3

I'm trying to translate the following syntax found in tidyverse into base R as a function, though I'm having difficulties following the same output.

Here's the syntax:

x <- function(x) {x %>% 
    select(where(negate(is.numeric))) %>% 
    map_dfc(~ model.matrix(~ .x -1) %>% 
              as_tibble) %>% 
    rename_all(~ str_remove(., "\\.x")) 
}

I understand that select can be represented as indexing within a dataframe such as x[,]. As for the pipe function %>%, I can just index a function within a variable i.e. x <- ...

I can manage to transfer select(where(negate(is.numeric)))

into:

x <- function(x){
  x[, !sapply(x, is.numeric)]
  
}

Though, this makes it difficult, as I'm thinking it can be replaced with a conditional argument:

 map_dfc(~ model.matrix(~ .x -1)

Here's the expected output with some example data:

# A tibble: 12 x 5
   black brown white female  male
   <dbl> <dbl> <dbl>  <dbl> <dbl>
 1     1     0     0      1     0
 2     1     0     0      1     0
 3     1     0     0      1     0
 4     1     0     0      1     0
 5     0     0     1      1     0
 6     0     0     1      1     0
 7     0     0     1      0     1
 8     0     0     1      0     1
 9     0     1     0      0     1
10     0     1     0      0     1
11     0     1     0      0     1
12     0     1     0      0     1

reproducible code:

structure(list(wgt = c(64L, 71L, 53L, 67L, 55L, 58L, 77L, 57L, 
56L, 51L, 76L, 68L), hgt = c(57L, 59L, 49L, 62L, 51L, 50L, 55L, 
48L, 42L, 42L, 61L, 57L), age = c(8L, 10L, 6L, 11L, 8L, 7L, 10L, 
9L, 10L, 6L, 12L, 9L), id = structure(c(1L, 1L, 1L, 1L, 3L, 3L, 
3L, 3L, 2L, 2L, 2L, 2L), .Label = c("black", "brown", "white"
), class = "factor"), sex = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 
2L, 2L, 2L, 2L, 2L, 2L), .Label = c("female", "male"), class = "factor")), class = "data.frame", row.names = c(NA, 
-12L))
Lime
  • 738
  • 5
  • 17

2 Answers2

4

1) If input is the input data frame, define a model matrix function mm and lapply it to the non-numeric columns and put them together into a single data frame. Finally fix up the names.

mm <- function(x) model.matrix(~ x - 1)
result <- do.call("data.frame", lapply(Filter(Negate(is.numeric), input), mm))
names(result) <- sub(".*\\.x", "", names(result))
result

giving:

   black brown white female male
1      1     0     0      1    0
2      1     0     0      1    0
3      1     0     0      1    0
4      1     0     0      1    0
5      0     0     1      1    0
6      0     0     1      1    0
7      0     0     1      0    1
8      0     0     1      0    1
9      0     1     0      0    1
10     0     1     0      0    1
11     0     1     0      0    1
12     0     1     0      0    1

2) To make it similar to the tidyverse version we can use the Bizarro pipe which does not require any packages.

input ->.;
  Filter(Negate(is.numeric), .) ->.;
  lapply(., function(x) model.matrix(~ x - 1)) ->.;
  do.call("data.frame", .) ->.;
  setNames(., sub(".*\\.x", "", names(.))) -> result
result
G. Grothendieck
  • 254,981
  • 17
  • 203
  • 341
3

Calling your input data xx,

onehot = function(data) {
  x = Filter(Negate(is.numeric), data)
  x = as.data.frame(Reduce(cbind, lapply(x, function(col) model.matrix(~ . - 1, data = data.frame(col)))))
  setNames(x, sub(pattern = "^col", replacement = "", names(x)))
}

onehot(xx)
#    black brown white female male
# 1      1     0     0      1    0
# 2      1     0     0      1    0
# 3      1     0     0      1    0
# 4      1     0     0      1    0
# 5      0     0     1      1    0
# 6      0     0     1      1    0
# 7      0     0     1      0    1
# 8      0     0     1      0    1
# 9      0     1     0      0    1
# 10     0     1     0      0    1
# 11     0     1     0      0    1
# 12     0     1     0      0    1

There are other packages that do one-hot encoding like this, see here for some examples, but the above is all base.

Gregor Thomas
  • 136,190
  • 20
  • 167
  • 294