This is an implementation of the paper Adaptive Bayesian Reticulum by Nuti et al. The model is a binary and multiclass classification tree with soft margins and a novel tree construction method.
To install you can either use conda or pip:
git clone https://github.com/UBS-IB/adaptive-bayesian-reticulum
cd adaptive-bayesian-reticulum
conda build conda.recipe
conda install --use-local adaptive-bayesian-reticulum
git clone https://github.com/UBS-IB/adaptive-bayesian-reticulum
cd adaptive-bayesian-reticulum
pip install -e .
Please see the Demo Scripts for usage examples.
Note that the model is fully compatible with scikit-learn, so you can use it for e.g. cross-validation or performance evaluation with scikit-learn classes and functions.
- Update links to paper once published in a journal