Table 11

Comparison of best classification accuracy for the Lung Cancer dataset

Methods (feature selection + classification)

#Selected genes

#Correctly classified samples (accuracy)

Rule-based classifier


depended degree + decision rules [this work]

1

145 (97.34%)

yes


2

144 (96.64%)


attribute reduction + k-NNs [9]

2

146 (97.99%)

no


PCLs [50]

unknown

146 (97.99%)

yes


C4.5 [50]

1

122 (81.88%)

yes


Bagging [50]

unknown

131 (87.92%)

yes


Boosting [50]

unknown

122 (81.88%)

yes


SVMs [50]

unknown

148 (99.33%)

no


k-NNs [50]

unknown

148 (99.33%)

no


discretization + decision trees [11]

unknown

139 (93.29%)

yes


RCBT [13]

10-40

146 (97.99%)

yes


gene expression ratios [15]

6

148 (99.33%)

no


Wang and Gotoh BMC Medical Genomics 2009 2:64   doi:10.1186/1755-8794-2-64

Open Data