Qualitative Discrimination of Intact Tobacco Leaves Based on Near-Infrared Technology

Table 3

Accuracy of the DPLS model using different preprocessing methods (average spectra).

Preprocessing method	Position-GV discrimination model		Grade discrimination model
	Training set	Validation set	Training set	Validation set
None	96.24	86.81	83.88	77.96
1st derivative	98.49	90.69	86.10	80.02
SNV	92.08	90.41	85.30	80.02
1st derivative + SNV	98.12	92.71	86.84	81.75