Table 6.

MAPK model structure obtained from two-stage, forward and backward subset selection

Selection order	Term index	Terms	Param’s	SSE
1	132	ut−2ut−3ut−5	−0.0990	2.1193
2	56	y2t−2	2.1606	0.9038
3	51	y2t−1	−3.3213	0.7320
4	251	y3t−1	1.5365	0.3664
5	57	yt−2yt−3	−1.1189	0.3410
6	8	yt−3	0.9515	0.3163
7	7	yt−2	−3.1124	0.2984
8	6	yt−1	3.8819	0.2851

The parameters P_k and regressor terms Θ_k selected are given for the case of 100 training samples. The two-stage method selected a new set of terms: {132, 56, 51, 251, 57, 8, 7, 6}