[R] Variable Column Names and Lists

[Duncan Murdoch]

On 27/02/2015 1:18 PM, Brian Trautman wrote:
> This should be a simple question, but I am at my wits end.
> 
> dt<-data.table(a=rep(1:10, 26), b=1:260, c=rep(1:2, 130))
> 
> sumvar <- 'mysum'
> bvar <- 'b'
> 
> dt_min <- dt[, list(sumvar = sum(get(bvar))), by=list(a)]
> print(dt_min)
> 
> I want the function to return two variables, "a" and "mysum".  However, it
> instead returns "a" and "sumvar", rather than evaluating "sumvar".
> 
> If I try replacing it with "get(sumvar)" or "eval(sumvar)", R just throws
> an error.  What am I missing?
> 
>      a sumvar
>  1:  1   3276
>  2:  2   3302
>  3:  3   3328
>  4:  4   3354
>  5:  5   3380
>  6:  6   3406
>  7:  7   3432
>  8:  8   3458
>  9:  9   3484
> 10: 10   3510

I don't know the data.table function, but

 list(sumvar = sum(get(bvar)))

will produce a list with one element named sumvar, holding the sum of
values of a variable named 'b'.  If you want that element to be named
'mysum', you could change it at the end, or construct that list as

structure(list(sum(get(bvar))), names=sumvar)

i.e. the full expression would be

dt_min <- dt[, structure(list(sum(get(bvar))), names=sumvar), by=list(a)]

The idea is that structure() produces list(sum(get(bvar))) with
attribute "names" set to the contents of sumvar.

Duncan Murdoch