Skip to main content

Table 5 Comparison of classifiers (ovarian cancer dataset)

From: Classification-based comparison of pre-processing methods for interpretation of mass spectrometry generated clinical datasets

  Ciphergen C Cromwell A
  CM10/ Q10 CM10 Q10 CM10/ Q10 CM10 Q10
DLDA 77 (57–93) 71 (50–93) 71 (50–86) 78 (64–93) 74 (57–93) 70 (43–93)
Naive Bayes 70 (50–93) 66 (43–86) 63 (43–86) 69 (50–86) 68 (50–86) 60 (36–86)
PCDA 73 (50–93) 70 (50–93) 67 (43–86) 77 (57–93) 73 (57–86) 77 (50–93)
SVM 77 (57–93) 71 (50–86) 70 (50–86) 79 (64–93) 77 (57–93) 79 (50–93)
Tree 69 (43–93) 69 (43–93) 64 (36–86) 58 (36–79) 58 (36–79) 52 (29–71)
  1. Average classification accuracy (95% confidence interval) of ovarian cancer versus control on 1000 test sets (size of training sets: 14, size of test sets: 14). Pre-processing was done using setting C for Ciphergen and setting A for Cromwell. Classifiers without feature selection.