Skip to main content

Table 5 P value of statistical comparisons among the five classifiers on the lip images data

From: Computer-assisted lip diagnosis on traditional Chinese medicine using multi-class support vector machines

 Feature selection methods          Classifier algorithm
  A-SVM; B-WSVM; C-MAPLSC; D-Naïve Bayes; E-kNN
Unused B > A(P = 3.2 × 10-7), A > C(P = 2.3 × 10-6), A > D(P = 1.8 × 10-19)
A > E(P = 1.5 × 10-61), B > C(P = 8.5 × 10-21),B > D(P = 1.5 × 10-40)
B > E(P = 2.7 × 10-90), C > D(P = 3.2 × 10-6), C > E(P = 2.2 × 10-36)
D > E(P = 4.5 × 10-17)
SVM-RFE A > B(P = 1.3 × 10-16), A > C(P = 3.5 × 10-26), A > D(P = 7.0 × 10-14)
A > E(P = 5.1 × 10-50), B > C(P = 3.3 × 10-3), B > E(P = 2.6 × 10-13)
D > C(P = 1.4 × 10-4), C > E(P = 6.2 × 10-6), D > E(P = 1.4 × 10-16),
mRMR B > A(P = 5.1 × 10-3), A > C(P = 1.2 × 10-18), A > E(P = 6.0 × 10-28)
B > C(P = 1.5 × 10-28), B > D(P = 3.1 × 10-3), B > E(P = 6.2 × 10-40)
D > C(P = 5.7 × 10-17), C > E(P = 2.5 × 10-2),D > E(P = 1.5 × 10-25)
IG A > C(P = 5.9 × 10-11), A > E(P = 6.2 × 10-19), B > C(P = 1.3 × 10-15)
B > D(P = 1.9 × 10-2), B > E(P = 5.0 × 10-25), D > C(P = 3.1 × 10-9)
C > E(P = 9.3 × 10-3), D > E(P = 1.7 × 10-16)
Total Rank WSVM(11) > SVM(9) > Naïve Bayes(1) > MAPLSC(−6) > kNN(−16)