Skip to main content

Table 4 Comparison of the performance of Support Vector Machine (SVM) classifier with sampling using polynomial, normalized polynomial and puk kernels using complexity parameters 0.1, 10 and 30

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

  Polynomial Normalized Polynomial Puk
C = 0.1 C = 10 C = 30 C = 0.1 C = 10 C = 30 C = 0.1 C = 10 C = 30
Sensitivity 36.18% 36.18% 36.18% 100% 95.10% 65.10% 47.38% 81.94% 80.26%
Specificity 94.37% 94.37% 94.37% 88.31% 88.79% 88.85% 88.58% 94.13% 95.19%
Precision 61.46% 61.41% 61.41% 0.02% 33.67% 5.62% 6.33% 53.64% 62.63%
F-score 45.55% 45.53% 45.53% 0.05% 49.73% 10.35% 11.17% 64.84% 70.36%
RMSE 0.41 0.42 0.42 0.34 0.34 0.34 0.35 0.26 0.25
AUC 0.74 0.74 0.74 0.5 0.52 0.53 0.53 0.76 0.8