Distance Variance Score: An Efficient Feature Selection Method in Text Classification

Table 2

Accuracy of classifiers (unit for mean and std: %).

(a) DBWorld data set

	LS				DVS
	DT(C5.0)		SVM		DT(C5.0)		SVM
	Mean std	Sig.	Mean std	Sig.	Mean std	Sig.	Mean std	Sig.
3721	91.30 7.54	—	85.73 6.73	—	91.30 7.54	—	85.73 6.73	—
2721	82.64 8.17	.004	84.14 7.06	.505	89.50 6.59	.736	85.22 6.10	.807
1721	70.27 8.22	.000	80.73 6.68	.023	89.61 5.57	.779	87.02 8.92	.291
721	58.53 10.52	.000	63.97 11.09	.000	91.74 6.07	.433	84.15 8.67	.129
521	56.49 6.94	.000	62.66 9.16	.000	89.07 7.08	.185	85.10 7.57	.584
321	53.58 10.79	.000	57.75 8.83	.000	92.16 4.58	.454	83.55 7.19	.192
121	52.44 10.34	.000	51.92 10.91	.000	92.74 6.61	.179	90.56 6.16	.098
21	58.03 6.73	.000	57.38 9.53	.000	88.26 7.68	.332	91.40 6.41	.722

(b) CNAE data set

	LS				DVS
	DT(C5.0)		SVM		DT(C5.0)		SVM
	Mean std	Sig.	Mean std	Sig.	Mean std	Sig.	Mean std	Sig.
856	87.59 2.25	—	92.93 1.58	—	87.59 2.25	—	92.93 1.58	—
656	86.67 2.21	.910	82.19 4.52	.000	87.64 1.77	.924	83.41 2.02	.000
456	86.87 2.62	.394	84.42 2.26	.000	88.25 1.86	.338	85.18 2.73	.000
256	73.11 2.77	.000	66.53 2.09	.000	87.93 2.31	.700	84.53 2.12	.000
156	17.60 1.48	.000	15.08 2.48	.000	86.54 2.46	.146	82.77 2.33	.000
56	8.86 0.930	.000	5.33 0.95	.000	81.99 1.90	.000	77.47 2.33	.000