Table 2.
Evaluation of the performance of the four algorithms.
Algorithm | Data set | Area under the curve | Optimal cutoff | Specificity | Sensitivity | Accuracy | Specificity/sensitivity |
Logistic regression | Validation | 0.812 | 0.116 | 0.785 | 0.682 | 0.768 | 1.151 |
Logistic regression | Test | 0.814 | 0.116 | 0.770 | 0.701 | 0.759 | 1.098 |
Random forest | Validation | 0.773 | 0.040 | 0.654 | 0.784 | 0.675 | 0.834 |
Random forest | Test | 0.780 | 0.040 | 0.645 | 0.793 | 0.669 | 0.813 |
Naïve Bayes | Validation | 0.804 | 0.214 | 0.796 | 0.688 | 0.778 | 1.157 |
Naïve Bayes | Test | 0.814 | 0.214 | 0.776 | 0.672 | 0.760 | 1.155 |
XGBoost | Validation | 0.815 | 0.302 | 0.753 | 0.744 | 0.752 | 1.012 |
XGBoost | Test | 0.814 | 0.302 | 0.738 | 0.724 | 0.736 | 1.019 |