Table 2.

Average accuracy of tenfold cross‐validation of discrimination between six VOCs analyte gases under diverse algorithms with different input vectors.

Input vector	Accuracy
	DT	SVM‐RBF	SVM‐LIN	LR	KNN	LDA
(S1, S2, S3)	86.7%	78.8%	75.2%	66.1%	86.7%	52.8%
$({\mathrm{R}}_1^1,{\mathrm{R}}_2^1,{\mathrm{R}}_3^1)$	89.2%	92.2%	89.7%	84.9%	89.8%	77.1%
$({\mathrm{R}}_1^2,{\mathrm{R}}_2^2,{\mathrm{R}}_3^2)$	89.1%	90.3%	91.0%	85.0%	84.3%	75.3%
(${\mathrm{R}}_1^3,{\mathrm{R}}_2^3,{\mathrm{R}}_3^3)$	84.4%	74.0%	60.0%	61.3%	79.5%	37.7%
$({\mathrm{S}}_1,{\mathrm{S}}_2,{\mathrm{S}}_3,{\mathrm{R}}_1^1,{\mathrm{R}}_2^1,{\mathrm{R}}_3^1)$	93.3%	98.2%	98.8%	97.6%	98.2%	86.7%
$({\mathrm{S}}_1,{\mathrm{S}}_2,{\mathrm{S}}_3,{\mathrm{R}}_1^1,{\mathrm{R}}_2^1,{\mathrm{R}}_3^1,$ ${\mathrm{R}}_1^2,{\mathrm{R}}_2^2,{\mathrm{R}}_3^2)$	95.8%	97.6%	98.2%	98.2%	97.6%	90.4%
$({\mathrm{S}}_1,{\mathrm{S}}_2,{\mathrm{S}}_3,{\mathrm{R}}_1^1,{\mathrm{R}}_2^1,{\mathrm{R}}_3^1,$ ${\mathrm{R}}_1^2,{\mathrm{R}}_2^2,{\mathrm{R}}_3^2,{\mathrm{R}}_1^3,{\mathrm{R}}_2^3,{\mathrm{R}}_3^3)$	95.8%	97.6%	98.2%	97.6%	96.9%	97.0%
$({\mathrm{R}}_1^1,{\mathrm{R}}_2^1,{\mathrm{R}}_3^1,{\mathrm{R}}_1^2,{\mathrm{R}}_2^2,{\mathrm{R}}_3^2)$	92.8%	90.4%	90.4%	88.6%	90.4%	77.2%
$({\mathrm{R}}_1^1,{\mathrm{R}}_2^1,{\mathrm{R}}_3^1,{\mathrm{R}}_1^2,{\mathrm{R}}_2^2,{\mathrm{R}}_3^2,$ ${\mathrm{R}}_1^3,{\mathrm{R}}_2^3,{\mathrm{R}}_3^3)$	95.1%	96.4%	94.6%	92.8%	95.8%	88.6%
$({\mathrm{R}}_1^2,{\mathrm{R}}_2^2,{\mathrm{R}}_3^2,{\mathrm{R}}_1^3,{\mathrm{R}}_2^3,{\mathrm{R}}_3^3)$	93.4%	94.6%	91.0%	89.2%	94.6%	79.9%