mud-pod: A Multivariate Unimodality Test

This package offers tools to analyze the unimodality of data sampled from multivariate distributions lying in the Euclidean Space. To read an independent explanation and summary of this paper, please refer to the write-up by AIModels.fyi here.

Features

• The mud-pod test: A multivariate unimodality test.
• The dip-means clustering algorithm: A wrapper of k-means that also detects the numbers of clusters.

Installation

To install mudpod, you can use pdm, which is a modern packaging tool that manages your Python packages without the need for creating a virtualenv in a traditional sense.

Prerequisites

Ensure you have pdm installed on your system. If not, install it using the following command:

curl -sSL https://pdm-project.org/install-pdm.py | python3 -

Project Setup

Please run:

pdm install -G core

Note: If you want to run the tests or the experiments, please install the additional dependencies, i.e., test and exps, respectively, using the following command:

pdm install -G GROUP_NAME

References

If you find this code useful in your research, please cite:

@inproceedings{
kolyvakis2025a,
title={A Multivariate Unimodality Test Harnessing the Dip Statistic of Mahalanobis Distances Over Random Projections},
author={Prodromos Kolyvakis and Aristidis Likas},
booktitle={The 41st Conference on Uncertainty in Artificial Intelligence},
year={2025},
url={https://openreview.net/forum?id=f3KdHTMabv}
}

License

This project is licensed under the GNU Affero General Public License v3.0 - see the LICENSE file for details.