Calls penalized::penalized().

  • distr is predicted using penalized::predict()

  • crank is predicted as the expectation of the survival distribution, distr

The penalized and unpenalized arguments in the learner are implemented slightly differently than in penalized::penalized(). Here, there is no parameter for penalized but instead it is assumed that every variable is penalized unless stated in the unpenalized parameter, see examples.

Format

R6::R6Class() inheriting from LearnerSurv.

Construction

LearnerSurvPenalized$new()
mlr_learners$get("surv.penalized")
lrn("surv.penalized")

Meta Information

  • Type: "surv"

  • Predict Types: distr, crank

  • Feature Types: integer, numeric, factor, ordered

  • Packages: penalized distr6

References

Goeman JJ (2009). “L1Penalized Estimation in the Cox Proportional Hazards Model.” Biometrical Journal, NA--NA. doi: 10.1002/bimj.200900028 .

See also

Examples

library(mlr3) task = tgen("simsurv")$generate(20) learner = lrn("surv.penalized") resampling = rsmp("cv", folds = 2) resample(task, learner, resampling)
#> 1234 #> 1234
#> <ResampleResult> of 2 iterations #> * Task: simsurv #> * Learner: surv.penalized #> * Warnings: 0 in 0 iterations #> * Errors: 0 in 0 iterations
# specifying penalized and unpenalized variables task = tgen("simsurv")$generate(20) learner = lrn("surv.penalized", unpenalized = c("height")) learner$train(task)
#> 1234
learner$model@penalized
#> treatment weight #> 0.97867392 0.02127612
learner$model@unpenalized
#> height #> 0.008976014