gxgboost

Lightweight GAUSS wrapper for the popular XGBoost library.

XGBoost is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable. It implements machine learning algorithms under the Gradient Boosting framework.

Installation

$ git clone --recursive https://github.com/aptech/gxgboost
$ cd gxgboost
$ mkdir build
$ cd build

GAUSS Package manager

The GAUSS gxgboost library can be installed and updated directly in GAUSS using the GAUSS package manager.

Windows

$ cmake -G"NMake Makefiles" -DCMAKE_BUILD_TYPE=Release ..
$ nmake install OR jom install

Linux / macOS

$ cmake -G"Unix Makefiles" -DCMAKE_BUILD_TYPE=Release ..
$ make -jN install

GAUSS application installer

The gxgboost.zip release folder can be easily installed using the GAUSS Application Installer, as shown below:

1. Download the zipped folder gxgboost.zip from the gxgoost release page.
2. Select Tools > Install Application from the main GAUSS menu.
3. Follow the installer prompts, making sure to navigate to the downloaded gxgboost.zip.
4. Before using the functions created by gxgboost you will need to load the newly created gxgboost library. This can be done in a number of ways:

• Navigate to the library tool view window and click the small wrench located next to the gxgboost library. Select Load Library.
  
• Enter library gxgboost in the program input/output window.
• Put the line library gxgboost; at the beginning of your program files.

GAUSS package installer CLI

In addition, you can also use the GAUSS package installer CLI (gpkg(.exe)) in the GAUSS installation directory:

• gpkg install gxgboost.zip

Requirements

OpenMP must be installed on the target system to successfully compile or use the gxgboost library.

Windows

Install the Visual C++ Redistributable for Visual Studio 2015 if it is not already installed.

Linux

Built against GCC OpenMP (libgomp1)

# Ubuntu
$ sudo apt install libgomp1

# RHEL/CentOS
$ sudo yum install libgomp

macOS

LLVM's OpenMP runtime library:

# Homebrew
$ brew install libomp

Usage

Refer to <data/gxgboost.e> for a full example and data/gxgboost.sdf for full description of parameters.

Setup

Initialize a control structure for the boosting method of your choice.

Train

The following code samples are all verbose to showcase utilizing their associated control structures. Providing a control structure is not necessary if you wish to use the default values. These can all be referenced in <data/gxgboost.sdf>

Tree Booster

library gxgboost;

/*
** Declare instance of the
** xgbTree control structure
*/
struct xgbTree ctl;

/*
** Initialize control structure with
** default values
*/
ctl = xgbCreateCtl("tree");

// Declare instance of xgbModel structure
struct xgbModel model;

// Call xgbTreeFit
model = xgbTreeFit(labels, train_data, ctl);

Dart Boooster

library gxgboost;

/*
** Declare instance of the
** xgbDart control structure
*/
struct xgbDart ctl;

/*
** Initialize control structure with
** default values
*/
ctl = xgbCreateCtl("dart");

// Declare instance of xgbModel structure
struct xgbModel model;

// Call xgbDartFit
model = xgbDartFit(labels, train_data, ctl);

Linear Booster

library gxgboost;

/*
** Declare instance of the
** xgbLinear control structure
*/
struct xgbLinear ctl;

/*
** Initialize control structure with
** default values
*/
ctl = xgbCreateCtl("linear");

// Declare instance of xgbModel structure
struct xgbModel model;

// Call xgbLinearFit
model = xgbLinearFit(labels, train_data, ctl);

Predict

pred = xgbPredict(model, test_data);

License

---

© Contributors, 2018. Licensed under an Apache-2 license.