R: creating a named vector from variables

Question

Inside a function I define a bunch of scalar variables like this:

a <- 10
b <- a*100
c <- a + b

At the end of the function, I want to return a,b,c in a named vector, with the same names as the variables, with minimal coding, i.e. I do not want to do:

c( a = a, b = b, c = c )

Is there a language construct that does this? For example, if I simply do return(c(a,b,c)) it returns an unnamed vector, which is not what I want. I currently have a hacky way of doing this:

> cbind(a,b,c)[1,]
   a    b    c 
  10 1000 1010

Is there perhaps a better, less hacky, way?

What could be less minimal than c(a=a,b=b,c=c)? You have to specify the names and values somewhere. Or do you want it to magically realise that 'a' is called 'a'? — Spacedman
– Spacedman, Commented Feb 18, 2011 at 15:01
Yes that's what I want -- e.g. the cbind above does that, and I'm quite happy with its minimality, but I wondered if there was a more straightforward way. Variable names could be long... and I want to save on typing :) — Prasad Chalasani
– Prasad Chalasani, Commented Feb 18, 2011 at 15:05
you want to save on typing, but you obscure your code (if I see such a construct, I get it out of the code asap) and you lose quite some performance when dealing with big and/or many vectors. I'd really think again whether you're not finding a solution to something that can be solved in a different way. — Joris Meys
– Joris Meys, Commented Feb 18, 2011 at 15:34
Why are you concerned with saving on typing within a function? You only write it once. I would encourage you to instead focus on execution speed and clarity. — Joshua Ulrich
– Joshua Ulrich, Commented Feb 18, 2011 at 16:28
@Joshua I agree with you in general. I was just bothered by the fact that cbind() preserves variable names and c() doesn't, so I was checking if there was maybe a simple way to acheive this. — Prasad Chalasani
– Prasad Chalasani, Commented Feb 18, 2011 at 18:24

hadley · Accepted Answer · 2011-02-18 15:41:00Z

15

Here's a function to do that for you, which also allows you to optionally name some of the values. There's not much to it, except for the trick to get the unevaluated expression and deparse it into a single character vector.

c2 <- function(...) {
  vals <- c(...)

  if (is.null(names(vals))) {
    missing_names <- rep(TRUE, length(vals))
  } else {
    missing_names <- names(vals) == ""
  }
  if (any(missing_names)) {
    names <- vapply(substitute(list(...))[-1], deparse, character(1))
    names(vals)[missing_names] <- names[missing_names]
  }

  vals
}

a <- 1
b <- 2
c <- 3

c2(a, b, d = c)
# a b d 
# 1 2 3

Note that it's not guaranteed to produce syntactically valid names. If you want that, apply the make.names function to the names vector.

c2(mean(a,b,c))
# mean(a, b, c) 
#            1

Also, as with any function that uses substitute, c2 is more suited for interactive use than to be used within another function.

answered Feb 18, 2011 at 15:41

hadley

104k35 gold badges186 silver badges248 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Tommy Over a year ago

@hadley: substitute(list(...))[-1] can be shortened to with substitute(...())

hadley Over a year ago

Shortened, but made even more confusing! (I realise list isn't used, but it conveys the intent)

krlmlr Over a year ago

@hadley: I have added two functions nc and nlist to my kimisc package, crediting you and this answer.

Collectives™ on Stack Overflow

R: creating a named vector from variables

1 Answer 1

3 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related