Skip to main content

Table 17 Comparison of the performance of the different classification models using the SMOTE sampling methods. The models are: Decision Tree (DT), Support Vector Machine (SVM), Artificial Neural Networks (ANN), Naïve Bayesian Classifier (BC), Bayesian Network (BN), K-Nearest Neighbor (KNN) and Random Forest (RF)

From: Comparison of machine learning techniques to predict all-cause mortality using fitness data: the Henry ford exercIse testing (FIT) project

  DT SVM ANN BC BN KNN RF
Sensitivity 59.95% 80.26% 55.89% 37.41% 60.07% 78.43% 96.07%
Specificity 96.05% 95.19% 90.43% 93.32% 91.02% 96.98% 96.84%
Precision 70.91% 62.63% 24.37% 52.20% 27.32% 77.18% 75.50%
F-score 64.97% 70.36% 33.94% 43.59% 37.56% 77.80% 84.55%
RMSE 0.27 0.25 0.29 0.34 0.28 0.23 0.18
AUC 0.88 0.8 0.82 0.82 0.84 0.88 0.97
  1. The results of this experiment show that the RF model achieves the highest AUC (0.97), the lowest RMSE (0.18) and the highest sensitivity (94.65%)