Table 4 Hyper-parameter of feature selection
From: A novel EEG-based major depressive disorder detection framework with two-stage feature selection
Feature selection | Method | \(\tau\) | \(\zeta\) | F1 | R2 |
|---|---|---|---|---|---|
– | SVM | – | 154 | 0.4510 | – |
SVM-DF | – | 192 | 0.5144 | – | |
LNR | – | – | – | – | |
LNR-DF | – | – | – | – | |
PCC | SVM | 0.055 | – | 0.5306 | – |
SVM-DF | 0.055 | – | 0.5766 | – | |
LNR | 0.210 | – | – | 0.0644 | |
LNR-DF | 0.185 | – | – | 0.0927 | |
RFE | SVM | – | 43 | 0.9385 | – |
SVM-DF | – | 53 | 0.9667 | – | |
LNR | – | 53 | – | 0.8474 | |
LNR-DF | – | 66 | – | 0.8962 | |
PCC and RFE | SVM | 0.040 | 38 | 0.9581 | – |
SVM-DF | 0.015 | 36 | 0.9846 | – | |
LNR | 0.010 | 59 | – | 0.8812 | |
LNR-DF | 0.010 | 63 | – | 0.9479 |