Comparison of Machine Learning Classification Methods for Determining the Geographical Origin of Raw Milk Using Vibrational Spectroscopy

Table 1

Statistical results obtained by the PLS-DA method.

Preprocessing data	Number of LVs	Groups	R^2Calibrarion	RMSEC	R^2Cross-validation	RMSECV	Sensitivity (%)	Specificity (%)	% CCR
Raw	12	R 1	0.98	0.056	0.97	0.069	100	100	100
		R 2	0.99	0.036	0.99	0.041	100	100
		R 3	0.95	0.10	0.81	0.200	100	100
		R 4	0.93	0.11	0.79	0.200	100	100
Detrend polynomial 1	13	R 1	0.98	0.055	0.97	0.070	100	100	100
		R 2	0.99	0.034	0.99	0.040	100	100
		R 3	0.97	0.075	0.83	0.183	100	100
		R 4	0.97	0.071	0.83	0.178	100	100
Detrend polynomial 2	13	R 1	0.98	0.056	0.97	0.070	100	100	100
		R 2	0.99	0.034	0.99	0.041	100	100
		R 3	0.97	0.075	0.84	0.185	100	100
		R 4	0.97	0.072	0.84	0.182	100	100