CODEWITHSUNDEEP
rdataframedummy-variable
bfoste01
bfoste01

Reputation: 359

Creating an unorthodox dummy variable

I need to create some unorthodox dummy variables and I am having some trouble. Essentially in my dataset each teacher can teach multiple classes. I'm building a multilevel dataset, so it is ok that there are duplicate teacher IDs.

Here is an example of the data:

#generate data
teacher.id <- c(1:5, 1:5)
class.taught <- c("ELA", "Math", "Science", "ELA", "Math", "Science", "Math", "ELA", "ELA", "Math")

# combine into data frame
dat <- data.frame(teacher.id, class.taught)

As you can see teachers with IDs 1 and 3 both teach 2 different classes.

The conventional approach to creating dummy variables yields: 

# example of what I have done so far 
dat$teach.ELA <- ifelse(dat$class.taught == "ELA", 1, 0 )
dat$teach.MATH <- ifelse(dat$class.taught == "Math", 1, 0 )
dat$teach.SCIENCE <- ifelse(dat$class.taught == "Science", 1, 0 )
dat

However, here is how I would like the new dummy variables to look: 

desired.ELA <- c(1,0,1,1,0,1,0,1,1,0)
desired.MATH <- c(0,1,0,0,1,0,1,0,0,1)
desired.SCIENCE <- c(1,0,1,0,0,1,0,1,0,0)
dat.2 <- data.frame(dat, desired.ELA, desired.MATH, desired.SCIENCE)
dat.2

My hunch is that I need to loop through the ids to create these, but past that I really don't see my avenue to accomplish what I desire.

Upvotes: 2

Views: 87

Answers (4)

David_B
David_B

Reputation: 936

You can also do this using %in%:

dums <- function(dt, x){
  ix <- dt[, 2] %in% x
  dt[, 1] %in% unique(dt[ix, 1])
}

dums(dat, 'ELA')
dums(dat, 'Math')
dums(dat, 'Science')

This gives you TRUE/FALSE rather than 0/1 vectors, but as.integer will convert them to 0/1 if necessary.

Upvotes: 3

lmo
lmo

Reputation: 38520

Here is a base R method. The idea is that you create the dummies for each teacher and then merge these onto the original data:

# get dummies for each teacher
temp <- as.data.frame(with(dat, table(teacher.id, class.taught) > 0))
temp$teacher.id <- as.integer(row.names(temp))

# merge onto dataset
merge(dat, temp, by="teacher.id")

You could coerce the logicals to integer if it really bugged you, but R will do all that work for you.

Upvotes: 4

Mhairi McNeill
Mhairi McNeill

Reputation: 1991

I'd use dplyr and tidyr.

library(dplyr)
library(tidyr)

dummies <- 
dat %>%
   group_by(teacher.id, class.taught) %>%
   summarise(is_taught = as.numeric(n() > 0)) %>%
   spread(class.taught, is_taught, fill = 0)

> dummies
Source: local data frame [5 x 4]

  teacher.id   ELA  Math Science
       (int) (dbl) (dbl)   (dbl)
1          1     1     0       1
2          2     0     1       0
3          3     1     0       1
4          4     1     0       0
5          5     0     1       0

You can then have them in the original data using a join. 

> inner_join(dat, dummies)
Joining by: "teacher.id"
   teacher.id class.taught ELA Math Science
1           1          ELA   1    0       1
2           2         Math   0    1       0
3           3      Science   1    0       1
4           4          ELA   1    0       0
5           5         Math   0    1       0
6           1      Science   1    0       1
7           2         Math   0    1       0
8           3          ELA   1    0       1
9           4          ELA   1    0       0
10          5         Math   0    1       0

Upvotes: 1

Edward R. Mazurek
Edward R. Mazurek

Reputation: 2177

Just for fun, using dplyr:

library(dplyr)
dat %>% left_join(
    dat %>%
      group_by(teacher.id) %>%
      summarize(desired.ELA = ifelse(sum(teach.ELA), 1, 0),
        desired.MATH = ifelse(sum(teach.MATH), 1, 0),
        desired.SCIENCE = ifelse(sum(teach.SCIENCE), 1, 0))
  )

Output:

   teacher.id class.taught teach.ELA teach.MATH teach.SCIENCE desired.ELA desired.MATH desired.SCIENCE
1           1          ELA         1          0             0           1            0               1
2           2         Math         0          1             0           0            1               0
3           3      Science         0          0             1           1            0               1
4           4          ELA         1          0             0           1            0               0
5           5         Math         0          1             0           0            1               0
6           1      Science         0          0             1           1            0               1
7           2         Math         0          1             0           0            1               0
8           3          ELA         1          0             0           1            0               1
9           4          ELA         1          0             0           1            0               0
10          5         Math         0          1             0           0            1               0

Upvotes: 2

Related Questions