Table 11 Testing accuracies under different procedures for lung cancer data
From: A new regularized least squares support vector regression for gene selection
Procedures | Classifier | No. of genes | Accuracy | SD |
|---|---|---|---|---|
A: Proposed selection and criterion | ||||
RLS-SVR | Â | Â | Â | Â |
δ j ≥ 1/q | +KFDA | 5 | 0.903 | 0.0082 |
| Â | +SVM | 5 | 0.9051 | 0.0111 |
∑ δ j ≥ 80% | +KFDA | 7 | 0.9179 | 0.0059 |
| Â | +SVM | 7 | 0.9097 | 0.0065 |
q genes | +KFDA | 10 | 0.9222 | 0.009 |
| Â | +SVM | 10 | 0.9071 | 0.0104 |
B: Other selection procedures | ||||
SGS1 | +KFDA | 10 | 0.9005 | 0.0062 |
| Â | +SVM | 10 | 0.9005 | 0.0052 |
SGS2 | +KFDA | 10 | 0.8077 | 0.0164 |
| Â | +SVM | 10 | 0.8513 | 0.0015 |
C: Selection and classification together | ||||
| Â | SGS1 | 98 | 0.938 | n.a. |
| Â | SGS2 | 99 | 0.931 | n.a. |