data.table subsetting by variable containing a column name

Question

I have a data table whose column names are not known in advance

set.seed(1)
titles <- rep(letters[sample.int(3,replace = T)],2)
dt <- data.table( x = c(1, 1, 3, 1, 3, 1), y = c(1, 2, 1, 2, 1, 2) )
dt = dcast(dt, x+y ~ titles, fill=0, value.var = 'x')
> dt
   x y a b
1: 1 1 1 0
2: 1 2 1 2
3: 3 1 0 2

I wish to remove rows whose last column is 0, but 'subset' doesn't work. Why?

dt <- subset( dt, last(titles) > 0 )

I'd preferrably use data.table methods.

@akrun he doesn't know it's `b`. I guess one way is `dt[dt[[last(titles)]] > 0]`, but it's a good question on what's the idiomatic way to do this. — David Arenburg, Jan 20 '16 at 11:48
Then `dt[get(tail(names(dt),1))!=0]` or `dt[get(tail(titles,1))!=0]` — akrun, Jan 20 '16 at 11:49
I would guess [this](http://stackoverflow.com/questions/27677283/evaluating-both-column-name-and-the-target-value-within-j-expression-within-d) somewhat related — David Arenburg, Jan 20 '16 at 11:55
The link David Arenburg directs to, suggest also dt[,.SD[[last(titles)]] — Amitai, Jan 20 '16 at 14:38

score 1 · Accepted Answer · answered Jan 20 '16 at 11:52

1

We can use get

 dt[get(last(titles))!=0]

answered Jan 20 '16 at 11:52

akrun

874,273
37
540
662

data.table subsetting by variable containing a column name

1 Answers1