Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.



princomp interpretation
Posted:
Oct 4, 2011 3:17 AM


Hi. I've a data set comprised of 32 channels. So I've a matrix of 32 x 100000 where 32 is the channels and 100000 are the data collected for each channel.
I'm trying to use principal component method to reduce the number of channels for my subsequent analysis.
I did this [pc, scores, variance, t] = princomp (mydata);
From there, how do I decide which of the 32 channels do not contain enough information for me to spend time analyzing? Do I read the pc or scores or variance, or t value? I read the manual but it does not make much sense to me. And I'm running out of time to slowly dig through the maths behind it. Is there a proper way to decide which channel has enough information based on the info. I gather from the princomp?
Many thanks. Marc



