Bugzilla – Bug 661
timef.m: absolute power for wavelet and FFT
Last modified: 2015-02-20 13:29:25 PST
Yes, test with newtimef. For wavelets use wavelet factor of 0 or 1 (same as FFT) and make sure you have the same window size for both FFT and wavelets.
We don't use this function anymore. Should we test this with newtimef??
Yes, still a problem. We should email the eeglablist.
Message for external help posted on wiki web site.
Still a problem as of October 2009
This is true when we look at the atachement. When plotting the difference from baseline, the scale is similar but when plotting the absolute value, it is very different.
Created attachment 388 [details]
Different scaling (right=fft, left=wavelet)
Dear eeglablist subscribers,
I have a question concerning the timef function.
As I want to calculate the absolute power of my data, I set the 'baseline'
parameter to 'NaN'.
Now, first thing that concerns me, I get totally different results using
either FFT or wavelets (freqs from 2-50Hz, FFT yielded values from -25 to
10, wavelet values from 36 to 48).
Besides this, even after reading carefully through previous eeglablist
discussions on this topic, I'm still not shure how to calculate the mean
power (in uV) from the values returned by the function.
I'd appreciate any help, thanks in advance,