subset argument in nls() and possibly other functions

Hi John,

In scanning some of the more popular model functions (e.g. lm(), glm(), 
lme(), coxph(), etc.), none seem to provide examples of the use of the 
'subset' argument, even though it is documented for them.

That being said, there is some old (2003) documentation by Prof Ripley here:

https://developer.r-project.org/model-fitting-functions.html

that may be helpful, and where the link to the lm() function source code 
on the above page should be:

 ? https://svn.r-project.org/R/trunk/src/library/stats/R/lm.R

Within that source file, you might want to focus upon the 
model.frame.lm() function, the basic form which is used internally in 
many (most?, all?) of the typical model related functions in R to create 
the internal data frame from the specified formula, that is then used to 
create the model.

There is a parallel model.frame.glm() function for glm() here:

https://svn.r-project.org/R/trunk/src/library/stats/R/glm.R

There is also a 2003 paper by Thomas Lumley on non-standard evaluation 
that may be helpful:

https://developer.r-project.org/nonstandard-eval.pdf

The help for the generic ?model.frame has the following text for the 
'subset' argument:

I cannot recall off-hand, using the 'subset' argument myself in ~20 
years of using R, but do seem to recall some old discussions on the 
e-mail lists, which I cannot seem to locate at present. A search via 
rseek.org may yield some benefit.

Regards,

Marc Schwartz

J C Nash wrote on 7/13/21 7:21 PM:

subset argument in nls() and possibly other functions

Thread (3 messages)