Table 3 Responsiveness and reliability in predicting length of stay for the 10 models developed using all 15 variables
LOS | ||||||
|---|---|---|---|---|---|---|
| Â | Reliability (Accuracy) | Responsiveness (AUC) | ||||
| Â | Training | Testing | Validation | Training | Testing | Validation |
Random Forest | 91.44% | 60.86% | 61.30% | 0.94 | 0.632 | 0.636 |
Neural Network | 62.81% | 62.84% | 62.79% | 0.662 | 0.661 | 0.668 |
XGT Boost Tree | 61.44% | 61.40% | 61.44% | 0.619 | 0.615 | 0.61 |
XGT Boost linear | 61.44% | 61.40% | 61.44% | 0.603 | 0.6 | 0.595 |
LSVM | 66.64% | 66.84% | 66.55% | 0.689 | 0.689 | 0.684 |
CHAID | 65.54% | 65.41% | 65.63% | 0.665 | 0.665 | 0.663 |
Decision List | 85.57% | 85.39% | 85.44% | 0.59 | 0.593 | 0.59 |
Discriminant | 59.29% | 59.55% | 59.12% | 0.616 | 0.622 | 0.615 |
Logistic Regression | 62.84% | 62.87% | 62.79% | 0.662 | 0.662 | 0.661 |
Bayesian Network | 62.99% | 63.22% | 63.03% | 0.664 | 0.665 | 0.664 |