Table 3.

PAIFE performance over genomic (1–11) and proteomic (12) datasets.

ID	Data source	Features	Sample size		Features Removed (%)	Classification Accuracy (%)
			Class 1	Class 0		Reduced data	Full data
1	Alon et al.16	6,584	40	21	70.05	95.12	95.95
2	Beer et al.17	5,372	69	17	84.07	92.08	81.6
3	Bhattacharjee et al.18	5,372	17	52	92.68	85.71	73.1
4	Golub, et al.19	7,129	25	47	67.91	96.61	92.41
5	Hedenfalk et al.20	7,464	18	18	74.81	97.5	39.58
6	Iizuka, Oka et al.21	7,129	20	40	90.19	81.67	70
7	Pomeroy, et al.22	7,129	21	39	93.08	76.67	60.83
8	Rosenwald, et al.24	7,399	102	138	85.35	68.13	62.71
9	Singh, et al.25	12,599	52	50	46.72	64.23	57.55
10	Van’t Veer, et al.26	24,481	34	44	82.64	77.14	67.59
11	Yeoh, et al.27	12,625	48	201	90.76	87.94	80.55
12	Pusztai, et al.23	11,170	101	58	32.51	71.69	70.13