Nonlinear Model Order Selection: A GMM Clustering Approach Based on a Genetic Version of EM Algorithm

Table 1

The FNN ratio when R is obtained by different algorithms.

	FNN %; Input order (n)	Output order (m)
		0	1	2	3	4
(a) FNN ratios in ex. 1 when R is equals to constant 1	0	100	50.71	6.91	2.88	1.43
	1	87.57	26.84	4.50	1.87	0.92
	2	75.63	19.43	4.03	1.71	0.85
	3	65.83	14.35	3.53	1.56	0.79
	4	56.57	11.45	3.08	1.44	0.74
(b) FNN ratios in ex. 1 when R is acquired by EM algorithm	0	100	55.13	6.56	0.48	0.09
	1	50.48	13.52	2.03	0.17	0.04
	2	2.10	0.10	0.05	0.05	0
	3	0.35	0.08	0.02	0	0
	4	0.12	0.04	0	0	0
(c) FNN ratios in ex. 1 when R is acquired by GA-EM algorithm	0	100	68.05	6.30	0.94	0.17
	1	49.50	17.01	3.02	0.23	0.05
	2	3.90	2.33	0.07	0	0
	3	0.35	0.08	0	0	0
	4	0.11	0.01	0	0	0