Table 3.
Train the model with a bias to reduce FN and increase TP value.
| Cognitively Impaired Class (Imbalanced Data) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Model Type | TN | FN | FP | TP | Sensitivity | Specificity | Precision | Accuracy | F-Score | Cost |
| Fine Decision Tree | 159 | 14 | 35 | 42 | 75.00% | 81.96% | 54.55% | 80.40% | 78.33% | 455 |
| Discriminant Linear |
59 | 2 | 135 | 54 | 96.43% | 30.41% | 28.57% | 45.20% | 46.24% | 175 |
| Naïve Bayes Kernel |
150 | 11 | 44 | 45 | 80.36% | 77.32% | 50.56% | 78.00% | 78.81% | 264 |
| SVM Quadratic |
133 | 21 | 61 | 35 | 62.50% | 68.56% | 36.46% | 67.20% | 65.39% | 481 |
| KNN Cubic |
173 | 30 | 21 | 26 | 46.43% | 89.18% | 55.32% | 79.60% | 61.06% | 621 |
| Ensemble RUSBoosted |
158 | 14 | 36 | 42 | 75.00% | 81.44% | 53.85% | 80.00% | 78.09% | 456 |