[R] Warning message as a result of logistic regression performed

[Gavin Simpson]

Try reading this thread:

http://thread.gmane.org/gmane.comp.lang.r.general/134368/focus=134475

especially the posts by I Kosmidis which show you how to diagnose
problems in logit model fits like this.

There is a statement about this warning in ?glm as well and a pointer to
a reference which discusses a source of the warning.

G

On Wed, 2009-05-27 at 11:22 +0100, Winter, Katherine wrote:
> I am sorry if this question sounds basic but I am having trouble understanding a warning message I have been receiving in R after attempting logistic regression.
> 
> I have been using the logistic regression function in R to analyse a simulated data set. The dependent variable "failure" has an outcome of either 0 (success) or 1 (failure). Both the independent variables have been previously generated in a mathematical model and stored in a data.frame for analysis. I am currently using a sample size of 1000 and I use the following commands in R:
> 
>        log.reg.1 <- glm(failure ~ age +weight +init.para.log.value +k.d1,family=binomial(logit), data=test)
> 	log.reg.1.summary <- summary(log.reg.1); print(log.reg.1.summary)
> 	log.reg.1.exp <- exp(log.reg.1$coef); print(log.reg.1.exp)
> 
> When I execute these commands I get the following warning message: 
> 
> "In glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart,  :fitted probabilities numerically 0 or 1 occurred"
> 
> I am unsure what this warning is referring to. I have tried using google to answer this question but have had no luck. 
> 
> I have been on the following website https://stat.ethz.ch/pipermail/r-sig-ecology/2008-July/000278.html but found it was not helpful as I when I ran the example given I received no warning message (I am using R version 2.8.1). 
> 
> I am working with simulated data so there are no missing values in the data set.
> 
> I have also looked at the following website http://tolstoy.newcastle.edu.au/R/help/05/07/7759.html they suggest that the warning is as a result of "perfect separation" of the results (a possibility with simulated data). However, when I added an extra row to my data.frame of results that I knew to be false and hence to prevent "perfect separation" subsequent logistic regression still resulted in the same warning message. 
> 
> I am still at a loss as to the meaning of this message and any help in understanding this warning would be much appreciated.
> ______________________________________________
> 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.
-- 
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
 Dr. Gavin Simpson             [t] +44 (0)20 7679 0522
 ECRC, UCL Geography,          [f] +44 (0)20 7679 0565
 Pearson Building,             [e] gavin.simpsonATNOSPAMucl.ac.uk
 Gower Street, London          [w] http://www.ucl.ac.uk/~ucfagls/
 UK. WC1E 6BT.                 [w] http://www.freshwaters.org.uk
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20090527/25eba3a6/attachment-0002.bin>