multiply each cell of a data.frame with it's weight

Question

What I want to do is embarrassing simple - nevertheless I fail.

I have a data.frame with "characters" and "numerics". One of the columns of the data.frame represents the weights.

I want to multiply every cell of the data frame with the corresponding weight (if it's a numeric).

How do I do that (best without using a nested loop).

Thank you in advance!

Example:

   c1   c2   w   
l1 abc  2    1
l2 dxf  3    0.5
l3 ghi  4    1.5

should become

   c1   c2   w   
l1 abc  2    1
l2 dxf  1.5  0.5
l3 ghi  6    1.5

Answer 1

For a reproducible example, dd is a data frame with a mixture of variable types, with W being the weights.

dd <- data.frame(G=gl(2,2), X=rnorm(4), Y=1L:4L, Z=letters[1:4], W=0.3:3.3)
num.vars <- names(dd)[sapply(dd, is.numeric)]  #select numeric variables
num.vars <- setdiff(num.vars, "W")  # remove the weight variable
dd[num.vars] <- dd[num.vars] * dd$W  # multiply

Answer 2

Vectorise!

> dat <- data.frame(c1 = c("abc","dxf","ghi"), c2 = 2:4, w = c(1,0.5,1.5))

Effectively, you want c2 * w, but we need to tell R to look inside the data frame:

> with(dat, c2 * w)
[1] 2.0 1.5 6.0

Which we can insert back into dat in a single line:

> dat <- within(dat, c3 <- c2 * w)
> dat
   c1 c2   w  c3
1 abc  2 1.0 2.0
2 dxf  3 0.5 1.5
3 ghi  4 1.5 6.0

(Replace c3 with c2 if you want to overwrite the existing c2.)

If you have more than one numeric column other than weights, a slighlty different strategy is required if you want to automate it (i.e. not tell R which columns to multiply by w).

> ## dummy data
> dat2 <- data.frame(c1 = c("abc","dxf","ghi"), c2 = 2:4, w = c(1,0.5,1.5),
                     c3 = 5:7, c4 = 3:5)
> ## select the columns we want, all numerics, but not `w`
> want <- sapply(dat2, is.numeric) & names(dat2) != "w"
> ## then use want to index into dat2
> dat2[, want] <- with(dat2, dat2[, want] * w)
> dat2
   c1  c2   w   c3  c4
1 abc 2.0 1.0  5.0 3.0
2 dxf 1.5 0.5  3.0 2.0
3 ghi 6.0 1.5 10.5 7.5