows users, compilation with MinGW-w64 is not supported and CMake (version 3.8 or higher) is strongly required.

MPI libraries are needed: details for installation can be found in Installation Guide.

pip install lightgbm --install-option=--gpu

All requirements from Build from Sources section apply for this installation option as well.

For Windows users, CMake (version 3.8 or higher) is strongly required.

Boost and OpenCL are needed: details for installation can be found in Installation Guide. Almost always you also need to pass OpenCL_INCLUDE_DIR, OpenCL_LIBRARY options for Linux and BOOST_ROOT, BOOST_LIBRARYDIR options for Windows to CMake via pip options, like

pip install lightgbm --install-option=--gpu --install-option="--opencl-include-dir=/usr/local/cuda/include/" --install-option="--opencl-library=/usr/local/cuda/lib64/libOpenCL.so"

All available options:

• boost-root
• boost-dir
• boost-include-dir
• boost-librarydir
• opencl-include-dir
• opencl-library

For more details see FindBoost and FindOpenCL.

pip install lightgbm --install-option=--cuda

All requirements from Build from Sources section apply for this installation option as well, and CMake (version 3.16 or higher) is strongly required.

CUDA library (version 10.0 or higher) is needed: details for installation can be found in Installation Guide.

To use the CUDA version within Python, pass {"device": "cuda"} respectively in parameters.

pip install lightgbm --install-option=--hdfs

All requirements from Build from Sources section apply for this installation option as well.

HDFS library is needed: details for installation can be found in Installation Guide.

Note that the installation process of HDFS version was tested only on Linux.

pip install lightgbm --install-option=--mingw

CMake and MinGW-w64 should be installed first.

It is recommended to use Visual Studio for its better multithreading efficiency in Windows for many-core systems (see Question 4 and Question 8).

pip install lightgbm --install-option=--bit32

By default, installation in environment with 32-bit Python is prohibited. However, you can remove this prohibition on your own risk by passing bit32 option.

It is strongly not recommended to use this version of LightGBM!

pip install lightgbm --install-option=--time-costs

Use this option to make LightGBM output time costs for different internal routines, to investigate and benchmark its performance.

If you use conda to manage Python dependencies, you can install LightGBM using conda install.

We strongly recommend installation from the conda-forge channel and not from the default one due to many reasons. The main ones are less time delay for new releases, greater number of supported architectures and better handling of dependency conflicts, especially workaround for OpenMP is crucial for LightGBM. More details can be found in this comment.

Note: The lightgbm conda-forge feedstock is not maintained by LightGBM maintainers.

conda install -c conda-forge lightgbm

All requirements from Build from Sources section apply for this installation option as well.

For Windows users, if you get any errors during installation and there is the warning WARNING:LightGBM:Compilation with MSBuild from existing solution file failed. in the log, you should install CMake (version 3.8 or higher).

git clone --recursive https://github.com/microsoft/LightGBM.git
cd LightGBM/python-package
# export CXX=g++-7 CC=gcc-7  # macOS users, if you decided to compile with gcc, don't forget to specify compilers (replace "7" with version of gcc installed on your machine)
python setup.py install

Note: sudo (or administrator rights in Windows) may be needed to perform the command.

Run python setup.py install --nomp to disable OpenMP support. All requirements from Build Threadless Version section apply for this installation option as well.

Run python setup.py install --mpi to enable MPI support. All requirements from Build MPI Version section apply for this installation option as well.

Run python setup.py install --mingw, if you want to use MinGW-w64 on Windows instead of Visual Studio. All requirements from Build with MinGW-w64 on Windows section apply for this installation option as well.

Run python setup.py install --gpu to enable GPU support. All requirements from Build GPU Version section apply for this installation option as well. To pass additional options to CMake use the following syntax: python setup.py install --gpu --opencl-include-dir=/usr/local/cuda/include/, see Build GPU Version section for the complete list of them.

Run python setup.py install --cuda to enable CUDA support. All requirements from Build CUDA Version section apply for this installation option as well.

Run python setup.py install --hdfs to enable HDFS support. All requirements from Build HDFS Version section apply for this installation option as well.

Run python setup.py install --bit32, if you want to use 32-bit version. All requirements from Build 32-bit Version with 32-bit Python section apply for this installation option as well.

Run python setup.py install --time-costs, if you want to output time costs for different internal routines. All requirements from Build with Time Costs Output section apply for this installation option as well.

If you get any errors during installation or due to any other reasons, you may want to build dynamic library from sources by any method you prefer (see Installation Guide) and then just run python setup.py install --precompile.

You can use python setup.py bdist_wheel instead of python setup.py install to build wheel file and use it for installation later. This might be useful for systems with restricted or completely without network access.

To install all additional dependencies required for Dask-package, you can append [dask] to LightGBM package name:

pip install lightgbm[dask]

Or replace python setup.py install with pip install -e .[dask] if you are installing the package from source files.

In case you are facing any errors during the installation process, you can examine $HOME/LightGBM_compilation.log file, in which all operations are logged, to get more details about occurred problem. Also, please attach this file to the issue on GitHub to help faster indicate the cause of the error.

Refer to FAQ.

Refer to the walk through examples in Python guide folder.

The code style of Python-package follows PEP 8. If you would like to make a contribution and not familiar with PEP 8, please check the PEP 8 style guide first. Otherwise, the check won't pass. Only E501 (line too long) and W503 (line break occurred before a binary operator) can be ignored.

Documentation strings (docstrings) are written in the NumPy style.