mkearney / dapr

☝🏼👉🏼👇🏼👈🏼 Dependency-free purrr-like apply/map/iterate functions

dapr

Dependency-free purrr-like apply/map/iterate functions

Installation

Install the development version from Github with:

## install remotes pkg if not already
if (!requireNamespace("remotes", quietly = TRUE)) {
  install.packages("remotes")
}

## install from github
remotes::install_github("mkearney/dapr")

{dapr} vs. {base} & {purrr}?

{dapr} provides the ease and consistency of {purrr}, (see also: simple benchmark results plot below) including use of ~ and .x, without all the dependencies. In other words, use {dapr} when you want a purrr-like experience but you need a lightweight solution.

Use

Function names use the convention *ap() where * is the first letter of output data type.

• vap for vectors
• lap for lists
• dap for data frames

Common inputs:

• .data Input object–numeric, character, list, data frame, etc.–over which elements will be iterated. If matrix or data frame, each column will be treated as the elements which are to be iterated over.
• .f Function to apply to each element of input object. This can be written as a single function name e.g., mean, a formula-like function call where .x is assumed to be the iterated over element of input data e.g., ~ mean(.x), or an in-line function definition e.g., function(x) mean(x).

Vectors

Functions that apply expressions to input data objects and return atomic vectors e.g., numeric (double), character, logical.

• vap_dbl() Iterate and return numeric vector.
• vap_int() Iterate and return integer vector.
• vap_lgl() Iterate and return logical vector.
• vap_chr() Iterate and return character vector.

## create data
set.seed(2018)
d <- replicate(5, rnorm(10), simplify = FALSE)
e <- replicate(5, sample(letters, 10), simplify = FALSE)

## numeric
vap_dbl(d, ~ mean(.x))
#> [1]  0.26934527 -0.55232322  0.05559290 -0.06253258 -0.11183760

## integer
vap_int(d, length)
#> [1] 10 10 10 10 10

## logical
vap_lgl(d, ~ max(.x) > 3)
#> [1] FALSE FALSE FALSE FALSE FALSE

## character
vap_chr(e, paste, collapse = "")
#> [1] "hizjpgcexk" "rbeovimtxh" "ujrimwgvzs" "euwrlytgbj" "qkrhylgmnx"

Lists

Function(s) that apply expressions to input data objects and return lists.

• lap() Iterate and return a list vector.

## list of strings
lap(e[1:2], ~ paste0(.x, "."))
#> [[1]]
#>  [1] "h." "i." "z." "j." "p." "g." "c." "e." "x." "k."
#> 
#> [[2]]
#>  [1] "r." "b." "e." "o." "v." "i." "m." "t." "x." "h."

• ilap() Iterate over sequence length .i (instead of .x) and return a list vector.

## list of strings
ilap(1:4, ~ paste0(letters[.i], rev(LETTERS)[.i]))
#> [[1]]
#> [1] "aZ"
#> 
#> [[2]]
#> [1] "bY"
#> 
#> [[3]]
#> [1] "cX"
#> 
#> [[4]]
#> [1] "dW"

Data frames

Functions that apply expressions to input data objects and return data frames.

• dap*() Iterate and return a data frame
  • dapc() Iterate over columns
  • dapr() Iterate over rows
• dap*_if() Conditionally iterate
  • dapc_if() Conditionally iterate over columns
  • dapr_if() Conditionally iterate over rows

## some data
d <- data.frame(
  a = letters[1:3],
  b = rnorm(3),
  c = rnorm(3),
  stringsAsFactors = FALSE
)

## column explicit (same as dap)
dapc(d[-1], ~ round(.x, 2))
#>       b     c
#> 1 -0.50 -0.09
#> 2 -1.87  1.08
#> 3  0.74 -1.36

## rows
dapr(d[-1], round, 3)
#>        b      c
#> 1 -0.499 -0.089
#> 2 -1.869  1.081
#> 3  0.743 -1.365

## conditional COLUMNS
dapc_if(d, is.numeric, ~ round(.x, 4))
#>   a       b       c
#> 1 a -0.4994 -0.0892
#> 2 b -1.8686  1.0812
#> 3 c  0.7434 -1.3646

## conditional ROWS
dapr_if(d[-1], ~ sum(.x) >= -.7, ~ round(.x, 0))
#>           b         c
#> 1  0.000000  0.000000
#> 2 -1.868615  1.081164
#> 3  1.000000 -1.000000

dapr's Issues

taking number, or "name" as input, like the purrr map

Nice package. In the purrr, map function could also takes number, or "name". Would it be possible to add these features to the package as well?