[R] a weighted average by group

[Gabor Grothendieck]

On 11/17/05, Weiwei Shi <helprhelp at gmail.com> wrote:
> Hi,
> I have a data.frame a like this:
> > a
>   v1 v2 v3 v4
> 1   1  1  A  X
> 2   2  2  A  X
> 3   3  3  A  X
> 4   4  4  B  X
> 5   5  5  B  X
> 6   6  6  C  Y
> 7   7  7  C  Y
> 8   8  8  C  Y
> 9   9  9  D  Y
> 10 10 10  D  Y
>
> I want to get a weighted average for each combination of v3 and v4,
> using v1 as values and v2 as weights i.e.,
> for A and X, the final result should be (1*1+2*2+3*3)/(1+2+3)
>
> not sure which function I should use in R ?


All our solutions will make use of this function f:

f <- function(i) weighted.mean(a$v1[i], a$v2[i])


# Here are three solutions that only give combos
# that are already present in the data:

aggregate(rownames(a), a[,3:4], f)

aggregate(1:nrow(a), a[,3:4], f)

sapply(split(1:nrow(a), a[,3:4], drop = TRUE), f)

# and here are two solutions that give all combos
# returning results in different ways:

tapply(1:nrow(a), a[,3:4], f)

sapply(split(1:nrow(a), a[,3:4]), f)

Aside:
I found it strange that rownames only works in the
first case.