Table 3.
False discovery rates and false-negative rates for each simulated dataset and selection method.
| N | q | Training fraction | FDR/FNR | |||||
|---|---|---|---|---|---|---|---|---|
| LASSO-cvl | LASSO-pcvl | Elastic Net | BSS Enet | CoxBoost | RSF | |||
| 500 | 0 | 1/2 | 0.43/. | 0.51/. | 0.45/. | 1/. | 0.49/. | 1/. |
| 2/3 | 0.39/. | 0.48/. | 0.4/. | 1/. | 0.49/. | 1/. | ||
| 12 | 1/2 | 0.62/0.29 | 0.32/0.36 | 0.69/0.19 | 0.56/0.24 | 0.65/0.3 | 0.92/0.56 | |
| 2/3 | 0.61/0.18 | 0.28/0.23 | 0.69/0.1 | 0.6/0.12 | 0.64/0.19 | 0.9/0.51 | ||
| 50 | 1/2 | 0.51/0.39 | 0.35/0.45 | 0.56/0.3 | 0.3/0.39 | 0.47/0.41 | 0.83/0.74 | |
| 2/3 | 0.52/0.27 | 0.34/0.31 | 0.58/0.19 | 0.32/0.25 | 0.47/0.29 | 0.78/0.72 | ||
|
| ||||||||
| 750 | 0 | 1/2 | 0.44/. | 0.5/. | 0.46/. | 1/. | 0.46/. | 1/. |
| 2/3 | 0.23/. | 0.42/. | 0.24/. | 1/. | 0.38/. | 1/. | ||
| 12 | 1/2 | 0.62/0.24 | 0.31/0.27 | 0.67/0.16 | 0.59/0.16 | 0.64/0.25 | 0.85/0.4 | |
| 2/3 | 0.57/0.21 | 0.19/0.23 | 0.62/0.12 | 0.63/0.11 | 0.61/0.2 | 0.65/0.29 | ||
| 50 | 1/2 | 0.49/0.29 | 0.27/0.35 | 0.54/0.2 | 0.32/0.27 | 0.44/0.3 | 0.75/0.71 | |
| 2/3 | 0.48/0.17 | 0.2/0.26 | 0.54/0.1 | 0.33/0.15 | 0.4/0.2 | 0.54/0.66 | ||
|
| ||||||||
| 1000 | 0 | 1/2 | 0.45/. | 0.48/. | 0.45/. | 1/. | 0.51/. | 1/. |
| 2/3 | 0.55/. | 0.57/. | 0.56/. | 1/. | 0.62/. | 1/. | ||
| 12 | 1/2 | 0.71/0.23 | 0.4/0.27 | 0.74/0.16 | 0.7/0.16 | 0.73/0.23 | 0.71/0.32 | |
| 2/3 | 0.75/0.19 | 0.36/0.23 | 0.78/0.12 | 0.77/0.1 | 0.77/0.19 | 0.22/0.4 | ||
| 50 | 1/2 | 0.52/0.15 | 0.27/0.18 | 0.58/0.1 | 0.35/0.11 | 0.44/0.16 | 0.54/0.65 | |
| 2/3 | 0.56/0.1 | 0.24/0.13 | 0.61/0.06 | 0.42/0.07 | 0.44/0.12 | 0.15/0.7 | ||