Table 1.

Comparison of precision, recall, and F-measure obtained with stepwise regression (SW), elastic net (EN), logistic lasso (LL) and BOSS

Scenario	Average precision				Average recall				Average F-measure
	SW	EN	LL	BOSS	SW	EN	LL	BOSS	SW	EN	LL	BOSS
A90	0.29	0.26	0.33	0.93	1	0.99	0.99	1	0.44	0.40	0.49	0.96
A70	0.28	0.35	0.45	0.78	0.93	0.89	0.89	0.80	0.42	0.48	0.57	0.75
A50	0.23	0.40	0.52	0.68	0.73	0.54	0.50	0.46	0.34	0.44	0.46	0.53
B90	0.35	0.28	0.39	0.95	1	1	1	0.98	0.52	0.43	0.56	0.96
B70	0.31	0.35	0.44	0.78	0.82	0.95	0.93	0.77	0.45	0.50	0.59	0.76
B50	0.26	0.48	0.59	0.83	0.63	0.68	0.61	0.47	0.36	0.49	0.52	0.58

Each value is the average across the 50 datasets simulated in each scenario. In terms of F-measure, BOSS performed better than the three reference methods (p<0.01, one-sided paired t test).