Unstacking a dataframe in R that only has one row for each category

Question

If I have the following data frame called pets in R:

> pets
     name animal
1     Amy    dog
2     Bob    cat
3   Carol    dog
4   Danny    cat
5 Eustace  horse
6 Frances  horse

I can unstack the data frame to set the animal categories as columns headers like so:

> unstack(pets, pets$name ~ pets$animal)
    cat   dog   horse
1   Bob   Amy Eustace
2 Danny Carol Frances

If, however, the pets data frame only has one instance of each animal:

> pets
     name animal
1     Amy    dog
2     Bob    cat

then running the same code produces this result:

> unstack(pets, pets$name ~ pets$animal)
    res
cat Bob
dog Amy

I need something that puts those animal categories as the column headers, regardless of how many rows fall into each category. Does anyone have any ideas, please?

Answer 1

data.frame and split will do it:

pets <- data.frame(
 name=c("Amy", "Bob", "Carol", "Danny", "Eustace", "Frances"),
 animal=c("dog", "cat", "dog", "cat", "horse", "horse")
)

data.frame(split(pets$name,pets$animal,drop=TRUE))

#    cat   dog   horse
#1   Bob   Amy Eustace
#2 Danny Carol Frances

pets2 <- pets[1:2,]

data.frame(split(pets2$name,pets2$animal,drop=TRUE))

#  cat dog
#1 Bob Amy

Answer 2

You can simply transpose the result:

> t(unstack(pets))
    cat   dog  
res "Bob" "Amy"

If you want a solution that will work for both:

unstack <- function(..., drop=FALSE) {
  u <- utils::unstack(...)
  if (!drop && ncol(u) == 1) {
    u <- t(u)
    rownames(u) <- 1
    return(u)
  } else {
    return(u)
  }
}

This will overwrite the unstack function. drop=TRUE preserves the default behaviour, while drop=FALSE gives you your desired output.

Answer 3

Another alternative:

> p2 <- pets[!duplicated(pets$animal), ]
> data.frame(matrix(p2$name, nrow = 1, ncol = nrow(p2), 
                    dimnames = list(NULL, p2$animal)))
#   dog cat   horse
# 1 Amy Bob Eustace