Table 4. Statistics of best predictive models generated by 10 fold cross-validation of training dataset.
| Classifier* | CSC J48 | CSC LibSVM | CSC Naïve Bayes | CSC Random Forest |
| Misclassification Cost | 42 | 18 | 4200 | 55.4 |
| Accuracy | 79.50 | 78.25 | 78.25 | 80.61 |
| Sensitivity | 73.03 | 62.92 | 65.17 | 80.90 |
| Specificity | 80.41 | 80.41 | 80.09 | 80.57 |
| BCR$ | 76.72 | 71.67 | 72.63 | 80.73 |
| F-value | 0.47 | 0.42 | 0.42 | 0.51 |
| MCC∧ | 0.40 | 0.33 | 0.34 | 0.46 |
| auROC# | 0.79 | 0.72 | 0.79 | 0.91 |
*CSC denotes Cost Sensitive Classifier; BCR$ is Balanced Classification Rate; MCC∧ is Matthews Correlation Coefficient; auROC# is area under Receiver Operating characteristic curve. Highest numerical value in each row is highlighted as bold.