R: Accessing and using names of data.frames within a list during an `apply` function

Question

I'm trying to write a function that takes a list of data.frames (somelist) and writes them out to .csv files, each named the same as the data.frame being written (i.e., a1.csv, a2.csv, a3.csv).

somelist <- list(a1 = data.frame(a = 1, b = 2, c = 3),
                 a2 = data.frame(d = 4, e = 5, f = 6),
                 a3 = data.frame(g = 7, h = 8, i = 9))

csvout <- function (y) {
    z <- deparse(substitute(y))
    write.table(y, file = paste0("~/somefolder/",
                                 z,
                                 ".csv"),
                sep = ",",
                row.names = FALSE)
    print(z)
}

sapply(somelist, csvout)

That is the closest I get, and instead, the files are named what is printed for z:

[1] "X[[1L]]"
[1] "X[[2L]]"
[1] "X[[3L]]"
       a1        a2        a3 
"X[[1L]]" "X[[2L]]" "X[[3L]]"

In searching the documentation for an answer, I think I'm on the right track here in ?sapply, but I've been unable to connect the dots:

For historical reasons, the calls created by lapply are unevaluated, and code has been written (e.g., bquote) that relies on this. This means that the recorded call is always of the form FUN(X[[i]], ...), with i replaced by the current (integer or double) index

Update

I can use a for loop using syntax similar to C programming...

for (i in 1:length(somelist)) {
    write.table(i, file = paste0("~/somefolder/",
                                 names(somelist)[i],
                                 ".csv"),
                sep = ",",
                row.names = FALSE)
}

But I'm trying to code more natively in R and I'm aware that the apply family, or at least using a vectorized for loop (i.e., for (all in somelist) { instead of for (i in 1:length(somelist)) {) is preferable for efficiency and proper coding etiquette.

You FOR-loop solution is missing "somelist[i]" in the second line. — Juergen, Mar 08 '15 at 16:46
It worked for me. In this instance, I removed the extra function `csvout` and just piped the data.frames in to the `write.table` function directly. — mcjudd, Mar 08 '15 at 17:20

Juergen · Accepted Answer · 2015-03-08T16:38:41.053

This is really, really, really dirty, but I think it works as you described. Of course for more than nine data.frames it needs adjustment in the substitutes part.

csvout <- function (y, csvnames) {
  write.table(y, file = paste0("test",
                                   csvnames[as.numeric(substr(deparse(substitute(y)),4,4))],
                               ".csv"),
              sep = ",",
              row.names = FALSE)
}

sapply(somelist, FUN=csvout, names(somelist))

I suppose you know that, but if you implemented a FOR-loop instead of sapply this would be much easier because you could directly reference the data.frame names with the names function.

Edit: This is the FOR-loop solution which works no matter how many data.frames you've got:

csvout <- function (y) {
  for (i in 1:length(y)){
    write.table(y[i], file = paste0("test",
                                 names(y)[i],
                                 ".csv"),
                sep = ",",
                row.names = FALSE)  
  }  
}

csvout(somelist)

Thanks @Juergen, can you append your answer with the `for` loop alternative? — mcjudd, Mar 08 '15 at 16:24
Added it. Although I see you already found the solution. I'm not aware of something like "for all in something" / For-each-loop in R. Usually you make use of (s/l)apply. The only drawback known to me is the problematic you described regarding the names. For more information on workarounds see this thread and especially this solution: http://stackoverflow.com/a/10959033/4166885. In your case, you could add another attribute to the data.frame, e.g. rowIndex. I tried it and it worked. — Juergen, Mar 08 '15 at 16:43

score 1 · Answer 2 · answered Mar 08 '15 at 16:40

1

You can use apply in similar fashion like apply(1:length,function(i){})

answered Mar 08 '15 at 16:40

Lev Kuznetsov

3,520
5
20
33

Thanks for this -- was unaware of this functionality. I've got a small-ish list of data.frames to write, so efficiency is not a big issue, but are you familiar with the relative efficiencies of using `for (i in length:x) {}` vs `apply(1:length, function(i) {})` when performing lots of calculations, or memory-intensive calculations? – mcjudd Mar 08 '15 at 17:18
1

I was hoping someone would chime in since I'm only speculating, but as I see it the only difference is `apply ` is deferred and `for ` is forced. So apply will only calculate the fields actually used and for will calculate all – Lev Kuznetsov Mar 09 '15 at 12:42

R: Accessing and using names of data.frames within a list during an `apply` function

2 Answers2