Table 3.

Results of classification models using NIR spectra based on full spectra and effective wavelengths.

Temperature (°C)	Classifier	Parameter 1	Full Spectra (%)		Parameter	Effective Wavelengths (%)
			Calibration	Prediction		Calibration	Prediction
4	PLS-DA	10	100	100	10	98.75	100
	SVM	(106, 103)	100	97.50	(106, 104)	98.75	92.50
	ELM	12	100	100	18	100	100
20	PLS-DA	4	100	100	4	100	100
	SVM	(103, 103)	100	100	(103, 105)	100	100
	ELM	7	100	100	8	100	100

¹ Parameter means the parameters of partial least squares discriminant analysis (PLS-DA), support vector machine (SVM), and extreme learning machine (ELM) models with optimal performances. The parameter for PLS-DA is the optimal number of latent variables; the parameters for SVM models are the regularization parameter c and kernel function parameter g; the parameter of the ELM model is the number of hidden layer neurons.