Table 9.
Performance of the three classifiers for classification of variable (ING).
Dataset B test performance (training on non-variable
IN
and
ING
data) Actual -in rates: 62.4% |
Dataset C performance (training on non-variable IN and ING data) Actual -in rates: 60.3% | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Pred. -in rate | Accuracy | Precision | Recall | F1 score | AUC | Pred. -in rate | Accuracy | Precision | Recall | F1 score | AUC | |
Linear SVM | 64.6% | 69.6% | 0.77 | 0.75 | 2.99 | 0.67 | 61.8% | 69.8% | 0.76 | 0.74 | 2.97 | 0.68 |
Radial SVM | 64.6% | 78.8% | 0.85 | 0.82 | 3.28 | 0.77 | 60.3% | 75.2% | 0.79 | 0.79 | 3.18 | 0.74 |
RF | 70.0% | 75.5% | 0.86 | 0.77 | 3.08 | 0.72 | 64.5% | 76.4% | 0.84 | 0.78 | 3.14 | 0.74 |