Repository navigation
dcast naming: 1 vs. 2+ value.var #6582
Description
Suppose I have a datatable with columns id, part, etc. & want to widen it with id ~ part.
- If there is a single
value.varcolumn, the input & output look like this:
id part foo
<num> <num> <int>
1: 1 1 1
2: 1 2 2
3: 2 1 3
id 1 2
<num> <int> <int>
1: 1 1 2
2: 2 3 NA
- If there are multiple
value.varcolumns, the input & output look like this:
id part foo bar
<num> <num> <int> <int>
1: 1 1 1 4
2: 1 2 2 5
3: 2 1 3 6
id foo_1 foo_2 bar_1 bar_2
<num> <int> <int> <int> <int>
1: 1 1 2 4 5
2: 2 3 NA 6 NA
For interactive use this is fine, but when I'm using this code in an automated pipeline, I find myself wishing for a consistent treatment of column names, i.e. that the first result would be:
id foo_1 foo_2
<num> <int> <int>
1: 1 1 2
2: 2 3 NA
Could we switch to this treatment or provide an argument for doing so?
library(data.table) # 1.16.2
dt1 <- data.table(id = c(1, 1, 2), part = c(1, 2, 1), foo = 1:3)
dcast(dt1, id ~ part, value.var = 'foo')
dt1[, bar := 4:6]
dcast(dt1, id ~ part, value.var = c('foo', 'bar'))
Reacted by iagogv3
Activity
[-]dcast naming [/-][+]dcast naming: 1 vs. 2+ value.var[/+]tdhock commentedon Oct 23, 2024
good question, I always thought this behavior was funny / annoying sometimes too.
tldr; I think it would be difficult to support.
also there could be a third part of the dcast names which is added when there is more than one
fun.aggregate,I think these differences in names may be annoying in some cases, but I also suspect that always outputting full three /two part names could be annoying in some cases too.
tdhock commentedon Oct 23, 2024
here is another funny inconsistency involving dcast output names
tdhock commentedon Jun 8, 2025
duplicate of #2364