Table 3

Performance of classification algorithms for the van 't Veer et al. breast cancer genomic and clinical/histopathological data based on 20 trials of 10-fold CV

Algorithm

Accuracy

Sensitivity^*

Specificity^†

PPV^‡

NPV^§

CERP

63.3 (.024)

52.5 (.042)

71.6 (.027)

58.8 (.031)

66.1 (.022)

RF

63.0 (.023)

48.2 (.034)

74.4 (.034)

59.4 (.034)

65.1 (.016)

AdaBoost

61.9 (.045)

38.7 (.090)

79.8 (.065)

59.9 (.085)

62.8 (.034)

SVM

57.4 (.027)

40.3 (.044)

70.7 (.037)

51.5 (.040)

60.5 (.021)

DLDA

62.9 (.017)

52.6 (.025)

70.9 (.027)

58.4 (.023)

66.0 (.013)

SC

62.2 (.018)

53.8 (.025)

68.8 (.018)

57.1 (.022)

65.8 (.016)

CART

54.7 (.031)

21.6 (.096)

80.3 (.063)

44.3 (.103)

57.2 (.022)

CRUISE

57.5 (.047)

24.0 (.100)

83.4 (.063)

51.9 (.120)

58.8 (.032)

QUEST

56.3 (.036)

21.8 (.062)

83.1 (.071)

50.7 (.082)

57.8 (.021)

SD is given in parentheses. *Poor prognosis considered positive. ^†Good prognosis considered negative. ^‡Positive predictive value. ^§Negative predictive value.

Moon et al. Genome Biology 2006 7:R121   doi:10.1186/gb-2006-7-12-r121