Boosting and ensemble learning library in Python.

Algorithms supported:

• Classification and regression trees (work in progress)
• Random forests (work in progress)
• Gradient Boosting
• McRank
• LambdaMART

ivalice follows the scikit-learn API conventions. Computationally demanding parts are implemented using Numba.

ivalice needs Python >= 2.7, setuptools, Numpy >= 1.3, SciPy >= 0.7, scikit-learn >= 0.15.1 and Numba >= 0.13.4.

To run the tests you will also need nose >= 0.10.

To install ivalice from pip, type:

pip install https://github.com/mblondel/ivalice/archive/master.zip

To install ivalice from source, type:

git clone https://github.com/mblondel/ivalice.git
cd ivalice
sudo python setup.py install

https://github.com/mblondel/ivalice

Mathieu Blondel, 2014-present