Question

I have a data frame, and make selections based on some of the factors. I want a vector of names, created from the factor levels. Hopefully this suffices to show the intent:

test.results <- list(
  First  = factor(c('A', 'B'), levels=c('A', 'B', 'C')),
  Second = factor(c('E', 'F'), levels=c('E', 'F', 'G')),
  Third  = factor(c('X', 'Y'), levels=c('X', 'Y', 'Z'))
  )

# cols <-  c('First', 'Third'); TestName(test.results, cols) should return c('A X', 'B Y')

Here is an implementation. Is there a way to avoid the explicit ‘for’ loop?

TestName <- function(X, cols) {
  result <- character(length(cols))
  space <- '';
  for (i in cols) {
    result <- paste0(result, space, X[[i]]);
    space <- ' ';
  }
  return(result);
}

Answer 1

Your data is not a data.frame in the example, but nevermind, the following will work regardless

paste is vectorized, so as the question stands there should be no need for *apply or for loops

testname <- function(x, .names){ do.call(paste, x[.names])}
testname(test.results, c('First','Third'))
## [1] "A X" "B Y"

You could add checks about whether x is a list and that names exist in x.

EDIT — allowing sep to be set (or other variables) if you wished.

testname <- function(x, .names,...){ do.call(paste, c(x[.names], list(...)))}
testname(test.results, c('First','Third'), sep = '---')
## "A---X" "B---Y"

If your data was a data.table then you could do the following

library(data.table)
DT <- as.data.table(test.results)

DT[, paste(First, Third)]

Or you could just stick with lists and data.frames and use with or evalq

evalq(paste(First,Third), test.results)

or

with(test.results, paste(First, Third))