RMS Describing, Resampling, Validating, and Simplifying the Model

data analysis modeling strategy
, ,
158

Hi,
this post comes from this other one that was posted in the wrong place:

https://discourse.datamethods.org/t/validate-and-calibrate-cox-model-on-dataset-with-a-categorical-covariate-with-few-observations/8767?u=marc_vila_forteza

Professor Harrell’s answer:
You may have to combine infrequent levels. Hmisc::combine.levels can help with that. Or use the group= argument to calibrate and validate to do balanced bootstrap sampling. Specify something like group=mydataset$FLUID2 .

→ Thank you for your quick reply. I have another question with regards the use of group=argument. Is it possible to use it for grouping more than one covariate at the same time?

I mean to do something similar to ‘group=c(mydataset$FLUID2,mydataset$TYPE)’.
note: I tested this way to use a vector to specify more than one covariate but it didn’t work.

Thank you