A Multicriteria Approach to Find Predictive and Sparse Models with Stable Feature Selection for High-Dimensional Data

Table 7

Accuracy optimal configurations in Figure 10(a). The ID column references Table 8. The 67 Pareto optimal configurations are not displayed here.

ID	Classifier	Parameters	Filter	n.feats	Error; (train)	Error; (test)	Size; (train)	Size; (test)
12	Lasso Log. Reg.		Variance	10812	0	0.026	10811.9	10812.0
13	Lasso Log. Reg.		MRMR	9927	0	0.025	4546.2	4728.2
14	GLM Boosting		Variance	972	0	0.026	51.0	55.4
15	GLM Boosting		Variance	1586	0	0.040	25.7	42.5