thr3ads.net - R help - [R] Passing the name of a variable to a function [Aug 2010]

If this information is useful, please help other people find it:
Share via:

Anthony Staines

2010-Aug-04 13:56 UTC

[R] Passing the name of a variable to a function

Dear colleagues,

I have a problem which has bitten me occasionally. I often need to
prepare graphs for many variables in a data set, but seldom for all.
or for any large number of sequential or sequentially named variables.
Often I need several graphs for different subsets of the dataset
for a given variable.  I run into similar problems with other needs
besides graphing.

 What I would like to do is something like "write a function which
takes the *name* of a variable, presumably a s a character string,
from a dataframe, as one argument, and the dataframe, as a second
argument".

For example, where y is to be the the name of a variable in a given
dataframe d, and the other variables needed, T, M and so on, are
to be found in the same dataframe :-

pf <- function (y,data,...) {
p1 <- xyplot(y~x|T,data)
p2 <- xyplot(y~x|T,subset(data,M == 2))
p3 <- xyplot(y~x|T,subset(data,M == 4))
#print(p1,p2,p3....)
}
 pf(Score1,data)
pf(Score2,data)


This fails, because, of course, Score 1, Score 2 etc.. are  not
defined, or if you pass them as pf(data$Score2,data), then when you
subset the
data, data$Score2 is now the wrong shape.  I've come up with various
inelegant hacks, (often with for loops), for getting around this over
the
last few years, but I can't help feeling that I'm missing something
obvious, which I've been too dim to spot.

Please note the answer to my requirements is *not* a clever use of
lattice. This question goes beyond graphs.
All suggestions gratefully received!

Best wishes,
Anthony Staines
-- 
Anthony Staines,? Professor of Health Systems Research,
School? of Nursing, Dublin City University, Dublin 9, Ireland.
Tel:- +353 1 7007807. Mobile:- +353 86 606 9713

Erik Iverson

2010-Aug-04 19:09 UTC

head link

[R] Passing the name of a variable to a function

Hello,

> I have a problem which has bitten me occasionally. I often need to
> prepare graphs for many variables in a data set, but seldom for all.
> or for any large number of sequential or sequentially named variables.
> Often I need several graphs for different subsets of the dataset
> for a given variable.  I run into similar problems with other needs
> besides graphing.
> 
>  What I would like to do is something like "write a function which
> takes the *name* of a variable, presumably a s a character string,
> from a dataframe, as one argument, and the dataframe, as a second
argument".
> 
> For example, where y is to be the the name of a variable in a given
> dataframe d, and the other variables needed, T, M and so on, are
> to be found in the same dataframe :-
> 
> pf <- function (y,data,...) {
> p1 <- xyplot(y~x|T,data)
> p2 <- xyplot(y~x|T,subset(data,M == 2))
> p3 <- xyplot(y~x|T,subset(data,M == 4))
> #print(p1,p2,p3....)
> }
>  pf(Score1,data)
> pf(Score2,data)
> 
> 
> This fails, because, of course, Score 1, Score 2 etc.. are  not
> defined, or if you pass them as pf(data$Score2,data), then when you
> subset the
> data, data$Score2 is now the wrong shape.  
> I've come up with various
> inelegant hacks, (often with for loops), for getting around this over
> the
> last few years, but I can't help feeling that I'm missing something
> obvious, which I've been too dim to spot.
Depending on your needs (e.g., you use formulas, which can be trickier),
I think I often do something like:

# I prefer this, I quote the variable name...

df1 <- data.frame(x = rnorm(100),
                   score1 = rnorm(100),
                   M = sample(c(2, 4), 100, replace = TRUE))

pf <- function (y,data,...) {
   data$y <- data[[y]]
   xyplot(y~x, subset(data, M == 2))
}

pf("score1", df1)

# as an alternative, use eval/substitute, don't have to quote

pf2 <- function (y,data,...) {
   data$y <- eval(substitute(y), data)
   xyplot(y~x, subset(data, M == 2))
}

pf2(score1, df1)

Joshua Wiley

2010-Aug-04 19:28 UTC

head link

[R] Passing the name of a variable to a function

Hi Anthony,

I don't know if this will help you.  A similar method should work for
any function that accepts a formula, which is your main issue I think.
 Outside of formulae, you should just be able to pass the arguments
directly (as with the data argument of xyplot).

The only other thought that comes to mind involves a combination of
eval() and parse(), but its obvious messiness and fortune(106),
suggest that that is not worth sharing.

#Some data to work with
mydata <- data.frame(x = 1:10, Score1 = 1:10,
                     T = rep(c(1,2), 5), M = rep(c(2,4), each = 5))

#Function
pf <- function (y, data, ...) {
  #Save the old par() settings, and restore at the end
  old.par <- par(no.readonly = TRUE)
  on.exit(par(old.par))
  #Set it so it asks before changing graphs
  par(ask = TRUE)
  #Create a text string of your formula
  my.f <- paste(y, " ~ x|T", sep = "")

  #Initialize a list and calculate plots
  p <- vector("list", length = 3)
  p[[1]] <- xyplot(x = formula(my.f),
                   data = data)
  p[[2]] <- xyplot(x = formula(my.f),
                   data = subset(x = data, subset = M == 2))
  p[[3]] <- xyplot(x = formula(my.f),
                   data = subset(x = data, subset = M == 4))

  #Graph the plots
  for(i in 1:length(p)) {
    print(p[[i]])
  }
}

#Graphs!
pf(y = "Score1", data = mydata)

HTH,

Josh

On Wed, Aug 4, 2010 at 6:56 AM, Anthony Staines
<anthony.staines at gmail.com> wrote:> Dear colleagues,
>
> I have a problem which has bitten me occasionally. I often need to
> prepare graphs for many variables in a data set, but seldom for all.
> or for any large number of sequential or sequentially named variables.
> Often I need several graphs for different subsets of the dataset
> for a given variable. ?I run into similar problems with other needs
> besides graphing.
>
> ?What I would like to do is something like "write a function which
> takes the *name* of a variable, presumably a s a character string,
> from a dataframe, as one argument, and the dataframe, as a second
argument".
>
> For example, where y is to be the the name of a variable in a given
> dataframe d, and the other variables needed, T, M and so on, are
> to be found in the same dataframe :-
>
> pf <- function (y,data,...) {
> p1 <- xyplot(y~x|T,data)
> p2 <- xyplot(y~x|T,subset(data,M == 2))
> p3 <- xyplot(y~x|T,subset(data,M == 4))
> #print(p1,p2,p3....)
> }
> ?pf(Score1,data)
> pf(Score2,data)
>
>
> This fails, because, of course, Score 1, Score 2 etc.. are ?not
> defined, or if you pass them as pf(data$Score2,data), then when you
> subset the
> data, data$Score2 is now the wrong shape. ?I've come up with various
> inelegant hacks, (often with for loops), for getting around this over
> the
> last few years, but I can't help feeling that I'm missing something
> obvious, which I've been too dim to spot.
>
> Please note the answer to my requirements is *not* a clever use of
> lattice. This question goes beyond graphs.
> All suggestions gratefully received!
>
> Best wishes,
> Anthony Staines
> --
> Anthony Staines,? Professor of Health Systems Research,
> School? of Nursing, Dublin City University, Dublin 9, Ireland.
> Tel:- +353 1 7007807. Mobile:- +353 86 606 9713
>
> ______________________________________________
> 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.
>


-- 
Joshua Wiley
Ph.D. Student, Health Psychology
University of California, Los Angeles
http://www.joshuawiley.com/

Anthony Staines

2010-Aug-04 19:43 UTC

head link

[R] Passing the name of a variable to a function

Thanks to Joshua and Erik for very helpful suggestions. This clarifies
what I had found to be a tricky area of the documentation!
Best wishes,
Anthony Staines

-- 
Anthony Staines,? Professor of Health Systems Research,
School? of Nursing, Dublin City University, Dublin 9, Ireland.
Tel:- +353 1 7007807. Mobile:- +353 86 606 9713

Liviu Andronic

2010-Aug-04 20:24 UTC

head link

[R] Passing the name of a variable to a function

Dear Anthony

On Wed, 4 Aug 2010 14:56:58 +0100
Anthony Staines <anthony.staines at gmail.com>
wrote:>  What I would like to do is something like "write a function which
> takes the *name* of a variable, presumably a s a character string,
> from a dataframe, as one argument, and the dataframe, as a second
> argument".
> I am not sure that this is what you're looking for, but I was working
on this all day and am pretty happy with the result. The following code
takes a list of character variable names x.vars, apply-ies the function
to each variable .var in data frame .data (with optional subset .subs)
and stores the results in a list names-ed after the original
variables list. 

Best regards
Liviu

my.fun <- function(.var, .data, .subs) {
    require(lattice)
    res <- list()
    x <- .data[[.var]][.subs]
    res <- bwplot(x, main=.var)
    return(res)
}

x.vars <- as.list(names(mtcars)[2:5])
names(x.vars) <- names(mtcars)[2:5]
x.vars  ##there's probably a nicer way of doing this

x <- lapply(x.vars, my.fun, .data=mtcars)
x  ##if you use Sweave, don't forget to print(x)

Maybe Matching Threads

Search for more reasonably related threads

R help - Aug 2010 - Passing the name of a variable to a function

[R] Passing the name of a variable to a function

[R] Passing the name of a variable to a function

[R] Passing the name of a variable to a function

[R] Passing the name of a variable to a function

[R] Passing the name of a variable to a function

Maybe Matching Threads