sandy4321/trot

A python implementation of discrete optimal transport with a Tsallis entropy regularization.

TROT

This is a Python implementation of Tsallis Regularized Optimal Transport (TROT) for ecological inference, following

Boris Muzellec, Richard Nock, Giorgio Patrini, Frank Nielsen. Tsallis Regularized Optimal Transport and Ecological Inference. arXiv:1609.04495

It contains both scripts implementing algorithms for solving TROT, and notebooks which reproduce the ecological inference pipeline from the article.

Dependencies

numpy, scipy, pickle

Usage

To run the Ecological Inference notebook, you will first want to download the Florida dataset (600 MB):

wget https://www.dropbox.com/s/pvxqi8hzcf4fshr/Fl_Data.csv

and put it in the root folder of the repo.

You can then run Notebooks/Ecological/Inference.ipynb for a reproduction of the article's ecological inference pipeline, and Notebooks/Tsallis/Plots.ipynb for a visualization of the impact of parameter $q$ and $\lambda$.

The code under Trot/ contains the basics for building a TROT-based application.