Section 5 Functions

So far, we’ve used a few built-in tools like mean() - if you’re coming from a language like SPSS you might think of these tools as “commands”, but in R we call them functions. Functions are basically reusable chunks of code, that take a certain set of inputs (also called arguments) and either produce an output (a return value), or just do a task like showing a plot. When you plug a specific set of inputs into the function and “run” it, we say that you’re calling the function.

The mean() function in R can take a vector of numbers as an input, and return a single number as an output:

## [1] 4.833333

5.1 Arguments

The arguments of a function are the set of inputs it accepts. Some of the inputs will be used to calculate the output, while some might be different options that affect how the calculation happens.

If we look at the arguments for the default mean() function in R, accessed by entering ?mean in the console, we see:

Since the first argument x appears on its own, it’s a mandatory argument. You have to provide a value for x, otherwise you get an error:

Arguments like trim = 0 are optional when you’re calling the function: the value after the = is the default value that will be used if you don’t supply one. The default values tell you what types of input that argument accepts (numeric, logical, character, etc.), but it’s also good to read the information on the function’s help page for more detail.

## [1] 23.5
## [1] 23.5
## [1] 23.3125

5.1.1 Positional or Named Arguments

If you provide arguments for a function without specifying an argument name, they will be used in order, left to right, i.e. by position. R has a plot(x, y) function, and if you call it with plot(1:5, 10:6), 1:5 will be matched up with x and 10:6 will be matched up with y. Most functions take a vector or dataframe as their first argument, so you often end up passing the first argument by position.

If you specify the argument name, you are passing the argument by name. It doesn’t matter which order you pass named arguments in, and you can skip over any arguments you don’t want to provide. Because the order doesn’t matter, both of these are equivalent to plot(1:5, 10:6):

You can provide arguments by both position and name. Usually, you provide one or two positional arguments first, and then as many named arguments as you need4:

If in doubt, just name all the arguments.

5.2 Writing your own functions

You can write your own functions in R. In fact, there’s no real difference between functions you write yourself and those that are built-in or provided by other packages. Generally, using existing functions saves a lot of time (and debugging), but there’s nothing stopping you from creating your own tools if the existing ones don’t quite give you what you need.

The basic template for creating a function is:

We set out what arguments are going to be used within the parentheses () after function. Then, within the braces {} (called the body of the function), we write code to transform the inputs into whatever form we need. return(result) specificies that we want the output of the function to be the value of the result variable5.

A simple example of a useful function you might write yourself is:

## [1] 35

The advantages of writing functions are:

  • If you’re doing the same thing multiple times on different pieces of data, you can create a function and stop writing the same code over and over.
  • A function with a sensible name can make it clearer what is happening in the code: e.g. get_percent(score) vs sum(score) / length(score).
  • You can share functions you’ve written with other people - they’re flexible by nature so they’re not tied to your specific data.

  1. You can mix and match positional and named arguments in different ways, but it gets confusing and is not recommended.

  2. return() is not strictly needed in R: R will return the last value you calculate in the function by default, but return() makes it clearer.