Summarizing by subgroup percentage in R

Question

I have a dataset like this:

df = data.frame(group = c(rep('A',4), rep('B',3)),
                subgroup = c('a', 'b', 'c', 'd', 'a', 'b', 'c'),
                value = c(1,4,2,1,1,2,3))


group | subgroup | value
------------------------
  A   |    a     |  1
  A   |    b     |  4
  A   |    c     |  2
  A   |    d     |  1
  B   |    a     |  1
  B   |    b     |  2
  B   |    c     |  3

What I want is to get the percentage of the values of each subgroup within each group, i.e. the output should be:

group | subgroup | percent
------------------------
  A   |    a     |  0.125
  A   |    b     |  0.500
  A   |    c     |  0.250
  A   |    d     |  0.125
  B   |    a     |  0.167
  B   |    b     |  0.333
  B   |    c     |  0.500

Example for group A, subgroup A: the value was 1, the sum of the whole group A is 8 (a=1, b=4, c=2, d=1) - hence 1/8 = 0.125

So far I've only found fairly simple aggregates like this, but I cannot figure out how to do the "divide by a sum within a subgroup" part.

Adara

Answer 1

Per your comment, if the subgroups are unique you can do

library(dplyr)
group_by(df, group) %>% mutate(percent = value/sum(value))
#   group subgroup value   percent
# 1     A        a     1 0.1250000
# 2     A        b     4 0.5000000
# 3     A        c     2 0.2500000
# 4     A        d     1 0.1250000
# 5     B        a     1 0.1666667
# 6     B        b     2 0.3333333
# 7     B        c     3 0.5000000

Or to remove the value column and add the percent column at the same time, use transmute

group_by(df, group) %>% transmute(subgroup, percent = value/sum(value))
#   group subgroup   percent
# 1     A        a 0.1250000
# 2     A        b 0.5000000
# 3     A        c 0.2500000
# 4     A        d 0.1250000
# 5     B        a 0.1666667
# 6     B        b 0.3333333
# 7     B        c 0.5000000

Mather

Answer 2

We can use prop.table to calculate percentage/ratio.

Base R :

transform(df, percent = ave(value, group, FUN = prop.table))

#  group subgroup value percent
#1     A        a     1   0.125
#2     A        b     4   0.500
#3     A        c     2   0.250
#4     A        d     1   0.125
#5     B        a     1   0.167
#6     B        b     2   0.333
#7     B        c     3   0.500

dplyr :

library(dplyr)
df %>% group_by(group) %>% mutate(percent = prop.table(value))

data.table :

library(data.table)
setDT(df)[, percent := prop.table(value), group]

Noyes