Skip to main content

Table 15 Weighted accuracies of various models constructed on the support set of 231 genes identified by van 't Veer and coworkers

From: Breast cancer prognosis by combinatorial analysis of gene expression data

Method

Support set of 231 genes (van 't Veer [33])

 

Training set (78 cases)

Test set (19 cases)

Entire dataset (78 + 19 cases)

 

Direct classification (%)

Cross-validation (%)

Direct classification (%)

Cross-validation (%)

Artificial neural networks (1 hidden layer)

100.00

72.24

73.68

73.96

Support vector machines (linear kernel)

100.00

72.79

73.68

74.88

Logistic regression

100.00

71.21

73.68

75.63

Nearest neighbors

100.00

72.94

78.94

77.15

Decision trees (C4.5)

97.44

60.70

73.68

66.64

95% CI

98.48–100.00

65.39–74.56

72.67–76.79

70.07–77.24

  1. The 70-gene set was reported by van 't Veer and coworkers elsewhere [4]. CI, confidence interval; LAD, logical analysis of data.