Granger Causality

[Debanjan Borthakur]

Hi, if the data is stationary, is it recommended to use a sliding window for the calculation of Granger causality? The MVGC toolbox does not use a sliding window. Only in the non-stationary script (https://users.sussex.ac.uk/~lionelb/MVGC/html/mvgc_demo_nonstationary.html) they used windowing. So If I have a trial length of 180 seconds, do you recommend windowing or should I consider the whole trial length for the Granger Causality analysis? This is a motion capture data.

Thanks,

Debanjan

[Mike X Cohen]

180 seconds is quite long. I would estimate the autoregressive models in, e.g., 2-second windows, and then average the results across the windows.

Mike

--
You received this message because you are subscribed to the Google Groups "AnalyzingNeuralTimeSeriesData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to analyzingneuraltimes...@googlegroups.com.
Visit this group at https://groups.google.com/group/analyzingneuraltimeseriesdata.
To view this discussion on the web visit https://groups.google.com/d/msgid/analyzingneuraltimeseriesdata/cc770e57-310b-470e-a281-a19452ccf9a4%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Mike X Cohen, PhD
Fresh look: mikexcohen.com

[Debanjan Borthakur]

Thanks Mike