moabb icon indicating copy to clipboard operation
moabb copied to clipboard

Published 20 hours ago • verified publisher icon NeuroTechX

BNCI2014009 scaling issues?

Open jsosulski opened this issue 3 years ago • 1 comments
trafficstars

The overall shape of the ERPs for Target / NonTarget seem to be plausible for this dataset. However, the absolute scalings appear to be off.

grafik

Intuitively, I would guess its off by a factor of 10, but I do not know exactly. This does not matter for most classifiers, but can impact preprocessing, e.g. when you reject epochs where minmax exceeds 100 muV or so.

jsosulski avatar Mar 04 '22 07:03 jsosulski

I think it is ok for most classifiers. Did you see any bad scaling factor in the code?

sylvchev avatar Mar 26 '22 14:03 sylvchev