Table 1.

Comparison of means of classification recall, F1 score and AUC for each Savitzky-Golay window size by algorithm and derivative order.

			Window size
Algorithm	Performance statistic	Derivative order	5	9	15	Significance
SVM	Recall	First	0.8839	0.8648	0.8351	n.s.
		Second	0.9024	0.8947	0.8842	n.s.
	F1 score	First	0.8934	0.8747	0.8450	n.s.
		Second	0.9142	0.9058	0.8938	n.s.
	AUC	First	0.9309	0.9305	0.9297	n.s.
		Second	0.9339	0.9328	0.9265	n.s.
MLP	Recall	First	0.9796 a	0.9678 b	0.9404 c	***
		Second	0.9905 a	0.9842 b	0.9804 c	***
	F1 score	First	0.9796 a	0.9687 b	0.9404 c	***
		Second	0.9905 a	0.9842 b	0.9804 c	***
	AUC	First	0.9998 a	0.9995 b	0.9986 c	***
		Second	0.9999 a	0.9998 b	0.9996 c	***

The values represent the average recall.

Dissimilar lowercase letters within rows represent statistically different means among different window sizes, using Tukey's range test at a significance level p = 0.05.

AUC, area under the receiver operating characteristic curve; SVM, support vector machine; MLP, multilayer perceptron. n.s., not significant (p≥0.05);

^***p < 0.001.