Unique combination of all elements from two (or more) vectors

Question

I am trying to create a unique combination of all elements from two vectors of different size in R.

For example, the first vector is

a <- c("ABC", "DEF", "GHI")

and the second one is dates stored as strings currently

b <- c("2012-05-01", "2012-05-02", "2012-05-03", "2012-05-04", "2012-05-05")

I need to create a data frame with two columns like this

> data
    a          b
1  ABC 2012-05-01
2  ABC 2012-05-02
3  ABC 2012-05-03
4  ABC 2012-05-04
5  ABC 2012-05-05
6  DEF 2012-05-01
7  DEF 2012-05-02
8  DEF 2012-05-03
9  DEF 2012-05-04
10 DEF 2012-05-05
11 GHI 2012-05-01
12 GHI 2012-05-02
13 GHI 2012-05-03
14 GHI 2012-05-04
15 GHI 2012-05-05

So basically, I am looking for a unique combination by considering all the elements of one vector (a) juxtaposed with all the elements of the second vector (b).

An ideal solution would generalize to more input vectors.

---

See also:
How to generate a matrix of combinations

Answer 1

this maybe what you are after

> expand.grid(a,b)
   Var1       Var2
1   ABC 2012-05-01
2   DEF 2012-05-01
3   GHI 2012-05-01
4   ABC 2012-05-02
5   DEF 2012-05-02
6   GHI 2012-05-02
7   ABC 2012-05-03
8   DEF 2012-05-03
9   GHI 2012-05-03
10  ABC 2012-05-04
11  DEF 2012-05-04
12  GHI 2012-05-04
13  ABC 2012-05-05
14  DEF 2012-05-05
15  GHI 2012-05-05

If the resulting order isn't what you want, you can sort afterwards. If you name the arguments to expand.grid, they will become column names:

df = expand.grid(a = a, b = b)
df[order(df$a), ]

And expand.grid generalizes to any number of input columns.

Answer 2

The tidyr package provides the nice alternative crossing, which works better than the classic expand.grid function because (1) strings are not converted into factors and (2) the sorting is more intuitive:

library(tidyr)

a <- c("ABC", "DEF", "GHI")
b <- c("2012-05-01", "2012-05-02", "2012-05-03", "2012-05-04", "2012-05-05")

crossing(a, b)

# A tibble: 15 x 2
       a          b
   <chr>      <chr>
 1   ABC 2012-05-01
 2   ABC 2012-05-02
 3   ABC 2012-05-03
 4   ABC 2012-05-04
 5   ABC 2012-05-05
 6   DEF 2012-05-01
 7   DEF 2012-05-02
 8   DEF 2012-05-03
 9   DEF 2012-05-04
10   DEF 2012-05-05
11   GHI 2012-05-01
12   GHI 2012-05-02
13   GHI 2012-05-03
14   GHI 2012-05-04
15   GHI 2012-05-05

Answer 3

Missing in this r-faq overview is the CJ-function from the data.table-package. Using:

library(data.table)
CJ(a, b, unique = TRUE)

gives:

      a          b
 1: ABC 2012-05-01
 2: ABC 2012-05-02
 3: ABC 2012-05-03
 4: ABC 2012-05-04
 5: ABC 2012-05-05
 6: DEF 2012-05-01
 7: DEF 2012-05-02
 8: DEF 2012-05-03
 9: DEF 2012-05-04
10: DEF 2012-05-05
11: GHI 2012-05-01
12: GHI 2012-05-02
13: GHI 2012-05-03
14: GHI 2012-05-04
15: GHI 2012-05-05

---

_{NOTE: since version 1.12.2 CJ autonames the resulting columns (see also here and here).}

Answer 4

Since version 1.0.0, tidyr offers its own version of expand.grid(). It completes the existing family of expand(), nesting(), and crossing() with a low-level function that works with vectors.

When compared to base::expand.grid():

Varies the first element fastest. Never converts strings to factors. Does not add any additional attributes. Returns a tibble, not a data frame. Can expand any generalised vector, including data frames.

a <- c("ABC", "DEF", "GHI")
b <- c("2012-05-01", "2012-05-02", "2012-05-03", "2012-05-04", "2012-05-05")

tidyr::expand_grid(a, b)

   a     b         
   <chr> <chr>     
 1 ABC   2012-05-01
 2 ABC   2012-05-02
 3 ABC   2012-05-03
 4 ABC   2012-05-04
 5 ABC   2012-05-05
 6 DEF   2012-05-01
 7 DEF   2012-05-02
 8 DEF   2012-05-03
 9 DEF   2012-05-04
10 DEF   2012-05-05
11 GHI   2012-05-01
12 GHI   2012-05-02
13 GHI   2012-05-03
14 GHI   2012-05-04
15 GHI   2012-05-05

Answer 5

you can use order function for sorting any number of columns. for your example

df <- expand.grid(a,b)
> df
   Var1       Var2
1   ABC 2012-05-01
2   DEF 2012-05-01
3   GHI 2012-05-01
4   ABC 2012-05-02
5   DEF 2012-05-02
6   GHI 2012-05-02
7   ABC 2012-05-03
8   DEF 2012-05-03
9   GHI 2012-05-03
10  ABC 2012-05-04
11  DEF 2012-05-04
12  GHI 2012-05-04
13  ABC 2012-05-05
14  DEF 2012-05-05
15  GHI 2012-05-05

> df[order( df[,1], df[,2] ),] 
   Var1       Var2
1   ABC 2012-05-01
4   ABC 2012-05-02
7   ABC 2012-05-03
10  ABC 2012-05-04
13  ABC 2012-05-05
2   DEF 2012-05-01
5   DEF 2012-05-02
8   DEF 2012-05-03
11  DEF 2012-05-04
14  DEF 2012-05-05
3   GHI 2012-05-01
6   GHI 2012-05-02
9   GHI 2012-05-03
12  GHI 2012-05-04
15  GHI 2012-05-05`

Answer 6

In base R you could try merge(), cbind() and expand.grid().

a <- seq(1E4)
b <- c("2012-05-01", "2012-05-02", "2012-05-03", "2012-05-04", "2012-05-05")

 microbenchmark(
  "merge (1)" = mmm <- as.matrix(merge(a, b)),
  "diy (2)"   = {ccc <- cbind( rep(a, length(b)),
                               b[rep(seq_along(b), each = length(a))]
                        )
                },
 "diy R (3)"  = {ccc <- cbind( a,
                               b[rep(seq_along(b), each = length(a))]
                        )
                },
  "grid (4)"  = ggg <- expand.grid(a, b),
  times       = 2
)

Output.

Unit: milliseconds
      expr      min       lq     mean   median       uq      max neval
 merge (1) 863.3100 863.3100 888.6573 888.6573 914.0046 914.0046     2
   diy (2) 117.1912 117.1912 142.1394 142.1394 167.0875 167.0875     2
 diy R (3)  34.9320  34.9320  49.4119  49.4119  63.8918  63.8918     2
  grid (4)  45.1876  45.1876  46.1592  46.1592  47.1308  47.1308     2

Answer 7

You could use rep and the fact that base R data frames recycle:

data.frame(
  a = rep(a, each = length(b)),
  b = b
)