1

I´ve seen a lot of example of transformations similars but not the same yet. Hope not to be wrong.

I wopuld like to transform this DF:

      Reference   Amount    Reference.2   Amount.2
1:   20171201     100,00 €    20171204    110,00 €

To something like that:

     Reference   Amount
   1: 20171201   100,00 €
   2: 20171204   110,00 €

Thank you.

Jaap
  • 81,064
  • 34
  • 182
  • 193
Álvaro Rodríguez
  • 123
  • 8

1 Answers1

3

If you are really just dealing with pairs of columns and you don't want a "variable" or "value" column, then maybe you can just do:

matrix(c(t(df)), ncol = 2, byrow = TRUE)
##      [,1]       [,2]     
## [1,] "20171201" "100,00€"
## [2,] "20171204" "110,00€"
## [3,] "20171202" "101,00€"
## [4,] "20171205" "110,00€"

From there, convert to data.frame or data.table or tbl or whatever you prefer to work with....

But, I don't know why you wouldn't just do:

library(data.table)
melt(as.data.table(df), measure.vars = patterns("Reference", "Amount"), 
     value.name = c("Reference", "Amount"))[, variable := NULL][]
#    Reference  Amount
# 1:  20171201 100,00€
# 2:  20171202 101,00€
# 3:  20171204 110,00€
# 4:  20171205 110,00€

Sample data (from a deleted answer by @Florian):

df = read.table(text='Reference   Amount    Reference.2   Amount.2
1:   20171201     100,00€    20171204    110,00€
2:   20171202     101,00€    20171205    110,00€',header=T)
A5C1D2H2I1M1N2O1R2T1
  • 190,393
  • 28
  • 405
  • 485
  • FWIW, the matrix approach has the disadvantage that all columns are coerced to the same data type, `character` in this case. If the resulting columns are required to keep their individual data types, `melt()` with multiple measure.vars is a better choice. – Uwe Feb 01 '18 at 09:30