Table 2 Prediction accuracies of the PLSR and PCR models for EC based on different spectral smoothing methods.

Method		Calibration		Cross-validation		Independent Prediction		Number of predictors or factors
Method		R²	RMSE	R²	RMSE	R²	RMSE	Number of predictors or factors
PLSR	1	0.8796	0.2272	0.6695	0.3867	0.3069	0.6189	10
	2^a	0.7695	0.3144	0.5600	0.4485	0.5806	0.4814	7
	3	0.8807	0.2262	0.6093	0.4192	0.6414	0.4452	10
	4	0.9042	0.2027	0.6698	0.3885	0.6090	0.4649	10
PCR	1	0.7660	0.3168	0.5926	0.4298	0.5766	0.4837	19
	2	0.7563	0.3233	0.5512	0.4532	0.5804	0.4815	19
	3	0.7636	0.3184	0.5356	0.4569	0.6799	0.4206	19
	4	0.7540	0.3248	0.5830	0.4355	0.6407	0.4456	17

“Independent Prediction” stands for the accuracy of the models by independent validation set (36 selected samples). “Number of predictors or factors” denotes the number of spectral principal components extracted. The numbers 1, 2, 3, and 4 represent the moving-average data smoothing technique (MA), the Savitzky-Golay data smoothing technique (SG), the median filtering data smoothing technique (MF), and the Gaussian filtering data smoothing technique (GF) methods, respectively. The data in rows marked with the letter “a” are referenced from the literature³⁷.

Quick links

Search