Robust and efficient feature selection algorithm to identify important features for predicting survival risk. The method is based on subsampling and averaging linear models obtained from the (preconditioned) Lasso algorithm, with an extra shrinking procedure to reduce the size of signatures. An evaluation procedure using subsampling is also provided.
| Version: | 1.0 |
| Depends: | R (≥ 2.15.0), survival, parallel |
| Imports: | BBmisc, glmnet, superpc, survcomp, Matrix |
| Suggests: | testthat |
| Published: | 2013-10-12 |
| Author: | Sangkyun Lee, Michel Lang |
| Maintainer: | Sangkyun Lee <sangkyun.lee at tu-dortmund.de> |
| License: | GPL-2 |
| NeedsCompilation: | no |
| In views: | Robust, Survival |
| CRAN checks: | rsig results |
| Reference manual: | rsig.pdf |
| Package source: | rsig_1.0.tar.gz |
| Windows binaries: | r-devel: rsig_1.0.zip, r-release: rsig_1.0.zip, r-oldrel: rsig_1.0.zip |
| OS X Snow Leopard binaries: | r-release: rsig_1.0.tgz, r-oldrel: rsig_1.0.tgz |
| OS X Mavericks binaries: | r-release: rsig_1.0.tgz |