[Eeglablist] Problem running ICA: rank of data lower than the number of input data channels

[Makoto Miyakoshi]

Dear Caro,

> I should mention that we had to reconstruct two channels before the
analysis due to equipment problems. Could that account for part of the
problem?

Yes that's it.

> I must also insist on the fact that even when chennels were reconstructed
for other subjects, I son-t always get the message.

Because it also depends on data quality. From my experience I would expect
that if you have a lot of EMGs you would see that message more often.

> A second question I have is, should I edit the rank to 30 or should I
just run it with 28 components? how would that affect my data?

Just run it with 28 components. Using PCA should be fine after channel
interpolation, but why don't you go simple if you can.

I have seen and experienced rank issues. I would avoid it whenever I can.
Try matlab function rank on your fully processed data. It often returns
much less number than your number of channels.

Makoto


2013/9/11 Caro Gattei <carogattei at gmail.com>

> Dear all,
>
> I am currently running an experiment with a 32 channel equipment (30+ ext)
> and when I try to run ICA in specific subjects I get a message saying that
> "EEGLAB has detected that the rank of your data matrix is lower the number
> of input data channels. This might be because you are including a reference
> channel or because you are running a second ICA decomposition. The proposed
> dimension for ICA is 28 (out of 30 channels) rana computation may be
> innacurate so you may edit this number below. If you do not understand,
> simply press OK"
>
> I found an explanation for the error message on the eegab list that said
> that the ranking could be reduced de to low-pass filtering (being that I do
> not have a reference  channel and hand't run ICA previously:
>  http://sccn.ucsd.edu/pipermail/eeglablist/2012/005110.html
>
> However, I get the same message when I try to run ICA with raw data.
>
> I should mention that we had to reconstruct two channels before the
> analysis due to equipment problems. Could that account for part of the
> problem?
>
> I must also insist on the fact that even when chennels were reconstructed
> for other subjects, I son-t always get the message.
>
> A second question I have is, should I edit the rank to 30 or should I just
> run it with 28 components? how would that affect my data?
>
> Best regards,
>
> Caro
>
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>



-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20130911/472da772/attachment.html>