6 Writing functions

(AST230) R for Data Science

Md Rasel Biswas

Writing own R functions

  • Every R function has three parts:
    • name
    • a body of code
    • a set of arguments
  • The basic format of an R function
my_fun <- function(arguments) {
  
  "body of code"
  
}

  • Functions are just another R object

  • Comas separate more than one argument

Example

  • Write a function that calculates the area of a right triangle
calculate_right_triangle_area <- function(base, height) {
  area <- 0.5 * base * height
  return(area)
}
  • It takes two arguments: base and height.
calculate_right_triangle_area(2, 3)
[1] 3

Simulating rolling dice

The following codes simulate an experiment “rolling a dice”

sample(x = 1:6, size = 1)
[1] 3
sample(x = 1:6, size = 1)
[1] 4

Write a function roll() for the simulation

roll <- function() {
   dice <- 1:6 
   out <- sample(x = dice, size = 1)
   return(out)
}
  • roll() has no arguments, and it will return a number between 1 to 6
  • To use the function roll(), first run the written R function codes in the R console. Then run roll()
# Start rolling dice
# roll 1
roll()
[1] 1
# roll 2
roll()
[1] 2
# See the function codes
roll
function() {
   dice <- 1:6 
   out <- sample(x = dice, size = 1)
   return(out)
}
<bytecode: 0x7fb187b33b60>

Functions with arguments

The number of heads in 10 fair coin toss

# Toss a fair coin 10 times
s10 <- sample(x = c(0,1), size = 10, replace = TRUE)
sum(s10)
[1] 5

The number of heads in 20 fair coin toss

# Toss a fair coin 20 times
s20 <- sample(x = c(0,1), size = 20, replace = TRUE)
sum(s20)
[1] 12

The number of heads in 100 fair coin toss

# Toss a fair coin 100 times
s100 <- sample(x = c(0,1), size = 100, replace = TRUE)
sum(s100)
[1] 67

  • The size argument of sample() is changing for different number of coin toss

  • Can you write a function in R that takes how many times to toss a fair coin and prints us how many times it lands on heads?

# Creating the function
toss_fair_coins <- function(times) {
  out <- sample(x = 0:1, size = times, replace = T)
  sum(out)
}

Tossing coins 200/500 times

toss_fair_coins(times = 200)
[1] 107
toss_fair_coins(times = 500)
[1] 241

What will happen if you run the function without specifying the argument

toss_fair_coins()
# Creating R function with default argument values
toss_fair_coins2 <- function(times = 100) {
  out <- sample(x = 0:1, size = times, replace = T)
  sum(out)
}

  • Default value of an argument can be specified

  • E.g., times = 100 will be considered if times is not supplied!

# No argument supplied
toss_fair_coins2()
[1] 45
# If argument supplied
toss_fair_coins2(times=10000)
[1] 5003

Exercise 6.1

  • Write an R function roll_dice() that rolls a dice n times (where n is 1 or more) and returns the sum of the dice rolls.

  • Write an R function toss_bias_coins() that tosses a biased coin n times (where n is 1 or more times) and returns the number of heads. The probability of getting heads is 0.70.

  • Write an R function that can simulate flipping a fair or biased coin. The probability of getting heads should not always be 0.70, but any value between 0 and 1!

Help for R functions

help("mean") or “?mean” to check the detail of mean()

help("mean") # equivalently ?mean
help.search("mean") #equivalently ??mean

  • The help.search() function searches through the help documentation, code demonstrations and package vignettes and displays the results as clickable links for further exploration

  • Another useful function is apropos()

apropos("mean")
 [1] ".colMeans"     ".rowMeans"     "colMeans"      "kmeans"       
 [5] "mean"          "mean.Date"     "mean.default"  "mean.difftime"
 [9] "mean.POSIXct"  "mean.POSIXlt"  "rowMeans"      "weighted.mean"

Pro Tip

RStudio snippet can save your time. Type fun and wait some milliseconds to see suggestions (press Tab if they don’t appear). Select (or press Tab) on the snippet option to insert it. You can see and customize your snippets from Tools>Global Options>Code.