ave function in R: First argument a vector - r

Question

I'm trying to use the following code in R:
ID=seq(1,11)
g=c(1,2,3,1,1,2,3,4,4,1,3)
x <- sample(11)
d <- data.frame(ID,g, x)
Ranking_Categoria<-function(d,var,category)
{
d$rank<-ave(d$var,d$category,FUN=rank)
return(d)
}
and I get the following error message:
Error in split.default(x, g) : first argument must be a vector.
Variables var and category (character) are columns of the dataframe d that user needs to specify in order to get the desired result. I need to refer to this names when I use the function ave() as you can see.

Answer 1

You need to use [[ to get the var and category columns by name:
Ranking_Categoria<-function(d,var,category)
{
d$rank<-ave(d[[var]],d[[category]],FUN=rank)
return(d)
}
... because d$var tries to get the column called "var", and there is none.
UPDATE
> Ranking_Categoria(d, "x", "g")
ID g x rank
1 1 1 10 3
2 2 2 9 2
3 3 3 4 1
4 4 1 11 4
5 5 1 1 1
6 6 2 8 1
7 7 3 6 2
8 8 4 2 1
9 9 4 5 2
10 10 1 3 2
11 11 3 7 3

Answer 2

The best solution would be not to use names at all:
Ranking_Categoria<-function(d,var,category)
{
d$rank<-ave(var,category,FUN=rank)
return(d)
}
Then call it as
Ranking_Categoria(d,d$x,d$g)
The reason why the function in your question didn't work as you thought it would is partially because R's syntax and DWIM-ness for string manipulation sucks. Here's a hacky, fragile solution using eval and parse:
Ranking_Categoria<-function(d,var,category)
{
string=paste('d$rank<-ave(d$',var,',d$',category,',FUN=rank)',sep="")
eval(parse(text=string))
return(d)
}
However, you still have to call it as
Ranking_Categoria(d,"x","g")
And if you already have objects with the names of x and g, then may the gods help you if you try to do Ranking_Categoria(d,x,g)... Crap like this is why I've gone from using Perl and R equally to sticking with Perl (my first and native programming language) and using R only when necessary.