Table 4 Performance of each method on downsampled data. The boldfaced numbers indicate the ones which outperformed the others
From: Feature selection using distributions of orthogonal PLS regression vectors in spectral data
Method | MSE | Q2 | P | N | |
|---|---|---|---|---|---|
OPLSRa | αf=0.01 | 10,972 | 0.83 | 0.883 | 8.90 |
|  | αf=0.05 | 9,105 | 0.87 | 0.777 | 24.4 |
|  | αf=0.10 | 5,709 | 0.91 | 0.658 | 22.8 |
OPLSRb | αf=0.01 | 10,133 | 0.85 | 0.884 | 10.4 |
|  | αf=0.05 | 8,811 | 0.89 | 0.881 | 31.0 |
|  | αf=0.10 | 7,782 | 0.93 | 0.848 | 48.2 |
FDR | q=0.01 | 23,116 | 0.65 | 0.701 | 6.0 |
| Â | q=0.05 | 10,330 | 0.85 | 0.685 | 39.5 |
| Â | q=0.10 | 13,194 | 0.85 | 0.554 | 40.0 |
Lasso | λ=0.01 | 6,235 | 0.90 | 0.502 | 6.3 |
|  | λ=0.10 | 9,886 | 0.88 | 0.494 | 3.2 |
|  | λ=0.50 | 13,124 | 0.80 | 0.498 | 2.1 |