Research Article
Comparison of Variable Selection Methods for Time-to-Event Data in High-Dimensional Settings
Table 7
Number of selected predictors and gene frequency occurrence for breast cancer dataset and each selection method (resampling results).
| Training fraction | Number of selected predictors med (min–max) | LASSO-cvl | LASSO-pcvl | Elastic Net | BSS Enet | CoxBoost | RSF |
| 1/2 | 26 (4-71) | 11 (0-59) | 47 (11-100) | 30 (17-49) | 26 (0-56) | 82 (71-93) | 2/3 | 40 (6-70) | 14 (4-53) | 58 (14-104) | 41 (31-62) | 40 (5-70) | 85 (72-93) | | Occurrence frequency med (min–max) | 1/2 | 2 (1-65) | 2 (1-57) | 2 (1-73) | 2 (1-77) | 2 (1-64) | 5 (1-57) | 2/3 | 3 (1-92) | 2 (1-73) | 3 (1-95) | 3 (1-98) | 3 (1-91) | 5 (1-61) |
|
|