Table 4.

Classification performance of the SFS-KNN algorithm with different neighborhood sizes.

Neighborhood size		K = 1	K = 3	K = 5	K = 7	K = 9
Pre-norm	Selected features	14, 16	14, 31, 5, 15, 26, 4, 27, 21, 24, 9, 32, 2, 25, 8, 28, 3, 16	14, 31, 10, 36, 3, 25, 2	12, 31, 8, 29, 3, 15, 33, 1	12, 31, 23, 26, 3, 24, 30, 16
	Accuracy	88.29%	93.68%	93.29%	91.71%	92.24%
Norm	Selected features	12, 30	20, 15, 11, 30, 5	12, 30, 31, 33, 14	12, 19, 20, 30, 5, 18, 25, 17, 34, 3, 32, 15, 24	12, 19, 29, 30, 8, 34, 33, 25, 15, 6, 24, 7, 10, 20, 17
	Accuracy	93.95%	96.45%	96.58%	96.18%	97.89%