Skip to main content
. 2022 Mar 31;3(5):100471. doi: 10.1016/j.patter.2022.100471

Table 2.

Out of sample RF, the percentage of misclassification for the binary-classification datasets

SPECTF [80 × 44; 187 × 44] Spambase [4,601 × 57] Relathe [1,427 × 4,322] Ovarian cancer [72 × 592]
GSO 24.60 (23) 4.37 ± 0.87 (22) 14.25 ± 2.66 (28) 27.78 (3)
mRMR Peng 24.60 (20) 4.39 ± 0.57 (29) 16.48 ± 3.26 (28) 26.39 (15)
mRMR Spearman 23.53 (18) 4.78 ± 1.67 (27) 20.14 ± 3.56 (27) 20.83 (15)
Information gain 24.06 (16) 4.65 ± 0.59 (29) 20.63 ± 3.93 (23) 27.78 (4)
RELIEF 22.99 (27) 5.61 ± 1.00 (24) 29.30 ± 3.99 (27) 29.17 (7)
CFS 25.13 (12) 5.85 ± 0.53 (17) 17.18 ± 2.33 (30) 29.17 (3)
CBF 28.34 (2) 5.07 ± 0.48 (18) 16.06 ± 2.30 (27) 31.94 (3)
SIMBA 25.13 (25) 4.89 ± 0.83 (25) 32.25 ± 3.23 (26) 37.50 (20)
LOGO 25.67 (21) 4.87 ± 0.97 (28) 23.66 ± 4.24 (29) 27.78 (4)
L1-LSMI 22.46 (10) 4.33 ± 1.10 (17) 27.54 ± 2.24 (25) 26.39 (5)
IAMB 28.88 (6) 4.80 ± 0.89 (24) 14.73 ± 2.31 (30) 27.78 (3)
HITON 24.06 (22) 4.26 ± 0.66 (28) 28.24 ± 5.26 (23) 29.17 (3)
JMI 24.06 (30) 4.63 ± 0.58 (30) 20.70 ± 3.36 (16) 18.06 (14)
DISR 24.06 (22) 5.00 ± 0.83 (29) 19.86 ± 2.69 (21) 19.44 (4)
QPFS 24.60 (25) 4.37 ± 0.74 (28) 20.21 ± 1.97 (23) 23.61 (12)
CMIM 24.06 (15) 4.61 ± 0.88 (27) 16.06 ± 2.95 (30) 16.67 (24)
CIFE 23.53 (15) 5.89 ± 1.61 (30) 16.83 ± 3.26 (27) 29.17 (11)
MIQ 24.06 (16) 4.83 ± 0.84 (19) 18.52 ± 3.17 (28) 22.22 (15)
SPECCMI 24.06 (27) 4.61 ± 1.14 (28) 20.14 ± 4.67 (29) 19.44 (14)
RRCT 22.99 (18) 4.72 ± 0.56 (29) 13.87 ± 2.48 (27) 13.89 (20)

The design matrices are summarized in the form N×M [number of samples × number of features]. When two design matrices are mentioned, the first was used for training (selecting features and training the statistical learner) and the second for testing performance. The presented results are the percentage of misclassification values, and the number in parentheses is the number of features that gave best-performance results searching for results with 1 … min(M,30) features (see Figure 2 for details). When the assessment was done using 10-fold cross validation, the results are presented in the form mean ± SD.