[R] Help using mapply to run multiple models

[David Winsemius]

On Dec 19, 2013, at 11:45 AM, William Dunlap wrote:

>> Just so I can see if I understand ... that is because `as.name` will evaluate `modelType`
>> whereas as.name("modelType") would look for the function `modelType` and not find
>> such a name in the namespace? 
> 
> Almost.  as.name(modelType) will evaluate modelType so modelType could be a
> character string or a name.   as.name itself does not do any lookups - that is eval's job.
> When eval() is given a name object it looks it up.
> 
>> So modelType needs to be a language-object and `f`
>> needs to be called with:
>> 
>> f(glm, ....) rather than f("glm", ...)
> 
> If you use as.name(modelType) then you could call f("glm",...).
> 
> f(glm, ...) does not pass a name into the function f, it passes in the object
> named "glm" (usually the function in package:stats by that name).
> as.name(glm) returns garbage.  If you wanted to be able to call
>   f(glm, predictors, response)
> you could just use
>   call[[1]] <- modelType
> in f().  I didn't recommend that because then the call attributes of glm's output
> does not look nice.  You can write code so that both f("glm",...) and f(glm,...) work
> but I usually prefer not to load up functions with so much heuristic argument
> processing (e.g., how should it deal with 'func<-"glm" ; f(func,...)' and the like).

So by the time the function `f` "saw" its arguments from a call:  `f(glm, ...) `, the name of the function would already have been removed and you would just be getting the argument list attached to the function body and as.name() would make a hash of it .... as we saw in the original portion of this question.


> 
> Bill Dunlap
> Spotfire, TIBCO Software
> wdunlap tibco.com
> 
> 
>> -----Original Message-----
>> From: David Winsemius [mailto:dwinsemius at comcast.net]
>> Sent: Thursday, December 19, 2013 11:31 AM
>> To: William Dunlap
>> Cc: Simon Kiss; r-help at r-project.org
>> Subject: Re: [R] Help using mapply to run multiple models
>> 
>> 
>> On Dec 19, 2013, at 11:10 AM, William Dunlap wrote:
>> 
>>>>>              call[[1]] <- quote(modelType) # '
>>> 
>>> makes call[[1]] the same as as.name("modelType").  You want
>>> as.name(modelType).
>> 
>> Just so I can see if I understand ... that is because `as.name` will evaluate `modelType`
>> whereas as.name("modelType") would look for the function `modelType` and not find
>> such a name in the namespace? So modelType needs to be a language-object and `f`
>> needs to be called with:
>> 
>> f(glm, ....) rather than f("glm", ...)
>> 
>> 
>>> 
>>> Bill Dunlap
>>> Spotfire, TIBCO Software
>>> wdunlap tibco.com
>>> 
>>> 
>>>> -----Original Message-----
>>>> From: Simon Kiss [mailto:sjkiss at gmail.com]
>>>> Sent: Thursday, December 19, 2013 10:56 AM
>>>> To: William Dunlap
>>>> Cc: Dennis Murphy; r-help at r-project.org
>>>> Subject: Re: [R] Help using mapply to run multiple models
>>>> 
>>>> Hello Bill, that is fantastic and it's quite a bit above what I could write. Is there a way
>> to
>>>> make the model type an argument to the function so that you can specify whether
>> one is
>>>> running glm, lm and such?
>>>> I tried to modify it by inserting an argument modelType below, but that doesn't work.
>>>> Yours, simon Kiss
>>>>> f <- function (modelType, responseName, predictorNames, data, ..., envir =
>>>> parent.frame())
>>>>>  {
>>>>>      call <- match.call()
>>>>>      call$formula <- formula(envir = envir, paste(responseName, sep = " ~ ",
>>>>>          paste0("`", predictorNames, "`", collapse = " + ")))
>>>>>              call[[1]] <- quote(modelType) # '
>>>>>      call$responseName <- NULL # omit responseName=
>>>>>      call$predictorNames <- NULL # omit 'predictorNames='
>>>>>              eval(call, envir = envir)
>>>>>  }
>>>> On 2013-12-18, at 3:07 PM, William Dunlap <wdunlap at tibco.com> wrote:
>>>> 
>>>>> f <- function (responseName, predictorNames, data, ..., envir = parent.frame())
>>>>>  {
>>>>>      call <- match.call()
>>>>>      call$formula <- formula(envir = envir, paste(responseName, sep = " ~ ",
>>>>>          paste0("`", predictorNames, "`", collapse = " + ")))
>>>>>              call[[1]] <- quote(glm) # 'f' -> 'glm'
>>>>>      call$responseName <- NULL # omit responseName=
>>>>>      call$predictorNames <- NULL # omit 'predictorNames='
>>>>>              eval(call, envir = envir)
>>>>>  }
>>>>> as in
>>>>>  z <- lapply(list(c("hp","drat"), c("cyl"), c("am","gear")), FUN=function(preds)f("carb",
>>>> preds, data=mtcars, family=poisson))
>>>>>  lapply(z, summary)
>>>> 
>>>> *********************************
>>>> Simon J. Kiss, PhD
>>>> Assistant Professor, Wilfrid Laurier University
>>>> 73 George Street
>>>> Brantford, Ontario, Canada
>>>> N3T 2C9
>>>> Cell: +1 905 746 7606
>>>> 
>>>> 
>>> 
>>> ______________________________________________
>>> 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.
>> 
>> David Winsemius
>> Alameda, CA, USA
> 

David Winsemius
Alameda, CA, USA