Table 4: Comparison of various discriminant methods for a thyroid disease database.

Model Apparent error rate
Training sampleTest sample

Multiple-group neural discriminant model ( = 2) 0.0090.017
Multiple-group logistic discrimination with linear effects of covariates 0.0280.041
Multiple-group logistic discrimination with linear + quadratic effects of covariates 0.0190.032
Tree-based model 0.0020.073
2-nearest neighbor 0.0170.074
3-nearest neighbor 0.0380.089
Kernel smoother 0.0540.062
SVM 0.0450.052
Proportional odds model 0.0480.055
VGAM 0.0000.021