Table 3.
Accuracy and recall of each neural network model and logistic regression with a cutoff of >0.5 for the dichotomous classification.
| Models | Accuracy | Recall | |
| Neural network model | |||
| No undersampling | 0.60 | 0.11 | |
| Sampling ratio 1:2 | 0.72 | 0.37 | |
| Sampling ratio 1:4 | 0.65 | 0.37 | |
| Sampling ratio 1:8 | 0.70 | 0.52 | |
| Logistic regression | |||
| No undersampling | 0.58 | 0.05 | |
| Sampling ratio 1:2 | 0.65 | 0.21 | |
| Sampling ratio 1:4 | 0.67 | 0.26 | |
| Sampling ratio 1:8 | 0.81 | 0.68 | |