I’ve using Mike EEG recording to learn his course on applications of the Fourier transform (see the IPython Notebook here:).
I was wondering how to plot correctly the units in the color bar of a spectrogram, and also, how to normalize the amplitude (or power). In the way I do it now, I cannot see clearly the alpha band in the spectrogram.
Hi Jose. Very nice notebook!
You can see that the colorscaling is skewed by that low-frequency transient at around 61 seconds. In general, it’s often best to manually adjust the color scaling. In this particular dataset, a range of [0 10] looks pretty good.
As for normalization, that’s tricky with resting state data, because there isn’t a baseline period to normalize against. Some people take the log of power (10*np.log10(Z)), which helps to reduce the overall range of values and minimizes extreme large values.
