[R] ggplot in a function confusion!

[Dennis Murphy]

Hi:

Try this:

df<-data.frame(x=1:10,y=1:10)

plot.fun.one <- function(dff, x.var, y.var)
  print(ggplot(dff, aes_string(x = x.var, y = y.var)) + geom_point() )

plot.fun.two <- function(dff, x.var, y.var) {
   x <- names(dff)[x.var]
   y <- names(dff)[y.var]
   print(ggplot(dff, aes_string(x = x, y = y)) + geom_point())
  }

plot.fun.one(df, 'x', 'y')
plot.fun.two(df, 1, 2)

aes_string() allows you to pass an unevaluated R expression into a
call that gets evaluated at run time - see its help page for a better
explanation. Your third function is unnecessary because of the
existence of aes_string().

In plot.fun.one, the variable names are passed into aes_string() as
strings, which you can see when it is called. In the second function,
the column numbers are passed instead; I create two objects x and y
that find the names of those columns which are then passed into
aes_string(), just like the first function.

Questions about ggplot2 are more likely to get a quicker response on
the ggplot2 list, to which you can subscribe at
http://had.co.nz/ggplot2/ (the same place where the on-line help pages
are found).

HTH,
Dennis

On Mon, Aug 15, 2011 at 10:09 AM, Justin Haynes <jtor14 at gmail.com> wrote:
> Whats going on here?
>
> df<-data.frame(x=1:10,y=1:10)
>
> ggplot()+geom_point(data=df,aes(x=x,y=y))  ## this is the normal usage
> right?
>
> ggplot()+geom_point(data=df,aes(x=df[,1],y=df[,2]))  ## but I can also feed
> it column indices
> ggplot()+geom_point(aes(x=df[,'x'],y=df[,'y']))  ## or column names.
>
> ## but if i wrap it in a function...
>
> plot.func.one<-function(dff,x.var,y.var){
>    print(ggplot() + geom_point(aes(x=dff[,x.var],y=dff[,y.var])))
> }
>
> plot.func.two<-function(dff,x.var,y.var){
>    print(ggplot() + geom_point(data=dff,aes(x=dff[,x.var],y=dff[,y.var])))
> }
>
> plot.func.three<-function(dff,x.var,y.var){
>    print(ggplot() + geom_point(data=dff,aes(x=eval(x.var),y=eval(y.var))))
> }
>
> plot.func.one(df,1,2) ## i assume the dff not found error is happening in
> the aes call rather than the data= portion..
> plot.func.one(df,'x','y')  ## but why does it work in the global env and not
> within a function?
>
> plot.func.two(df,1,2)
> plot.func.two(df,'x','y')
>
> var.x<-'x'
> var.y<-'y'
> plot.func.three(df,var.x,var.y)  ## why does it give the error on y.var
> instead of x.var?
> plot.func.three(df,'x','y')
>
> dff<-df
> x.var<-var.x
> y.var<-var.y
>
> plot.func.one(dff,x.var,y.var)  ## now whats going on?  I assume this works
> because ggplot is looking globally rather than within the function...
> plot.func.two(dff,x.var,y.var)
> plot.func.three(dff,x.var,y.var)
>
> nothing seems to work right!  How do I plot within a function where I can
> feed the function a data.frame and the columns I want plotted?
>
> I assume this is some interesting name space issue but if you guys can
> enlighten me as to what's going on...
>
>
> Thanks,
> Justin
>
>
> P.S.  So before I sent this I dug some more and found my answer, aes_string:
>
> plot.func<-function(dff,x.var,y.var){
>    print(ggplot() + geom_point(data=dff,aes_string(x=x.var,y=y.var)))
> }
>
> plot.func(df,'x','y')
>
> works great.  But I still wouldn't mind some clarification on what's
> happening in my earlier examples.
>
>        [[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.
>