Table 4.

LASSO for logistic regression models based on case-control data with $n_1$ cases and $n_0$ controls

$F_{\mathbf{X}}$	n	${\mathit{\beta }}^{\ast }$	FP rate	FN rate	Coverage of zero for				Coverage of 95 % CIs of ${\mathit{\beta }}^{\ast }$ for $\widehat{\mathit{\beta }}\ne 0$
					$\mathit{\beta }=0$		${\mathit{\beta }}^{\ast }\ne 0$
					(Adapt${}^{\mathrm{a}}$)	(Oracle${}^{\mathrm{b}}$)	(Adapt)	(Oracle)	(Adapt)	(Oracle)
Normal	100	0.25	0.069	0.814	0.999	0.970	0.997	0.945	0.104	0.073
Normal	200	0.25	0.080	0.776	0.998	0.968	0.994	0.917	0.146	0.116
Normal	500	0.25	0.140	0.593	0.998	0.962	0.975	0.804	0.346	0.319
Normal	100	0.500	0.102	0.688	0.999	0.963	0.988	0.860	0.242	0.202
Normal	200	0.500	0.198	0.423	0.997	0.951	0.930	0.677	0.532	0.489
Normal	500	0.500	0.365	0.063	0.994	0.946	0.634	0.287	0.828	0.896
Normal	100	1.000	0.262	0.229	0.997	0.934	0.853	0.488	0.700	0.668
Normal	200	1.00	0.385	0.022	0.994	0.936	0.488	0.166	0.832	0.912
Normal	500	1.00	0.443	0.000	0.993	0.945	0.029	0.004	0.850	0.937
Binomial	100	0.25	0.086	0.760	0.999	0.965	0.992	0.907	0.163	0.125
Binomial	200	0.25	0.120	0.646	0.998	0.960	0.976	0.828	0.288	0.25
Binomial	500	0.25	0.264	0.259	0.996	0.949	0.859	0.533	0.713	0.683
Binomial	100	0.50	0.177	0.451	0.998	0.947	0.941	0.683	0.502	0.445
Binomial	200	0.50	0.314	0.119	0.996	0.938	0.750	0.381	0.778	0.817
Binomial	500	0.50	0.425	0.002	0.994	0.944	0.192	0.046	0.859	0.940
Binomial	100	1.00	0.367	0.039	0.996	0.915	0.578	0.206	0.793	0.815
Binomial	200	1.00	0.442	0.001	0.994	0.931	0.121	0.020	0.824	0.897
Binomial	500	1.00	0.483	0.000	0.993	0.942	0.000	0.000	0.813	0.930

$F_X$ is the distribution of the predictors $\mathbf{X}$

FP false positive, FN false negative

${}^{\mathrm{a}}$ Corresponds to Algorithm 1, and ${}^{\mathrm{b}}$ to Algorithm 2 in the text