The survival_prob_*() and surival_time_*() functions typically take a model object as fitted by the engine as input. The exceptions are the two functions for glmnet objects (coxnet), which need the parsnip model_fit object (for some additional components that are not stored in the engine fit object directly).
These functions preferably all take the same input object. Since we do need the parsnip fit for glmnet, we'd need to change all the other functions. This would be a breaking change. They are technically exported but developer-facing and tagged with @keyword internal.
I think it makes sense to bite the bullet now and introduce that breaking change: Those functions are not advertised as user-facing, censored is still in its experimental lifecycle stage, we are still "before the big release cascade" for survival analysis, and we might have to live with this rather big inconsistency for rather long.
This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.
The
survival_prob_*()
andsurival_time_*()
functions typically take a model object as fitted by the engine as input. The exceptions are the two functions for glmnet objects (coxnet), which need the parsnipmodel_fit
object (for some additional components that are not stored in the engine fit object directly).These functions preferably all take the same input object. Since we do need the parsnip fit for glmnet, we'd need to change all the other functions. This would be a breaking change. They are technically exported but developer-facing and tagged with
@keyword internal
.I think it makes sense to bite the bullet now and introduce that breaking change: Those functions are not advertised as user-facing, censored is still in its experimental lifecycle stage, we are still "before the big release cascade" for survival analysis, and we might have to live with this rather big inconsistency for rather long.