Table 5.

Performance measures of the best ordinal classifiers in comparison to eight popular non-ordinal classifiers.

	Performance Measures for Classification			Performance Measures for Ordinal Classification
	F-Score	Accuracy	AUC	MSE	${\mathit{\tau }}_{\mathit{b}}$
Non-ordinal classifiers
Naïve Bayes	0.246	0.305	0.478	0.916	−0.065
Logistic regression	0.453	0.505	0.560	1.189	0.060
Gradient boosting	0.347	0.356	0.481	1.611	−0.117
XGBoost	0.378	0.389	0.506	1.558	−0.077
K-nearest neighbor	0.433	0.453	0.543	1.305	0.013
AdaBoost	0.380	0.411	0.507	1.253	0.006
Random forest	0.405	0.400	0.540	1.421	0.035
CART	0.361	0.379	0.493	1.537	−0.079
Ordinal classifiers
Ordinal CART–OBE($EV$)	0.409	0.442	0.535	1.316	0.016
Ordinal AdaBoost—OBE($v(c^{max})$)	0.475	0.526	0.578	1.137	0.163
Ordinal RF—OBE($v(c^{mode})$)	0.439	0.484	0.570	1.147	0.185