Quantitative analysis of cadmium in rice roots based on LIBS and chemometrics methods

Environmental Sciences Europe

Table 5 Multivariate analysis results based on different variables

Pretreatment methods	Variable (nm)	Model	Parameter^a	Calibration set		Prediction set
Pretreatment methods	Variable (nm)	Model	Parameter^a	R_c	RMSEC (mg/kg)	R_p	RMSEP (mg/kg)
Raw	214.44 + 226.50 + 228.80	PLSR	2	0.9708	37.7	0.9612	46.1
		LS-SVM	(588,390.13, 86.94)	0.9869	25.3	0.9743	38.6
		ELM	36	0.9511	48.6	0.9429	54.3
	Full	PLSR	3	0.9778	33.0	0.9659	42.1
		LS-SVM	(1974.88, 16,315.65)	0.9923	19.5	0.9806	31.6
		ELM	28	0.9573	45.5	0.9557	49.4
After pretreatment	214.44 + 226.50 + 228.80	PLSR	2	0.9859	26.3	0.9791	32.4
		LS-SVM	(2026.25, 151.09)	0.9911	21.0	0.9828	29.4
		ELM	31	0.9921	19.7	0.9869	25.6
	Full	PLSR	3	0.9863	26.0	0.9808	31.1
		LS-SVM	(767.71, 35,346.51)	0.9935	17.9	0.9837	28.7
		ELM	15	0.9943	16.8	0.9896	26.0

Bold emphasis: the optimal model for given conditions
^aParameters of different models: the optimal number of latent variables (LVs) for PLSR, the bandwidth of kernel function (sig2) and the trade-off between the minimum model complexity and the minimum training error (gam) for LS-SVM, the number of hidden nodes for ELM.