Skip to content

Latest commit

 

History

History
269 lines (192 loc) · 5.17 KB

purrr_pres.md

File metadata and controls

269 lines (192 loc) · 5.17 KB

purrr::A_Primer

author: Karl Hailperin date: November 30, 2017 autosize: true

https://github.com/khailper/purrr_pres

What is purrr?

  • Package created by Lionel Henry and co-authored by Hadley Wickham, both of RStudio
  • Part of the broader tidyverse
  • Contains a variety of functions to make programing easier, mostly by reducing for loops

So What?

incremental:true

Which would you rather write?

This:

x <- 1:5
sum <- 0
for (i in seq(length(x))){
        sum <- sum + x[i]
}
sum/length(x)
[1] 3

Or this:

x <- 1:5
mean(x)
[1] 3

The map() Family, Part 1: The basics

  • Basic form: map(.x, .f, ...)
  • Creates results like this:
results = vector("list", length(.x))
for (i in seq(length(.x))){
        results[i] <- .f(.x[i], ...)
}
results

library(purrr)
numbers <- 1:3
map(numbers, sqrt)
[[1]]
[1] 1

[[2]]
[1] 1.414214

[[3]]
[1] 1.732051

The map() Family, Part 2: map_*

  • Returns a list by default, but the map_* family returns a vector
  • For example, map_chr returns a character vector
  • Other suffixes include lgl, int, dbl
map_dbl(numbers, sqrt)
[1] 1.000000 1.414214 1.732051

The map() Family, Part 3: Other ways to use .f

  • If .x is a nested list, using a number for .f returns the nth element of each element of .x
  • If the nested lists are named, you can string inputs to get that input
people <- list(c(first_name = "Kristen", last_name = "Bell"), 
               c(first_name = "Ted", last_name = "Danson"), 
               c(first_name = "William", middle_name = "Jackson", last_name = "Harper"))

map_chr(people, 2)
[1] "Bell"    "Danson"  "Jackson"

map_chr(people, "last_name")
[1] "Bell"   "Danson" "Harper"

# map_chr(people, 3) returns error
map_chr(people, 3, .default = NA)
[1] NA       NA       "Harper"

The map() Family, Part 4: anonymous functions

  • You can add more complex functions in .f with ~ notation
plots <- list(Plot_1 = plot_1, Plot_2 = plot_2, Plot_3 = plot_3)
plot_paths <- map_chr(names(plots), ~ paste0("Plots/", ., ".png"))

The map() Family, Part 5: map2()

  • map2 is like map, but takes two lists, and has the form map2(.x, .y, .f, ...)
  • has map2_* functions
vector_1 <- c("Spite", "Research", "Boom", "Far")
vector_2 <- c("Malice", "Development", "Bust", "Away")
map2_chr(vector_1, vector_2, paste, sep = " and ")
[1] "Spite and Malice"         "Research and Development"
[3] "Boom and Bust"            "Far and Away"

The map() Family, Part 6: pmap()

  • Instead of creating abitrarily many mapnumber functions, purrr has pwalk(.l, .f, ...)
  • .l is a list of lists
  • There are pmap_[type] functions
observations <- 1:3
sizes <- 6 # will be recycled
probabilities <- 0.1 * (6:8)
pmap(list(observations,sizes, probabilities), rbinom)
[[1]]
[1] 3

[[2]]
[1] 4 5

[[3]]
[1] 3 5 6

The map() Family, Part 7: invoke_map()

  • invoke_map(.f, .x, ...) is map in reverse, constant input, list of functions
  • There are invoke_map_[type] functions
sample_size <- 5
function_list <- list(rnorm, runif)
invoke_map(function_list, sample_size)
[[1]]
[1] -0.7606657 -0.5510715 -0.7123391  0.0662748 -0.2277455

[[2]]
[1] 0.42801122 0.17679530 0.52070898 0.03660668 0.68454829

The map() Family, Part 8: walk()

  • walk() is map, but for functions that like write_csv() where we only care about the side effects (the file creation, in this case)
walk2(plot_paths, plots, ggsave)

The map() Family, Part 9: modify()

  • modify(.x , .f, ...) is like map, but it always returns an object of the same type as .x
example_df <- data.frame(x = 1:3, y = 4:6, z = 7:9)
map(example_df, log)
$x
[1] 0.0000000 0.6931472 1.0986123

$y
[1] 1.386294 1.609438 1.791759

$z
[1] 1.945910 2.079442 2.197225

modify(example_df, log)
          x        y        z
1 0.0000000 1.386294 1.945910
2 0.6931472 1.609438 2.079442
3 1.0986123 1.791759 2.197225

The map() Family, Part 10: modify_if/at()

  • modify_if(.x, .p, .f, ...) applies .f where .p is TRUE, and leavesit alone otherwise
  • modify_at(.x, .at, .f, ...) is like modify_if, but .at is character or numeric vector indicating which positions to evaluate .f at
Warning message:
package 'knitr' was built under R version 4.1.1 


processing file: purrr_pres.Rpres
Quitting from lines 151-153 (purrr_pres.Rpres) 
Error: Can't coerce element 1 from a double to a integer
Execution halted