Table 3. Classification results on the training and test sets.
| Classification Method | Accuracy | Sensitivity | Specificity |
| Training Set | |||
| Support Vector Machine (SVM) | 90.9% | 91.5% | 90.4% |
| Random Forest (RF) | 86.9% | 87.2% | 86.5% |
| Linear Discriminant Analysis (LDA) | 90.9% | 89.4% | 92.3% |
| Predictive Analysis of Microarray (PAM) | 88.9% | 89.4% | 88.5% |
| Lasso | 91.9% | 91.5% | 92.3% |
| Boosting | 88.9% | 89.4% | 88.5% |
| Naïve Bayes | 88.9% | 89.4% | 88.5% |
| Majority Vote (Combined Classifiers) | 89.9% | 91.5% | 88.5% |
| Test Set | |||
| Support Vector Machine (SVM) | 83.3% | 80.0% | 85.0% |
| Random Forest (RF) | 73.3% | 90.0% | 65.0% |
| Linear Discriminant Analysis (LDA) | 76.7% | 80.0% | 75.0% |
| Predictive Analysis of Microarray (PAM) | 86.7% | 100.0% | 80.0% |
| Lasso | 86.7% | 80.0% | 90.0% |
| Boosting | 86.7% | 90.0% | 85.0% |
| Naïve Bayes | 83.3% | 100.0% | 75.0% |
| Majority Vote (Combined Classifiers) | 80.0% | 90.0% | 75.0% |
The accuracy, sensitivity, specificity of the ileal gene signature selected by the boosting method [16] are calculated using Leaving-One-Out cross validation on the training and subsequently, direct classification of the test set based on the training set.