LightGBM

LightGBM is a gradient boosting framework that was developed by Microsoft that uses the tree-based learning algorithm in a different fashion than other GBMs, favoring exploration of more promising leaves (leaf-wise) instead of developing level-wise.

In graph terminology, LightGBM is pursuing a depth-first search strategy than a breadth-first search one.

It has been designed to be distributed (Parallel and GPU learning supported), and its unique approach really achieves faster training speed with lower memory usage (thus allowing for the handling of the larger scale of data):

Website: https://github.com/Microsoft/LightGBM
Version at the time of print: 2.1.0

The installation of XGBoost requires some more actions on your side than usual Python packages. If you are operating on a Windows system, open a shell and issue the following commands:

$> git clone --recursive https://github.com/Microsoft/LightGBM
$> cd LightGBM
$> mkdir build
$> cd build
$> cmake -G "MinGW Makefiles" ..
$> mingw32-make.exe -j4

You may need to install CMake on your system first ( https://cmake.org), and you also may need to run cmake -G "MinGW Makefiles" .. if a sh.exe was found in your PATH error is reported.

If you are instead operating on a Linux system, you just need to digit on a shell:

$> git clone --recursive https://github.com/Microsoft/LightGBM
$> cd LightGBM
$> mkdir build
$> cd build
$> cmake ..
$> make -j4

After you have completed compiling the package, no matter whether you are on Windows or Linux, you just import it on your Python command line:

import lightgbm as lgbm

You can also build the package using MPI for parallel computing architectures, HDFS, or GPU versions. You can find all the detailed instructions at https://github.com/Microsoft/LightGBM/blob/master/docs/Installation-Guide.rst.

官术网_书友最值得收藏!

Python Data Science Essentials

LightGBM