Table 1.
Performance of four classifiers using the training dataset.
| Feature selection + classifier | Feature type | Feature number | AUC | Average accuracy | Average AUC |
|---|---|---|---|---|---|
| ANOVA + SVM | GE | 56 | 0.9962 | 0.7553 | 0.8446 |
| CNA | 30 | 0.6586 | 0.5937 | 0.5159 | |
| ANOVA + naïve bayes | GE | 46 | 0.9299 | 0.6755 | 0.8291 |
| CNA | 24 | 0.6019 | 0.5234 | 0.5506 | |
| ANOVA + logistic regression | GE | 44 | 0.9703 | 0.7059 | 0.6053 |
| CNA | 15 | 0.6782 | 0.6135 | 0.5699 | |
| Xgboost | GE | 64 | 0.9602 | 0.7338 | 0.8025 |
| CNA | 30 | 0.954 | 0.6559 | 0.6317 |
GE, gene expression; CAN, copy number alteration; ANOVA, analysis of variance; SVM, support vector machine; AUC, area under the curve; Average accuracy, average of the accuracies from 10-fold cross-validation. Average AUC, average of the AUC values from 10-fold cross-validation.