Table 2.

Comparison of mean accuracy (%) of combination of four features and ten classifiers.

Classifier	Feature
	SE	FE	AE	PE	Mean ± SD
AB	73.2 ± 4.4	84.2 ± 3.6	72.9 ± 5.5	65.3 ± 6.1	73.9 ± 8.4
DT	80.6 ± 3.3	89.7 ± 2.9	80.2 ± 4.2	72.7 ± 5.7	80.8 ± 7.3
GP	69.5 ± 5.4	81.7 ± 4.1	69.0 ± 6.4	62.8 ± 6.0	70.8 ± 8.8
LS	66.0 ± 5.0	79.3 ± 4.4	64.8 ± 6.4	57.3 ± 7.2	66.9 ± 9.9
GNB	67.5 ± 6.0	80.5 ± 4.3	66.8 ± 7.1	58.2 ± 7.1	68.3 ± 10.1
KNN	77.3 ± 4.4	85.8 ± 3.4	77.4 ± 5.0	71.5 ± 7.6	78.0 ± 7.4
MLP	67.7 ± 5.5	80.7 ± 4.3	67.0 ± 6.8	58.5 ± 7.2	68.4 ± 10.0
QDA	67.5 ± 6.0	80.5 ± 4.3	66.8 ± 7.1	59.3 ± 7.1	68.5 ± 9.8
RF	85.9 ± 3.1	91.8 ± 2.7	85.9 ± 3.3	79.1 ± 9.3	85.7 ± 7.0
RS	68.3 ± 5.7	81.2 ± 4.1	67.9 ± 6.8	59.2 ± 6.8	69.1 ± 9.8
Mean ± SD	72.3 ± 8.1	83.5 ± 5.6	71.9 ± 9.0	64.4 ± 10.1

Boldface indicates FE + RF is the optimal method.