R help - Apr 2010 - Help: coxph() in {survival} package

Hi All,

I'm runnning coxph() on 90 different datasets (in a loop).

1. I'm wondering how can I get "log-likelihood" value from coxph()
output.
Currently I can only see following:

Likelihood ratio test 
Wald test 
Score (logrank) test 
2. Once I have likelihood value, I would like to extract these from R
output (for each data set).
Can any one help please, thanks in advance,
Xin
P.S. OR, may be if anyone can suggest me some other function which can help
me with this.

	[[alternative HTML version deleted]]

Xin Ge wrote:
> Hi All,
> 
> I'm runnning coxph() on 90 different datasets (in a loop).
> 
> 1. I'm wondering how can I get "log-likelihood" value from
coxph() output.
> Currently I can only see following:
> 
> Likelihood ratio test > 
> Wald test > 
> Score (logrank) test > 
> 2. Once I have likelihood value, I would like to extract these from R
> output (for each data set).
> Can any one help please, thanks in advance,
> Xin
> P.S. OR, may be if anyone can suggest me some other function which can help
> me with this.
If you are trying to pick a "winner" it would be advisable to rank the
multiple models and to get bootstrap confidence intervals for the ranks. 
    I provided some code for this a couple of weeks ago.

Frank

-- 
Frank E Harrell Jr   Professor and Chairman        School of Medicine
                      Department of Biostatistics   Vanderbilt University

Hi Xin, to answer your question: say you have your regression

reg = coxph(...

Then you can assess the log likelihood by 

reg$loglik

This vector contains typically two values, the null log likelihood of the
restricted model that excludes the fixed effects and the log likelihood of
the unrestricted model that includes the fixed effects. The latter is
certainly the value you want. It is the second in the vector provided by
reg$loglik.

To store them you could just create a vector and add the log likelihood
values to the vector. Something along the lines of:

logliks=numeric(90)
for(i in 1:90){
	reg=coxph(... , data=aListOfYour90Datasets[[i]])
	logliks[i]=reg$loglik[2]
}

HTH,
Daniel
	 

-------------------------
cuncta stricte discussurus
-------------------------

-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On
Behalf Of Xin Ge
Sent: Monday, April 19, 2010 11:19 PM
To: r-help at r-project.org
Subject: [R] Help: coxph() in {survival} package

Hi All,

I'm runnning coxph() on 90 different datasets (in a loop).

1. I'm wondering how can I get "log-likelihood" value from coxph()
output.
Currently I can only see following:

Likelihood ratio test 
Wald test 
Score (logrank) test 
2. Once I have likelihood value, I would like to extract these from R
output (for each data set).
Can any one help please, thanks in advance,
Xin
P.S. OR, may be if anyone can suggest me some other function which can help
me with this.

	[[alternative HTML version deleted]]

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

The best answer to your question is
  help(coxph.object)

This is a manual page describing the contents of a coxph fit.

Terry Therneau