Table 4 Performance evaluation of models employed
From: Predicting hospital-acquired pneumonia among schizophrenic patients: a machine learning approach
Sample | Method | Accuracy(SD) | AUC(SD) | Sensitivity(SD) | Specificity(SD) | Kappa(SD) |
|---|---|---|---|---|---|---|
Train | CART | 0.804(0.089) | 0.851(0.074) | 0.739(0.094) | 0.851(0.106) | 0.597(0.180) |
C5.0 | 0.912(0.033) | 0.971(0.018) | 0.868(0.047) | 0.942(0.030) | 0.819(0.068) | |
KNN | 0.645(0.083) | 0.696(0.066) | 0.628(0.127) | 0.657(0.130) | 0.282(0.162) | |
NB | 0.675(0.095) | 0.798(0.094) | 0.868(0.096) | 0.544(0.098) | 0.376(0.178) | |
RF | 0.917(0.017) | 0.971(0.016) | 0.891(0.048) | 0.937(0.032) | 0.831(0.035) | |
SVM | 0.871(0.030) | 0.936(0.030) | 0.832(0.077) | 0.923(0.043) | 0.733(0.062) | |
LGR | 0.670(0.084) | 0.762(0.083) | 0.621(0.076) | 0.706(0.139) | 0.330(0.160) | |
Test | CART | 0.830 | 0.880 | 0.904 | 0.732 | 0.648 |
C5.0 | 0.945 | 0.993 | 0.989 | 0.887 | 0.887 | |
KNN | 0.667 | 0.701 | 0.745 | 0.563 | 0.312 | |
NB | 0.733 | 0.831 | 0.628 | 0.873 | 0.479 | |
RF | 0.927 | 0.994 | 1.000 | 0.831 | 0.849 | |
SVM | 0.897 | 0.953 | 0.968 | 0.803 | 0.786 | |
LGR | 0.739 | 0.823 | 0.798 | 0.662 | 0.464 |