Table 2.

Results of TSS predictive models of mulberries.

Number	Models	No.c	LVsd	Calibration		Cross-validation		Prediction		γ(103)	σ 2(103)
				R C	RMSEC	R CV	RMSECV	R P	RMSEP
(1)	Full-PLSR	460	21	0.984	0.279	0.958	0.445	0.959	0.411
(2)	Full-LS-SVM-RBF	460		0.999	<0.1	0.999	<0.1	0.999	<0.1	4.43 × 105
(3)	Full-LS-SVM-linear	460		0.999	<0.1	0.999	<0.1	0.999	<0.1	1.11 × 108	8.99 × 102
(4)	RF-PLSR	23	13	0.942	0.522	0.928	0.579	0.899	0.675
(5)	RF-LS-SVM-linear	23		0.944	0.515	0.931	0.571	0.886	0.714	4.93
(6)	RF-LS-SVM-RBF	23		0.999	0.061	0.958	0.453	0.956	0.430	9.56	4.29
(7)	RF-PLSR	11	7	0.811	0.912	0.792	0.951	0.834	0.800
(8)	RF-LS-SVM-linear	11		0.818	0.899	0.796	0.945	0.843	0.781	0.36
(9)	RF-LS-SVM-RBF	11		0.984	0.283	0.912	0.647	0.925	0.557	1.25 × 108	1.80

^cNumbers of the wavelengths used for analysis.

^dLatent variables of the PLSR model.