lssfind 0.1.1

Python implementation of LSSFind and LocalLSSFind

LSSFind

Python implementation of LSSFind and LocalLSSFind.

Installation

This package is available at PyPI and can be installed using

pip install lssfind

Usage

The main algorithms of LSSFind and LocalLSSFind are implemented as get_prevalent_interactions and get_sample_interactions respectively. They can be used with Random Forest from scikit-learn.

from lssfind import get_prevalent_interactions, get_sample_interactions
from sklearn.ensemble import RandomForestRegressor
from sklearn.datasets import make_regression
X, y = make_regression(n_features=4, n_informative=2, random_state=0, shuffle=False)
rf = RandomForestRegressor()
rf.fit(X, y)
get_prevalent_interactions(rf, impurity_decrease_threshold=1., min_weight=0.5)
get_sample_interactions(rf, impurity_decrease_threshold=1., testpoints=[[0, 0, 0, 0]], min_weight_dwp=0.5, min_weight_pp=0.25)

Literature

For theoretical background about LSSFind, see

M. Behr, Y. Wang, X. Li, & B. Yu, Provable Boolean interaction recovery from tree ensemble obtained via random forests, Proc. Natl. Acad. Sci. U.S.A. 119 (22) e2118636119, https://doi.org/10.1073/pnas.2118636119 (2022).

Theoretical background about LocalLSSFind can be found in

K. Vuk, N. A. Ihlo, & M. Behr, Provable Recovery of Locally Important Signed Features and Interactions from Random Forest, arXiv, https://arxiv.org/abs/2512.11081 (2025).

Acknowledgement

Part of our implementation is based on Python code which was provided by Yu Wang as part of iterative Random Forest, a project of the group of Bin Yu, see https://github.com/Yu-Group/iterative-Random-Forest/blob/master/irf/utils.py