Table 3.

Classifiers’ performance for each task, for the training data. Values not in bold are statistically significantly worse than the classifier with highest accuracy (using paired t-test with α = 0.05)

Exp.	Algorithm	CCI	K	F	AUROC
E 1	SMO	85.6 ±7.3	0.69 ±0.16	0.80 ±0.11	0.84 ±0.08
E 1	DTNB	81.6 ±8.2	0.60 ±0.18	0.74 ±0.13	0.88 ±0.07
E 1	NaiveBayes	81.3 +9.5	0.61 ±0.20	0.76 ±0.12	0.88 ±0.08
E 1	J48	80.7 ±9.3	0.59 ±0.20	0.75 ±0.13	0.79 ±0.11
E 2	SMO	83.9 ±7.7	0.66 ±0.17	0.78 ±0.11	0.82 ±0.08
E 2	NaiveBayes	80.3 ±9.3	0.59 ±0.19	0.75 ±0.12	0.87 ±0.09
E 2	DTNB	79.8 ±9.5	0.56 ±0.21	0.72 ±0.15	0.86 ±0.09
E 2	J48	75.4 ±9.5	0.47 ±0.21	0.65 ±0.15	0.73 ±0.12
E 3	SMO	83.8 ±7.7	0.65 ±0.17	0.78 ±0.11	0.82 ±0.09
E 3	J48	76.3 ±9.9	0.49 ±0.22	0.67 ±0.15	0.76 ±0.13
E 3	NaiveBayes	76.2 ±9.9	0.51 ±0.20	0.71 ±0.13	0.85 ±0.09
E 3	DTNB	75.7 ±9.0	0.48 ±0.19	0.67 ±0.13	0.81 ±0.10
E 4	SMO	81.3 ±8.2	0.52 ±0.21	0.64 ±0.17	0.75 ±0.11
E 4	J48	74.4 ±8.8	0.32 ±0.24	0.47 ±0.21	0.67 ±0.15
E 4	DTNB	73.5 ±10.0	0.34 ±0.24	0.51 ±0.19	0.76 ±0.12
E 4	NaiveBayes	72.8 ±9.9	0.37 ±0.23	0.56 ±0.18	0.77 ±0.11
E 5	NaiveBayes	67.2 ±12.1	0.33 ±0.25	0.62 ±0.15	0.72 ±0.14
E 5	SMO	66.8 ±10.7	0.31 ±0.22	0.55 ±0.16	0.65 ±0.11
E 5	J48	63.6 ±10.1	0.26 ±0.21	0.56 ±0.15	0.62 ±0.13
E 5	DTNB	62.1 ±11.9	0.22 ±0.24	0.54 ±0.16	0.64 ±0.14