Globally Induced Forest is a python package to build lightweight yet accurate decision forests. The core tree implementation is based on scikit-learn 0.18. The provided estimator are scikit-learn compatible.
If you use this package, please cite
Begon, J. M., Joly, A., & Geurts, P. (2017, July). Globally Induced Forest: A Prepruning Compression Scheme.
In International Conference on Machine Learning (pp. 420-428).
The paper is avaiblable at http://orbi.ulg.ac.be/handle/2268/214279.
The framework was tested under python 2.7 and 3.5 with the following dependencies:
- six 1.10.0
- numpy 1.11.0
- scipy 0.17.1
- scikit-learn 0.18.1
You may use pip install -r requirements.txt to install all requirements.
To install this package, clone the repository locally and use
python setup.py install
The examples folder contains a couple of snipets demonstrating the use of GIFClassifier and GIFRegressor.