Discriminating the Geographical Origins of Chinese White Lotus Seeds by Near-Infrared Spectroscopy and Chemometrics

Table 3

Model parameters and prediction results of LS-SVM and OCPLS.

Data pretreatment	OCPLS			LS-SVM			Models fusion
	LVs^a	Se^b	Sp^c	,	Se^a	Sp	Se	Sp
Raw data	4	0.800; (24/30)	0.541; (20/37)	18.04, 14.64	0.867; (26/30)	0.919; (34/37)	0.800; (24/30)	0.919; (34/37)
Smoothing	4	0.833; (25/30)	0.541; (20/37)	18.26, 14.88	0.867; (26/30)	0.919; (34/37)	0.833; (25/30)	0.946; (35/37)
D2	3	0.933; (28/30)	0.703; (26/37)	11.12, 9.41	0.900; (27/30)	0.973; (36/37)	0.900; (27/30)	0.973; (36/37)
SNV	1	0.900; (27/30)	0.595; 22/37	12.28, 11.90	0.900; (27/30)	0.919; (34/37)	0.900; (27/30)	0.946; (35/37)

Number of OCPLS components.
^bSe: sensitivity; the numbers in the brackets indicate TP/(TP + FN).
^cSp: specificity; the numbers in the brackets indicate TN/(TN + FP).