Functions

Jeff Stevens

2023-03-27

Introduction

Set-up

library(tidyverse)
library(palmerpenguins)

Why write your own functions?

Creating functions

functionname <- function(argument1, argument2) {
  # Function contents
}

Creating functions

mymean <- function(x) {
  sum(x) / length(x)
}

mymean(mtcars$mpg)

[1] 20.09062

mean(mtcars$mpg)

[1] 20.09062

Multi-line functions

mymean2 <- function(x) {
  mysum <- sum(x)
  mysum_divided <- mysum / length(x)
}

mymean2(mtcars$mpg)

Why no output?

(mymean_obj <- mymean2(mtcars$mpg))

[1] 20.09062

Returning output

Don’t assign last step to object

mymean3 <- function(x) {
  mysum <- sum(x)
  mysum / length(x)
}

mymean3(mtcars$mpg)

[1] 20.09062

Returning output

Or use return()

mymean4 <- function(x) {
  mysum <- sum(x)
  mysum_divided <- mysum / length(x)
  return(mysum_divided)
}

mymean4(mtcars$mpg)

[1] 20.09062

Return multiple output values

list()

mymean5 <- function(x) {
  mysum <- sum(x)
  mysum_divided <- mysum / length(x)
  list(sum = mysum, mean = mysum_divided)
}

mymean5(mtcars$mpg)

$sum
[1] 642.9

$mean
[1] 20.09062

Save intermediate objects

mymean6 <- function(x) {
  mysum <<- sum(x)
  mysum / length(x)
}

mymean6(mtcars$mpg)

[1] 20.09062

Note

This is fine when testing out a function but probably not great practice ‘in production’. Why?

Print messages to console

mymean7 <- function(x) {
  mysum <- sum(x)
  print(mysum)
  mysum / length(x)
}

mymean7(mtcars$mpg)

[1] 642.9

[1] 20.09062

Print messages to console

mymean8 <- function(x) {
  mysum <- sum(x)
  print(paste0("The sum is: ", mysum))
  mysum / length(x)
}

mymean8(mtcars$mpg)

[1] "The sum is: 642.9"

[1] 20.09062

Print messages to console

mymean9 <- function(x) {
  mysum <- sum(x)
  message(paste0("The sum is: ", mysum))
  mysum / length(x)
}

mymean9(mtcars$mpg)

The sum is: 642.9

[1] 20.09062

Note

Check out {cli} package for powerful messaging.

Arguments

multiplier <- function(x, constant) {
  x * constant
}

multiplier(x = 7, constant = 3)

[1] 21

multiplier(x = 1:10, constant = 3)

 [1]  3  6  9 12 15 18 21 24 27 30

multiplier(x = 1:10)

Error in multiplier(x = 1:10) : 
argument "constant" is missing, with no default

Argument default values

multiplier2 <- function(x, constant = 3) {
  x * constant
}

multiplier2(x = 1:10)

 [1]  3  6  9 12 15 18 21 24 27 30

multiplier2(x = 1:10, constant = 5)

 [1]  5 10 15 20 25 30 35 40 45 50

Let’s write a function!

Here’s the formula that reverse codes scale values

Value_max+ Value_min-Score

Write a function that calculates the reversed code score. What arguments do you need?

Conditional execution

head(penguins[, 1:5])

# A tibble: 6 × 5
  species island    bill_length_mm bill_depth_mm flipper_length_mm
  <fct>   <fct>              <dbl>         <dbl>             <int>
1 Adelie  Torgersen           39.1          18.7               181
2 Adelie  Torgersen           39.5          17.4               186
3 Adelie  Torgersen           40.3          18                 195
4 Adelie  Torgersen           NA            NA                  NA
5 Adelie  Torgersen           36.7          19.3               193
6 Adelie  Torgersen           39.3          20.6               190

mymean6(penguins$bill_length_mm)

[1] NA

Conditional execution

head(penguins[, 1:5])

# A tibble: 6 × 5
  species island    bill_length_mm bill_depth_mm flipper_length_mm
  <fct>   <fct>              <dbl>         <dbl>             <int>
1 Adelie  Torgersen           39.1          18.7               181
2 Adelie  Torgersen           39.5          17.4               186
3 Adelie  Torgersen           40.3          18                 195
4 Adelie  Torgersen           NA            NA                  NA
5 Adelie  Torgersen           36.7          19.3               193
6 Adelie  Torgersen           39.3          20.6               190

mymean10 <- function(x) {
  sum(x, na.rm = TRUE) / sum(!is.na(x))
}

mymean10(penguins$bill_length_mm)

[1] 43.92193

Conditional execution

But if you want the user to control whether NA is ignored

mymean11 <- function(x, ignore_na = TRUE) {
  if (ignore_na) {
    sum(x, na.rm = TRUE) / sum(!is.na(x))
  } else {
    sum(x) / length(x)
  }
}

mymean11(penguins$bill_length_mm)

[1] 43.92193

mymean11(penguins$bill_length_mm, ignore_na = FALSE)

[1] NA

Multiple conditions

Use else if

age_cutoffs <- function(x) {
  if(x <= 1.5) {
    "puppy"
  } else if (x <= 3) {
    "adolescent"
  } else if (x <= 10) {
    "adult"
  } else {
    "senior"
  }
}

age_cutoffs(1)

[1] "puppy"

age_cutoffs(2)

[1] "adolescent"

age_cutoffs(5)

[1] "adult"

Stopping based on conditionals

age_cutoffs2 <- function(x) {
  if(x <= 1.5) {
    "puppy"
  } else if (x <= 3) {
    "adolescent"
  } else if (x <= 10) {
    "adult"
  } else if (x <= 20) {
    "senior"
  } else {
    stop("Age exceeded 20.")
  }
}

age_cutoffs2(15)

[1] "senior"

age_cutoffs2(22)

Error in age_cutoffs2(22) : Age exceeded 20.

Multiple conditions

Use switch()

central_tend <- function(x, type) {
  switch(type,
         mean = mean(x),
         median = median(x),
         trimmed = mean(x, trim = .1))
}

vector <- rcauchy(100)
central_tend(x = vector, type = "mean")

[1] 3.548399

central_tend(x = vector, type = "median")

[1] 0.04851638

central_tend(x = vector, type = "trimmed")

[1] 0.3435384

Data frame functions

Creating functions in tidyverse

mean_species_bill <- function(df) {
  df |> 
    group_by(species) |> 
    summarise(mean(bill_length_mm, na.rm = TRUE))
}

mean_species_bill(penguins)

# A tibble: 3 × 2
  species   `mean(bill_length_mm, na.rm = TRUE)`
  <fct>                                    <dbl>
1 Adelie                                    38.8
2 Chinstrap                                 48.8
3 Gentoo                                    47.5

Creating functions in tidyverse

What if we want user to input grouping and response variable?

grouped_mean <- function(df, group_var, mean_var) {
  df |> 
    group_by(group_var) |> 
    summarise(mean(mean_var, na.rm = TRUE))
}

penguins |> 
  grouped_mean(group_var = species, mean_var = bill_length_mm)

Error in `group_by()`:
! Must group by variables found in `.data`.
✖ Column `group_var` is not found.
Run `rlang::last_trace()` to see where the error occurred.

Embracing

Embrace variables in {{ }}

grouped_mean2 <- function(df, group_var, mean_var) {
  df |> 
    group_by({{ group_var }}) |> 
    summarize(mean({{ mean_var }}, na.rm = TRUE))
}

penguins |> 
  grouped_mean2(group_var = species, mean_var = bill_length_mm)

# A tibble: 3 × 2
  species   `mean(bill_length_mm, na.rm = TRUE)`
  <fct>                                    <dbl>
1 Adelie                                    38.8
2 Chinstrap                                 48.8
3 Gentoo                                    47.5

Let’s code!

Functions [Rmd]