Make dummy variables from multiple columns and create new dataframe

Question

I need to make a dataframe of dummies from survey data where respondents have stated words in several columns in a dataframe. Here is a simplified example to illustrate what I need to do? A single word is here represented by a letter.

id <- c(1:6)
v.1 <- c("a","b","d","e","a","c")
v.2 <- c("b","a","a","a","b","a")
v.3 <- c("e","c","b","b","e","b")

df <- data.frame(id,v.1,v.2,v.3)

> df
  id v.1 v.2 v.3
1  1   a   b   e
2  2   b   a   c
3  3   d   a   b
4  4   e   a   b
5  5   a   b   e
6  6   c   a   b

Here is my wanted output?

> print(df.dummy)
   id  a b c d e
1:   1 1 1 0 0 1
2:   2 1 1 1 0 0
3:   3 1 1 0 1 0
4:   4 1 1 0 0 1
5:   5 1 1 0 0 1
6:   6 1 1 1 0 0

Any ideas?

What do v.X represent? How do you want these reducing into your desired output? Do you need a count, i.e. if `id 1` row was `"a", "b", "a"` would you want a to be 2? — MikeRSpencer, Sep 21 '15 at 12:04
the content in v.X are single words so they are character vectors. — jonas, Sep 21 '15 at 12:06
the respondents can only choose one word so no need for summing or counting.. — jonas, Sep 21 '15 at 12:07
Have a look at Frank's response for what I would have written! — MikeRSpencer, Sep 21 '15 at 12:12
@BenBolker It's not obvious to me where these previous answers can be found. Did you forget to mark a dupe? — Frank, Sep 21 '15 at 12:26

score 7 · Accepted Answer · edited Sep 21 '15 at 12:48

7

You can use reshape2:

library(reshape2)
dcast( melt(df,id.var="id"), id ~ value, length)

which gives

  id a b c d e
1  1 1 1 0 0 1
2  2 1 1 1 0 0
3  3 1 1 0 1 0
4  4 1 1 0 0 1
5  5 1 1 0 0 1
6  6 1 1 1 0 0

Or use recast to "melt and cast in a single step":

recast(df, id ~ value, id.var = "id", length)

Without using a package, you could do table( rep(df$id,ncol(df)-1), unlist(df[-1]) ).

edited Sep 21 '15 at 12:48

Henrik

65,555
14
143
159

answered Sep 21 '15 at 12:11

Frank

66,179
8
96
180

2

I was just about to say reshape2 can do it... good answer! – MikeRSpencer Sep 21 '15 at 12:12

Make dummy variables from multiple columns and create new dataframe

1 Answers1