Skip to contents

A ‘dibble’ (derived from ‘dimensional tibble’) is a data frame consisting of arrays with dimension names, known as data cubes. The columns of the dibbles are classified into dimensions or measures, and the operations on the measures are broadcasted by dimension names.

Installation

# the released version from CRAN:
install.packages("dibble")

# the development version from GitHub:
# install.packages("devtools")
devtools::install_github("UchidaMizuki/dibble")

Examples

Broadcasting

arr1 <- array(1:6, c(2, 3),
              list(axis1 = letters[1:2],
                   axis2 = letters[1:3]))
arr2 <- array(1:2, 2,
              list(axis2 = letters[1:2]))

try(arr1 * arr2)
#> Error in arr1 * arr2 : non-conformable arrays

ddf1 <- as_dibble(arr1)
ddf2 <- as_dibble(arr2)

ddf1 * ddf2
#> Warning: Broadcasting,
#> New axes, dim_names = c("axis1", "axis2")
#> New coordinates, 
#>  $ axis2: chr "c"
#> # A dibble:   6
#> # Dimensions: axis1 [2], axis2 [3]
#>   axis1 axis2     .
#>   <chr> <chr> <int>
#> 1 a     a         1
#> 2 a     b         6
#> 3 a     c        NA
#> 4 b     a         2
#> 5 b     b         8
#> 6 b     c        NA
# You can use broadcast() to suppress the warnings.
broadcast(ddf1 * ddf2,
          dim_names = c("axis1", "axis2"))
#> # A dibble:   6
#> # Dimensions: axis1 [2], axis2 [3]
#>   axis1 axis2     .
#>   <chr> <chr> <int>
#> 1 a     a         1
#> 2 a     b         6
#> 3 a     c        NA
#> 4 b     a         2
#> 5 b     b         8
#> 6 b     c        NA

dplyr methods

dibble provides some dplyr methods as follows,

How to build a dibble

From a data.frame

df <- expand_grid(axis1 = letters[1:2],
                  axis2 = letters[1:2]) |> 
  mutate(value1 = row_number(),
         value2 = value1 * 2)

ddf <- df |> 
  dibble_by(axis1, axis2)
ddf
#> # A dibble:   4 x 2
#> # Dimensions: axis1 [2], axis2 [2]
#> # Measures:   value1, value2
#>   axis1 axis2 value1 value2
#>   <chr> <chr>  <int>  <dbl>
#> 1 a     a          1      2
#> 2 a     b          2      4
#> 3 b     a          3      6
#> 4 b     b          4      8
# You can access the measures from the dibble with `$`.
ddf$value1
#> # A dibble:   4
#> # Dimensions: axis1 [2], axis2 [2]
#>   axis1 axis2     .
#>   <chr> <chr> <int>
#> 1 a     a         1
#> 2 a     b         2
#> 3 b     a         3
#> 4 b     b         4

df <- expand_grid(tibble(axis1_key = letters[1:2],
                         axis1_value = 1:2),
                  tibble(axis2_key = letters[1:2],
                         axis2_value = 1:2)) |>
  mutate(value1 = row_number(),
         value2 = value1 * 2)

# You can `pack` several columns into one dimension (See `tidyr::pack()`).
df |> 
  dibble_by(axis1 = c(axis1_key, axis1_value),
            axis2 = c(axis2_key, axis2_value),
            .names_sep = "_")
#> # A dibble:   4 x 2
#> # Dimensions: axis1 [2], axis2 [2]
#> # Measures:   value1, value2
#>   axis1$key $value axis2$key $value value1 value2
#>   <chr>      <int> <chr>      <int>  <int>  <dbl>
#> 1 a              1 a              1      1      2
#> 2 a              1 b              2      2      4
#> 3 b              2 a              1      3      6
#> 4 b              2 b              2      4      8

From an array with dimension names or a vector

dibble provides some dplyr methods as follows,

# from an array with dimension names
arr <- array(1:4, c(2, 2),
             list(axis1 = letters[1:2],
                  axis2 = letters[1:2]))

ddf1 <- as_dibble(arr)

# from a vector
ddf2 <- broadcast(1:4,
                  list(axis1 = letters[1:2],
                       axis2 = letters[1:2]))

arr
#>      axis2
#> axis1 a b
#>     a 1 3
#>     b 2 4
ddf1
#> # A dibble:   4
#> # Dimensions: axis1 [2], axis2 [2]
#>   axis1 axis2     .
#>   <chr> <chr> <int>
#> 1 a     a         1
#> 2 a     b         3
#> 3 b     a         2
#> 4 b     b         4
ddf2
#> # A dibble:   4
#> # Dimensions: axis1 [2], axis2 [2]
#>   axis1 axis2     .
#>   <chr> <chr> <int>
#> 1 a     a         1
#> 2 a     b         3
#> 3 b     a         2
#> 4 b     b         4