rdataframesumcombinationscombn

comb with sum function across the columns of a dataframe

I have the dataframe DATA with the first 5 rows shown:

>dput(DATA)
structure(list(e_1 = c(21.8, 21.8, 21.8, 21.8, 21.8), 
e_2 = c(9.8, 9.8, 9.8, 9.8, 9.8), e_3 = c(-2.2, -2.2, -2.2, 
-2.2, -2.2), e_4 = c(-14.2, -14.2, -14.2, -14.2, -14.2), 
e_5 = c(0, 0, 0, 0, 0)), row.names = c(NA, 5L), class = "data.frame")

I want to use combn to get the sum of all combinations of three of the columns of the dataframe across the columns. I use the following:

library(utils)
combn(DATA,3,sum)

The results are over all the rows and I get:

 [1] 147  87 158  27  98  38 -33  38 -22 -82

The desired output is the same number of rows as the dataframe DATA. For each row, give the sum the values in all possible combinations of 3 of the 5 columns in DATA. such as:

[1]  29.4  17.4  31.6   5.4  19.6   7.6  -6.6   7.6  -4.4 -16.4
[11]  29.4  17.4  31.6   5.4  19.6   7.6  -6.6   7.6  -4.4 -16.4
[21]  29.4  17.4  31.6   5.4  19.6   7.6  -6.6   7.6  -4.4 -16.4
[31]  29.4  17.4  31.6   5.4  19.6   7.6  -6.6   7.6  -4.4 -16.4
[41]  29.4  17.4  31.6   5.4  19.6   7.6  -6.6   7.6  -4.4 -16.4
Solution

  • To apply over the rows of your df you can use apply:

    d <- structure(list(e_1 = c(21.8, 21.8, 21.8, 21.8, 21.8), 
               e_2 = c(9.8, 9.8, 9.8, 9.8, 9.8), e_3 = c(-2.2, -2.2, -2.2, 
                                                         -2.2, -2.2), e_4 = c(-14.2, -14.2, -14.2, -14.2, -14.2), 
               e_5 = c(0, 0, 0, 0, 0)), row.names = c(NA, 5L), class = "data.frame")

t(apply(d, 1, combn, 3, sum))
#>   [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10]
#> 1 29.4 17.4 31.6  5.4 19.6  7.6 -6.6  7.6 -4.4 -16.4
#> 2 29.4 17.4 31.6  5.4 19.6  7.6 -6.6  7.6 -4.4 -16.4
#> 3 29.4 17.4 31.6  5.4 19.6  7.6 -6.6  7.6 -4.4 -16.4
#> 4 29.4 17.4 31.6  5.4 19.6  7.6 -6.6  7.6 -4.4 -16.4
#> 5 29.4 17.4 31.6  5.4 19.6  7.6 -6.6  7.6 -4.4 -16.4