Selection of predictive features in multivariable modeling, one-at-a-time screening of variables, and the cost of feature selection compared to using fuller models, possibly with penalization (shrinkage; regularization).
Hi there,
I want to asses the variable importance of a dataset that will be fitted with Cox PHM model.
The RMS book (section 5.4) shows a way to do that by bootstrapping the ranks of predictors with regards of the partial χ2 minus its degrees of freedom.
When fitting later the Cox models, it would be possible that some of them do not fit the Cox assumptions (e.g. linearity, proportionality), so, maybe later we would remove a variable or stratify the model by any of them.
It may be that a variable is important but does not meet the assumptions of the Cox models. One question is whether the importance test we have done is still valid for these variables or not.
The second question is how we can decide whether to incorporate an ‘important’ variable into the model if it does not meet the Cox assumptions.
Thank you!
Hello,
is there any thought or suggestion with regards the posted question?
Thank you!