Regularized F-Measure Maximization for Feature Selection and Classification

Table 3

Performance with different 𝛾's and 𝐿₁ F-measure maximization.

𝛾	Variables selected (1/0)	Sensitivity	Specificity	Test error	AUC
0.1	1101111	0.476	0.957	27.3	0.801
0.2	1111111	0.714	0.957	15.9	0.820
0.3	0101111	0.810	0.740	22.7	0.849
0.4	1111001	0.826	0.762	20.4	0.861
0.5	1111101	0.857	0.609	27.3	0.832
0.6	1100101	0.762	0.739	25	0.832
0.7	1110110	0.904	0.348	38	0.847
0.8	1011100	100	0.217	40.9	0.826
0.9	1100011	100	0	52.3	0.754