Are there any cross-validation packages in R that work with survival models, like a weibull model?
Asked
Active
Viewed 1,669 times
-2
-
I looked for an off-site resource but I couldn't find one. Do I really need to say that in my question? – user7340 Oct 20 '14 at 21:01
-
It certainly have helped to demonstrate some searching effort. What would have even been more convincing that you had read and understood the rules of SO would have been to construct a simulation and post the code for estimation of a single model and then request enhancements from perhaps the boot library or others. – IRTFM Oct 28 '14 at 20:29
2 Answers
2
The CoxBoost
package has the function cv.CoxBoost
for cross-validation of survival and competing risks models.
cvTools
is a package focused on cross-validation. I'm not sure if it has specific functions for survival models, but the package is intended for general cross-validation of regression models.

eipi10
- 91,525
- 24
- 209
- 285
1
There are calibration and validation tools in the rms/Hmisc packages that work with survival models. Harrell's RMS text is an extremely useful reference. He demonstrates using parametric (e.g. Weibull) models as well as the semi-parametric Cox models.

IRTFM
- 258,963
- 21
- 364
- 487