Computational and Mathematical Methods in Medicine / 2015 / Article / Tab 5

Research Article

Binary Matrix Shuffling Filter for Feature Selection in Neuronal Morphology Classification

Table 5

Breakdown of independent tests results of different models (%).

ClassifierFS methodPyramidalMotoneuronSensoryTripolarBipolarMultipolarPurkinje

NBAll 30.96 ± 1.96 18.24 ± 3.1241.62 ± 5.4761.26 ± 7.9094.16 ± 4.7498.34 ± 3.7196.00 ± 8.94
SVM-RFE29.22 ± 16.422.31 ± 3.2629.80 ± 60.556.25 ± 9.6692.50 ± 6.1888.33 ± 21.7396.0 ± 8.94
Rough set52.38 ± 4.4822.32 ± 3.0839.20 ± 3.9597.5 ± 2.794.16 ± 6.9885.0 ± 10.8796.0 ± 8.94
BMSF77.26 ± 7.6725.38 ± 3.7338.93 ± 4.6360.83 ± 4.090.83 ± 5.4351.67 ± 21.5792.0 ± 17.89

BPNNAll 99.10 ± 0.7582.46 ± 9.7857.84 ± 19.6442.94 ± 11.760.00 ± 0.000.00 ± 0.0052.00 ± 48.17
SVM-RFE99.12 ± 0.3683.22 ± 18.4445.24 ± 23.5962.50 ± 9.6415.84 ± 35.420.00 ± 0.0080.0 ± 34.61
Rough set99.08 ± 0.3678.92 ± 4.3771.80 ± 3.0957.06 ± 12.10.00 ± 0.000.00 ± 0.0076.0 ± 8.94
BMSF98.42 ± 1.0872.00 ± 6.0166.16 ± 16.6460.0 ± 18.4714.16 ± 31.670.00 ± 0.0076.0 ± 43.36

SVCAll 99.56 ± 0.1882.46 ± 6.9593.69 ± 5.2387.50 ± 6.0797.5 ± 2.2818.33 ± 17.0796.0 ± 8.94
SVM-RFE99.55 ± 0.1365.38 ± 4.5869.66 ± 15.6469.58 ± 7.0072.5 ± 31.260.00 ± 0.0088.0 ± 17.89
Rough set99.52 ± 0.1177.54 ± 5.5854.23 ± 15.0967.08 ± 7.7189.17 ± 5.590.00 ± 0.0092.0 ± 10.95
BMSF99.63 ± 0.1492.46 ± 4.995.84 ± 1.1083.33 ± 7.3799.17 ± 1.861.67 ± 3.7392.0 ± 17.89