Table 3.

RMSE values of SVR models based on average expert score for all EEGers for normalized and non-normalized data types, as well as various sets of inputs to the algorithm. The errors are calculated both for predictions made by peak-centered and shifted signals.

Data type → Normalization Approach ↓	Raw EEG waveforms			All features			Features selected with elastic net regression
	DWT Centered	DWT Shifted	DT-DWT Centered	DWT Centered	DWT Shifted	DT-DWT Centered	DWT Centered	DWT Shifted	DT-DWT Centered
No normalization	0.80	0.85	0.87	0.58	0.59	0.64	0.49	0.49	0.57
Local standardization	0.85	0.88	0.86	0.61	0.61	0.63	0.47	0.49	0.56
Global standardization	0.81	0.83	0.88	0.67	0.68	0.72	0.61	0.60	0.66
Median background RMS normalization	0.80	0.83	0.88	0.67	0.68	0.70	0.63	0.62	0.66