Table 4 Responsiveness and reliability in predicting discharge disposition for the 10 models developed using all 15 variables
Discharge | ||||||
|---|---|---|---|---|---|---|
| Â | Reliability (Accuracy) | Responsiveness (AUC) | ||||
| Â | Training | Testing | Validation | Training | Testing | Validation |
Random Forest | 91.50% | 74.25% | 74.05% | 0.955 | 0.671 | 0.675 |
Neural Network | 75.62% | 75.70% | 75.53% | 0.72 | 0.715 | 0.721 |
XGT Boost Tree | 79.81% | 79.81% | 79.53% | 0.749 | 0.741 | 0.747 |
XGT Boost linear | 79.81% | 79.81% | 79.53% | 0.719 | 0.715 | 0.722 |
LSVM | 80.43% | 80.43% | 80.26% | 0.745 | 0.742 | 0.747 |
CHAID | 80.04% | 80.02% | 79.80% | 0.712 | 0.711 | 0.713 |
Decision List | 89.97% | 90.03% | 89.81% | 0.648 | 0.647 | 0.648 |
Discriminant | 64.49% | 64.28% | 64.35% | 0.693 | 0.694 | 0.694 |
Logistic Regression | 75.50% | 75.63% | 75.44% | 0.716 | 0.713 | 0.718 |
Bayesian Network | 75.14% | 75.46% | 75.13% | 0.713 | 0.71 | 0.715 |