Estimate the baseline hazard function in the Cox model in R

[Peter B]

I am trying to estimate the baseline hazard function in the Cox model in R. The function basehaz in the survival package provides the cumulative hazard, not the hazard function. The “Muhaz” package can estimate the hazard function but only for one sample data as far as I know. Is there a tool for the baseline hazard function in the Cox model? If not, I will need to smooth the cumulative hazard on my own, but that might be painful.

I would appreciate any help.

Peter

[Rich Ulrich]

On Thu, 28 Feb 2013 16:49:41 -0800 (PST), Peter B <pio...@gmail.com>
wrote:

>I am trying to estimate the baseline hazard function in the Cox model in R. The function basehaz in the survival package provides the cumulative hazard, not the hazard function. The “Muhaz” package can estimate the hazard function but only for one sample data as far as I know. Is there a tool for the baseline hazard function in the Cox model? If not, I will need to smooth the cumulative hazard on my own, but that might be painful.
>
>I would appreciate any help.

Nobody has offered anything, so here is a shot.

I'm a little confused at what you are asking, or why you are
asking it.

The baseline hazard function cancels out in the Cox proportionate
hazard regression test. It is the function without the covariates
being tested. You can get a test of a single covariate by
comparing the maximum likelihood values for the fit.

So, I suppose it might be interesting to see the cumulative hazard
or the hazard function - to see that it fits reasonably - but it is
not very important.

My impression is that you have two groups, and you are trying
to fit one group at a time. That is not the method.

I don't know anything about it, but it sounds like you want to
use the Muhaz package for the "one sample" - which includes
all the cases - with and without the covariate of "group".
Or whatever.

Hope this helps. If someone else knows more, please jump in.

--
Rich Ulrich