Granger Components Analysis

Source code

This repository contains the source code for the paper "Granger Components Analysis: Unsupervised learning of latent temporal dependences". The paper is available on openreview.

MATLAB

A demonstration of how to apply GCA to simulated VAR(3) data is provided:

• as a live script here
  • the output of the live script is here
• as a standard script here

The core function is runGcaTrAlt.m.

Python

Disclaimer: this is a work-in-progress. The MATLAB code has been extensively tested and is the recommended implementation.

A demonstration of how to apply GCA to simulated VAR(3) data is provided as a Jupyter notebook here

The core functions required to implement GCA in Python are gca.py.