Could anyone advice (or direct to) on how to simulate data where I can expect/ control for time frequency granger causality? for learning purposes
I’ve been simulating time domain granger causality just using simulated ERPs - where one is determined by previous points of the other (as in the ANTS udemy series) - which is quite straight forward but gives clear results, but these don’t map into clear time frequency results
Can anyone advise please - I’ve been following the code for tf granger causality so am pretty sure that’s correct.
Hi Ira. People often simulate GC data using AR models. I show examples of this in code in various places… in my ANTS book, possible also in the MATLAB book, and I think also in my online courses.
Simulating spectral GC is interesting… to be honest, I haven’t thought about it. I suppose it could be done with an AR model that has a longer order, and you specify the AR coefficients to be a wavelet. Sorry I don’t have any plug-and-play code for that but perhaps there is some online somewhere.
I haven’t been able to find anything decent online there is this paper that you cite in your book https://doi.org/10.1016/j.neuroimage.2008.03.025
but I haven’t been able to successfully replicate their findings with simulated data (also they don’t really show spectral gc per se).
but perhaps there is some online somewhere.