Table 4.

Model prediction results of extracted variables subsets and their combinations under different preprocessing methods for metal ion concentration prediction.

	Feature Subset	MAPE(%)	MaxAPE(%)	RMSE${}_{\mathit{P}}$(mg/L)	${\mathit{R}}^2$
	Denoising	∖	∖	∖	∖
Zinc	1st derivative	6.942	9.433	3.569	0.948
	2nd derivative	6.271	9.185	3.412	0.956
	Combination	4.098	7.986	3.107	0.987
	Denoising	4.241	9.452	0.056	0.977
Copper	1st derivative	3.924	8.404	0.051	0.984
	2nd derivative	∖	∖	∖	∖
	Combination	3.515	6.939	0.043	0.990
	Denoising	∖	∖	∖	∖
Cobalt	1st derivative	∖	∖	∖	∖
	2nd derivative	3.083	7.414	0.041	0.993
	Combination	∖	∖	∖	∖
	Denoising	6.879	9.922	0.078	0.946
Nickel	1st derivative	4.612	9.067	0.072	0.953
	2nd derivative	∖	∖	∖	∖
	Combination	4.331	8.323	0.054	0.989