High-throughput prediction of eucalypt lignin syringyl/guaiacyl content using multivariate analysis: a comparison between mid-infrared, near-infrared, and Raman spectroscopies for model development

Table 1 Evaluation of literature multivariate models for lignin S/G prediction

Method	# of samples	SEC¹	SEP²	RM-SEC³	RM- SECV⁴	Random Val. Set	RM-SEP⁵	r-Val⁶	R²Cal⁷	R²Val⁸	F⁹	Ref
Raman	10	-	-	0.02 (S) 0.04 (G)	-	N	0.04 (S) 0.04 (G)	0.993 (S) 0.992 (G)	0.996 (S) 0.997 (G)	0.985 (S) 0.986 (G)	-	[15]
Raman	55 (Cal) 25 (Val)	0.07	0.32	-	-	Y	-	0.935	0.996	0.874	5	[16]
Raman	63 (Cal) 30 (Val)	0.32	0.35	-	-	Y	-	0.919	0.887	0.845	2	[23]
Raman	9	-	-	-	-	N	-	-	-	0.983	-	[20]
MIR	5	-	-	-	-	-	-	-	-	-	-	[14]
MIR	5	-	-	-	-	-	-	-	-	-	-	[24]
MIR	65	-	-	-	-	-	-	-	-	0.91-0.98	-	[25]
MIR	15	-	-	-	-	-	-	-	-	-	-	[26]
NIR	42 (Cal) 36 (Val)	-	-	-	0.025-0.033	Y	0.025-0.036	0.959-0.980	0.95-0.97	0.92-0.96	3-5	[13]
NIR	267 (Cal)	-	-	0.176 (S) 0.202 (G) 0.005 (H)	0.201 (S) 0.202 (G) 0.005 (H)	N	-	0.979 (S) 0.979 (G) 0.843 (H)	0.968 (S) 0.957 (G) 0.731 (H)	0.958 (S) 0.957 (G) 0.710 (H)	8	[17]
NIR	135 (Cal) 45 (Val)	-	0.124	-	0.121	Y	-	0.686	0.583	0.47	5-7	[19]
NIR	26 (Cal) 8 (Val)	0.26	0.3	-	-	Y	-	0.938	0.96	0.88	6	[22]

¹Standard Error of Calibration.
²Standard Error of Prediction.
³Root Mean Standard Error of Calibration.
⁴Root Mean Standard Error of Cross-Validation.
⁵Root Mean Standard Error of Prediction.
⁶Coefficient of Correlation for Validation Set.
⁷Coefficient of Determination for Calibration Set.
⁸Coefficient of Determination for Validation Set.
⁹#of Factors.
Italicized values calculated by the authors.
MIR = mid-infrared spectroscopy, NIR = near-infrared spectroscopy, Cal = calibration set, Val = validation set, S = syringyl, G = guaiacyl, H = p -hydroxyphenol, Y = yes, N = no.

ISSN: 2731-3654