. 2019 Sep 30;8(10):5249–5258. doi: 10.1002/fsn3.987

Table 1.

Selection of preprocessing spectral treatment

Type of spectra	Preprocessing	PLS		PLS		No. of factors
Type of spectra	Preprocessing	RMSE	R ²	RMSEP	R ²	No. of factors
Full Spectra (4,000–10,000 cm⁻¹)	Without spectral transformations	0.680	0.988	0.552	0.989	4
Spectra from (4,000–7,300 cm⁻¹)	Without spectral transformations	0.668	0.991	0.568	0.989	4
Full Spectra (4,000–10,000 cm⁻¹)	Baseline	0.627	0.986	0.631	0.986	5
Spectra (4,000–7,300 cm⁻¹)	Baseline	0.592	0.988	0.623	0.987	5
Full Spectra (4,000–10,000 cm⁻¹)	SNV	0.684	0.984	0.606	0.987	5
Spectra (4,000–7,350 cm⁻¹)	SNV	0.635	0.986	0.553	0.989	5
Full Spectra (4,000–10,000 cm⁻¹)	1st derv. spectral transformations	0.649	0.985	0.760	0.980	3
Spectra (4,000–6,400 cm⁻¹)	1st derv. spectral transformations	0.563	0.987	0.483	0.983	3

PLS‐DA: partial least‐squares discriminant analysis; RMSEP: root means square error of prediction; SNV: standard normal variate.