Question

so this is a simple question, and it’s unclear to me what’s going wrong with this plyr call.

library(plyr)

some silly data with a binomial outcome (y1) and two four-level classifying factors(x1 and x2):

df <- data.frame(x1 = sample(letters[1:4], 4000,T),
                 x2 = sample(LETTERS[5:8], 4000,T),
                 y1 = rbinom(n=4000,1,.5))

I want a table with the row proportions of each outcome–and i thought the following would work

foo <- ddply(df, .(x1,x2), function(i) prop.table(table(i$y1),1))

but instead it gives me:

head(foo)

  x1 x2 0 1
1  a  E 1 1
2  a  F 1 1
3  a  G 1 1
4  a  H 1 1
5  b  E 1 1
6  b  F 1 1

which is clearly not what i’m looking for. What’s my omission?

Answer 1

try

foo<-ddply(df, .(x1,x2), function(i) prop.table(table(i$y1)))

> head(foo)
  x1 x2         0         1
1  a  E 0.5365854 0.4634146
2  a  F 0.4936170 0.5063830
3  a  G 0.5176471 0.4823529
4  a  H 0.4568966 0.5431034
5  b  E 0.4780702 0.5219298
6  b  F 0.5617530 0.4382470

not sure if thats what you want