Table 2 Models used and test results to predict mortality among Brazilians aged 50 and over
Algorithms | AUC | AUC-PR | Accuracy | Precision | Recall | F1-Score | Specificity |
|---|---|---|---|---|---|---|---|
Logistic Regression | 0.78 0.75–0.80 | 0.30 | 0.71 | 0.22 | 0.74 | 0.34 | 0.70 |
Decision Tree | 0.71 0.68–0.74 | 0.22 | 0.72 | 0.21 | 0.61 | 0.31 | 0.72 |
Random Forest | 0.92 0.90–0.94 | 0.75 | 0.76 | 0.29 | 0.89 | 0.43 | 0.75 |
Gradient Boosting | 0.82 0.79–0.84 | 0.36 | 0.73 | 0.24 | 0.79 | 0.37 | 0.72 |
Support Vector Machine (SVM) | 0.84 0.81–0.86 | 0.37 | 0.73 | 0.24 | 0.80 | 0.37 | 0.72 |
K-Nearest Neighbors (KNN) | 0.91 0.89–0.93 | 0.79 | 0.75 | 0.27 | 0.88 | 0.41 | 0.73 |
XGBoost | 0.82 0.88–0.92 | 0.55 | 0.72 | 0.23 | 0.76 | 0.36 | 0.71 |
LightGBM | 0.81 0.78–0.83 | 0.35 | 0.72 | 0.23 | 0.74 | 0.35 | 0.72 |
CatBoost | 0.80 0.77–0.82 | 0.35 | 0.73 | 0.23 | 0.70 | 0.35 | 0.73 |
