Create feature files about ehg-oversampling HOT 2 CLOSED

Yes, I have seem some filtering in some of the papers, but they mostly mention those already applied to the data. I guess other databases might lack these or use different filtering, this should definitely be synchronized. Probably we could have separate preprocessing steps for the features implemented from the different papers. And also synchronize across datasets.

from ehg-oversampling.

Yes, I have seem some filtering in some of the papers, but they mostly mention those already applied to the data. I guess other databases might lack these or use different filtering, this should definitely be synchronized. Probably we could have separate preprocessing steps for the features implemented from the different papers. And also synchronize across datasets.

Exactly, for the icelandic dataset, we will have to apply the butterworth filtering ourselves and create bipolar signals from that (I already got that in code somewhere else).

Different preprocessing steps for different features would be nice, but will of course make the code a lot more complex.

Small note: extraction all features from one signal (one channel) currently takes 72.90 seconds, so it will take 900*73 seconds or roughly 18-19 hours to extract all of them. This is not too bad, but no good news for the Icelandic dataset (which has 10 times as much measurements due to a higher frequency) and 16 channels...

from ehg-oversampling.