Generating Cartesian product of strings in R

Question

Sometimes I need to create character vectors like this one:

datanames <- c(
  "europe_co2_min",
  "europe_co2_max",
  "europe_temperature_min",
  "europe_temperature_max",
  "asia_co2_min",
  "asia_co2_max",
  "asia_temperature_min",
  "asia_temperature_max"
)

Notice that this is actually the Cartesian product of the character vectors joined together with a _:

c('europe', 'asia')
c('co2', 'temperature')
c('min', 'max')

I came up with this helper function:

combine <- function(..., prefix='', sep='_') {
  combine.inner <- function(lx, ...) {
    if (length(c(...)) > 0) {
      sapply(sapply(lx, function(x) paste(x, combine.inner(...), sep=sep)), c)
    } else {
      lx
    }
  }
  paste(prefix, combine.inner(...), sep='')
}

It handles arbitrary number of character vector parameters of arbitrary lengths. The prefix parameter is for convenience. It's similar to having a single element vector as the first parameter, except the common separator sep will not be applied after it.

1. Is there an easier way that exists in R and I missed?
2. Is there anything smelly about this code?

Answer 1

Your approach is an example of good R code. However, there is a base function that allows creating a cartesian product of strings, interaction. This function creates a factor, and the levels are equivalent to the cartesian product.

Furthermore, instead of paste(..., sep = "") you can use paste0(...).

If you use these functions, your code will be much shorter:

combine <- function(..., prefix = "", sep = "_") {
  paste0(prefix, levels(interaction(..., sep = sep)))
}

Examples:

s1 <- c('europe', 'asia')
s2 <- c('co2', 'temperature')
s3 <- c('min', 'max')

combine(s1)
# [1] "europe" "asia"  
combine(s1, s2)
# [1] "europe_co2"         "europe_temperature" "asia_co2"           "asia_temperature"  
combine(s1, s2, s3)
# [1] "europe_co2_min"         "europe_co2_max"         "europe_temperature_min"
# [4] "europe_temperature_max" "asia_co2_min"           "asia_co2_max"          
# [7] "asia_temperature_min"   "asia_temperature_max"  

Answer 2

You can use expand.grid:

a <- c('europe', 'asia')
b <- c('co2', 'temperature')
c <- c('min', 'max')
expand.grid(a, b, c)

Output:

    Var1        Var2 Var3
1 europe         co2  min
2   asia         co2  min
3 europe temperature  min
4   asia temperature  min
5 europe         co2  max
6   asia         co2  max
7 europe temperature  max
8   asia temperature  max

Answer 3

Well, it can be still useful now-a-days.

a=c('europe', 'asia')
b=c('co2', 'temperature')
c=c('min', 'max')

apply(expand.grid(a,b,c), 1, function(x) paste0(x, collapse="_"))

Output:

[1] "europe_co2_min"         "asia_co2_min"           "europe_temperature_min"
[4] "asia_temperature_min"   "europe_co2_max"         "asia_co2_max"          
[7] "europe_temperature_max" "asia_temperature_max"

Answer 4

Maybe someone will still be served. I modified one of the codes they provided here, to create a tablet, although I'm not sure it's the best way to do it. Perhaps someone with more experience can tell us.

a <- c("a","b","o")
comb <- apply(expand.grid(a,a), 1, function(x) paste0(x, collapse=","))
tb <- matrix(comb,ncol = length(a),nrow=length(a))
colnames(tb) <- a
row.names(tb) <- a

Returns:

      a    b    o
a    a,a  a,b  a,c 
b    b,a  b,b  b,o 
o    o,a  o,b  o,o