Caret for Survival Analysis?

[annette987]

Hi,
Is there any update on being able to use caret for survival analysis? It is a fantastic package and I would really like to be able to tune the parameters for a random survival forest from the randomForestSRC package. In this post: https://stackoverflow.com/questions/29963011/using-caret-for-survival-analysis-random-survival-forest you indicated it might be coming soon and that was back in 2015.
Thanks,
Annette

[topepo]

No. I think that we are going to pour the survival analysis work into the tidymodels packages. The changes needed for caret are pretty invasive and it is easier to move that work to the new system.

I'm looking for more feedback on how these models should be processed and evaluated (I haven't done much of this since grad school, esp with Cox PH models). If you have a workflow, it would be helpful to share so that we can build the framework. I'd be happy to give you a script to do replicate what you want from caret if you can mock up some data.

We'll be discussing this specifically on the tidyverse developer day but would like to know what you think.

Currently, there are parts for fitting parametric models, resampling, and model evaluation. There are also some pending PRs for performance metrics for yardstick here and here. Feedback on those PRs would be helpful.

[anuj2054]

I would like to be involved in this.

[cysics]

me too.

[topepo]

This is happening in tidymodels see https://github.com/tidymodels/planning/tree/master/survival-analysis