Table 3.

Simulation results for evaluating prediction performance in the presence of high-dimensional data based on 400 Monte Carlo data sets, where n = 100. Lasso-PL, Lasso partly linear AFT model; Lasso-S_K, Lasso stratified model with K strata; Lasso-L, Lasso linear AFT model assuming a linear effect for both X_i and Z_i; and Lasso-Cox, Lasso linear Cox model assuming a linear effect for both X_i and Z_i. MSPE₁, the squared prediction error using both nonlinear and linear components; MSPE₂, the squared prediction error using only linear components; and c, the c-statistic for censored data. Range of SEs, range of SEs for the corresponding performance measure in each column. NA, a performance measure is not applicable for a estimator. All numbers are multiplied by 1000.

	d = 100		c	d = 500		c	d= 1500		c
	MSPE1	MSPE2		MSPE1	MSPE2		MSPE1	MSPE2
	ρ = 0
Lasso-PL	412	349	860	989	897	840	1685	1543	796
Lasso-S2	NA	676	811	NA	1589	768	NA	2310	711
Lasso-S4	NA	560	812	NA	1428	780	NA	2182	718
Lasso-S8	NA	529	811	NA	1454	775	NA	2208	716
Lasso-L	1441	568	829	2752	1666	784	3719	2496	697
Lasso-Cox	NA	NA	798	NA	NA	749	NA	NA	684
	ρ = 0.5
Lasso-PL	389	330	860	1034	937	839	1659	1518	797
Lasso-S2	NA	637	810	NA	1653	766	NA	2270	716
Lasso-S4	NA	525	812	NA	1472	777	NA	2152	725
Lasso-S8	NA	491	811	NA	1512	774	NA	2196	721
Lasso-L	1418	550	829	2803	1720	781	3703	2513	701
Lasso-Cox	NA	NA	799	NA	NA	749	NA	NA	690
	ρ = 0.9
Lasso-PL	387	328	875	1084	1124	852	1795	1909	811
Lasso-S2	NA	529	841	NA	1314	815	NA	2059	769
Lasso-S4	NA	474	842	NA	1422	812	NA	2253	759
Lasso-S8	NA	455	841	NA	1618	805	NA	2473	744
Lasso-L	1476	480	852	2274	1152	836	3179	1849	802
Lasso-Cox	NA	NA	840	NA	NA	825	NA	NA	796
Range of SEs	9-20	8-23	0.6-2	32-56	32-52	1-4	47-61	47-57	2-5