Table 3 Testing accuracies under different procedures for the acute leukemia data
From: A new regularized least squares support vector regression for gene selection
Binary classes | |||
|---|---|---|---|
Procedures | Classifier | No. of genes | Accuracy |
A: Proposed selection and criterion | |||
RLS-SVR | Â | Â | Â |
δ j ≥ 1/q | +KFDA | 4 | 0.9412 |
| Â | +SVM | 4 | 0.9412 |
∑ δ j ≥ 80% | +KFDA | 7 | 1 |
| Â | +SVM | 7 | 1 |
q genes | +KFDA | 10 | 1 |
| Â | +SVM | 10 | 1 |
B: Other selection procedures | |||
BVS | +KFDA | 5 | 0.9706 |
| Â | +SVM | 5 | 0.9706 |
BMA | +KFDA | 20 | 1 |
| Â | +SVM | 20 | 1 |
SGS1 | +KFDA | 10 | 0.9118 |
| Â | +SVM | 10 | 0.9118 |
SGS2 | +KFDA | 10 | 0.9412 |
| Â | +SVM | 10 | 0.9412 |
C: Selection and classification together | |||
| Â | IFFS | 14 | 1 |
| Â | SVM-RFE | 8 | 1 |
| Â | BVS | 5 | 0.9706 |
| Â | BMA | 20 | 0.9412 |
Three classes | |||
Procedures | Classifier | No. of genes | Accuracy |
A: Proposed selection and criterion | |||
RLS-SVR | Â | Â | Â |
δ j ≥ 1/q | +KFDA | 5 | 0.7353 |
| Â | +SVM | 5 | 0.9118 |
∑ δ j ≥ 80% | +KFDA | 7 | 1 |
| Â | +SVM | 7 | 1 |
q genes | +KFDA | 10 | 0.9706 |
| Â | +SVM | 10 | 0.9412 |
B: Other selection procedures | |||
BMA | +KFDA | 15 | 0.9706 |
| Â | +SVM | 15 | 0.9706 |
SGS1 | +KFDA | 10 | 0.9118 |
| Â | +SVM | 10 | 0.8529 |
SGS2 | +KFDA | 10 | 0.8824 |
| Â | +SVM | 10 | 0.8529 |
C: Selection and classification together | |||
| Â | IFFS | 23 | 1 |
| Â | BMA | 15 | 0.9706 |