Mixed-frequency time-frequency canonical correlation analysis (MF-TFCCA) is a method for identifying causal relationships between time series of different temporal resolutions.
This repository contains the MATLAB and Python codes for the paper: Inferring directed spectral information flow between mixed-frequency time series
Data: Data samples.MFTFCCA_MATLAB/src: MATLAB code for MF-TFCCA and related functions.MFTFCCA_MATLAB/demo: MATLAB scripts for running a simulation demo.MFTFCCA_Python/src: Python code for MF-TFCCA and related functions.MFTFCCA_Python/demo: Python scripts for running a simulation demo.
The MF-TFCCA package requires only a standard computer with enough RAM to support the in-memory operations. The runtimes are tested on a computer with the following specs: RAM: 16GB; CPU: 6 cores@2.38 GHz
The package is tested on Windows 10 operating system. The package should work on other operating systems as well.
Before setting up the package, please ensure that you have MATLAB installed on your system. The package is tested on MATLAB R2021a.
The following MATLAB toolboxes are required to run the code:
- The Multivariate Granger Causality (MVGC) Toolbox for generating VAR systems and estimating Granger causality. (version mvgc_v1.3)
This package can be installed using the given url and will install in within 30 seconds.
- MFVAR_toolbox (included in this repository) for estimating Granger causality in mixed-frequency VAR systems.
(For MATLAB) The package can be used directly by installing the required toolboxes and adding the src folder and its subfolders to the MATLAB path.
Please refer to demo/demo_VAR.m for an example of how to use the code. Each block of the code should take less than 1 minute to run on a standard computer using the given data samples.
For usage of the package and associated manuscript, please cite according to the enclosed citation.bib.