Table 4.

The classification of the accuracy of the logistic regression (LR), support vector machine (SVM), random forest (RF), convolution neural network (CNN), and residual neural network (ResNet).

Models	Categ	Parameter	Vis-NIR (%)			Parameter	NIR (%)
			Train a	Val b	Test c		Train	Val	Test
SVM	0	2.0, 0.1, poly	95.9	94.8	91.4	6.6, 1.0, linear	99.4	100.0	96.6
	1	1.2, 0.1, poly	98.4	96.3	92.7	1.0, 1.0, poly	100.0	100.0	96.3
	2	1.0, 1.0, poly	1.00	88.0	93.2	1.0, 1.0, poly	100.0	100.0	95.9
LR	0	1 × 105, liblinear	100.0	89.7	93.1	100, lbfgs	99.4	93.1	98.3
	1	1 × 105, liblinear	100.0	98.8	93.9	1 × 105, liblinear	100.0	100.0	100.0
	2	1 × 104, liblinear	100.0	92.0	95.9	100, newton-cg	100.0	98.7	97.3
RF	0	8, 450	100.0	77.6	79.3	6, 750	100.0	74.1	81.0
	1	7, 500	99.6	72.3	73.2	5, 550	98.8	86.7	87.8
	2	8, 200	100.0	66.7	75.7	4, 250	99.1	98.7	93.2
CNN	0	500, 32, 0.001	99.4	98.3	93.1	500, 32, 0.001	100.0	100.0	98.3
	1	500, 32, 0.001	97.6	97.6	92.7	500, 32, 0.001	100.0	100.0	98.8
	2	500, 32, 0.001	100.0	98.7	93.2	500, 32, 0.001	99.5	100.0	98.6
ResNet	0	1000, 32, 0.005	100.0	94.8	93.1	600, 32, 0.005	100.0	93.1	86.2
	1	1000, 32, 0.005	100.0	100.0	98.8	1000, 32, 0.005	100.0	100.0	97.6
	2	1000, 32, 0.005	100.0	97.3	94.6	600, 32, 0.005	97.7	100.0	97.3

^a,b,c represent training, validation, and test sets for the model; 0,1,2 represent Cabernet, Red grape and Munage, respectively, Categ mean Category of the grape. Parameters of the SVM, LR, RF, and CNN ResNet are shown. The parameters of the SVM, are (C, gamma, kernel); those of the LR are (C, solver); those of the RF are (n_estimator, max_depth); those of the CNN and ResNet are (epoch, batchsize, learning rate).