Skip to main content

Table 2 Comparison of prediction accuracies of various algorithms. Each figure represents the best accuracy obtained by the gene-selection condition shown as the value in the parenthesis. The greatest accuracy of each data set is shown as bold letters. Values in parentheses are numbers of diagnostic genes selected by recursive feature elimination for MC-SVM, shrinkage parameters for SC, and threshold p-value for others.

From: A multi-class predictor based on a probabilistic model: application to gene expression profiling-based diagnosis of thyroid tumors

 

11-SIS

11-SHS

1R-SIS

1R-SHS

1A-SIS

1A-SHS

AA-SHS

MC-SVM

SC

thyroid

79.8 (10-3)

79.8 (10-3)

74.8 (10-3)

74.8 (10-3)

79.8 (10-4)

79.8 (10-4)

79.0 (10-4)

74.8 (2000)

74.8 (0.5)

GCM

86.3 (10-3)

86.3 (10-3)

89.0 (10-5)

89.0 (10-5)

89.0 (10-6)

89.0 (10-6)

90.4 (10-6)

80.8 (32)

74.0 (0)

SRBCT

100

100

100

100

100

100

100

100 (32~2308)

100 (1~2)

esophageal

72.3 (10-1)

72.3 (10-1)

71.6 (10-1)

71.6 (10-1)

73.1 (10-1)

73.8 (10-1)

73.8 (10-1)

71.7 (8)

68.8 (≥ 1)