A comprehensive evaluation of multicategory classification methods for microbiomic data

Microbiome

Table 5 Classification accuracy without feature/operational taxonomic unit (OTU) selection, measured by relative classifier information (RCI)

Classifier	CBH	CS	CSS	FS	FSH	BP	PDX	PBS	Averages	P values
SVM, Linear C = 1	0.769	0.918	0.674	0.882	0.749	0.158	0.228	0.602	0.623	0.165*
SVM, Linear optimized C	0.771	0.915	0.674	0.958	0.751	0.157	0.241	0.607	0.634	0.294*
SVM, Poly	0.771	0.915	0.674	0.958	0.751	0.162	0.241	0.607	0.635	0.299*
SVM, RBF	0.689	0.907	0.631	0.942	0.731	0.156	0.202	0.561	0.602	0.059*
KRR, Poly	0.765	0.927	0.671	0.911	0.758	0.157	0.230	0.612	0.629	0.206*
KRR, RBF	0.774	0.913	0.675	0.935	0.759	0.163	0.242	0.598	0.632	0.265*
KNN, K = 1	0.344	0.329	0.377	0.163	0.355	0.167	0.074	0.078	0.236	0.003
KNN, K = 5	0.178	0.359	0.277	0.102	0.203	0.056	0.092	0.062	0.166	0.002
KNN, optimized K	0.337	0.402	0.354	0.028	0.207	0.089	0.122	0.196	0.217	0.003
PNN	0.325	0.292	0.411	0.236	0.342	0.041	0.070	0.089	0.226	0.002
L2-LR, C = 1	0.772	0.941	0.670	0.964	0.778	0.161	0.236	0.628	0.644	0.575*
L2-LR, optimized C	0.782	0.939	0.680	0.958	0.778	0.163	0.228	0.624	0.644	0.626*
L1-LR, C = 1	0.769	0.825	0.635	0.949	0.779	0.163	0.191	0.565	0.610	0.089*
L1-LR, optimized C	0.798	0.910	0.664	0.960	0.790	0.174	0.209	0.599	0.638	0.439*
RF, default	0.767	0.957	0.671	0.998	0.803	0.173	0.087	0.618	0.634	0.253*
*RF, optimized*	0.784	0.962	0.681	0.994	0.805	0.225	0.098	0.625	0.647	-
BLR, Laplace priors	0.759	0.932	0.679	0.922	0.770	0.166	0.090	0.619	0.617	0.085*
BLR, Gaussian priors	0.744	0.759	0.496	0.930	0.736	0.077	0.014	0.529	0.536	0.008

The nominally best performing classifier on average over all datasets is marked with bold, and P values of methods whose performance cannot be deemed statistically worse than the nominally best performing method are marked with “*”. The accuracy of the nominally best performing method for each dataset is underlined.

ISSN: 2049-2618