Coenjunction

Mutual information and transfer entropy estimation using copula entropy.

Academic foundation

This repository's mutual-information estimator was inspired by the paper bundled in this repo:

• Yves-Laurent Kom Samo, Inductive Mutual Information Estimation: A Convex Maximum-Entropy Copula Approach (MIND), arXiv:2102.13182v3.
• Local copy: 2102.13182v3.pdf

If you use this implementation in academic work, please cite that paper.

@article{komsamo2021mind,
  title   = {Inductive Mutual Information Estimation: A Convex Maximum-Entropy Copula Approach},
  author  = {Kom Samo, Yves-Laurent},
  journal = {arXiv preprint arXiv:2102.13182},
  year    = {2021},
  url     = {https://arxiv.org/abs/2102.13182}
}

Project layout

• coenjunction/: reusable Python modules.
• notebooks/: exploratory notebooks (*.ipynb).

Main APIs

• coenjunction.CopulaEntropyEstimator
• coenjunction.estimate_mi_from_ce
• coenjunction.calculate_transfer_entropy
• coenjunction.calculate_transfer_entropy_with_edge_lag