Exploring data with R

In this tutorial we will look deeper at the R programming language and how it can be used for exploratory data analysis (EDA).

You should run all the R code examples given yourself and answer the exercises at the end of each section. Remember to also make notes on key concepts as you encounter them.

A note on typography

R code, including literal values and the names of objects and functions, is denoted by code-style text. The names of R packages are linked the first time they are mentioned, but otherwise written with normal text. Specialist statistical and programming concepts and other ‘terms of art’ are highlighted, especially when they are first used.

Objectives

By the end of this tutorial, you should:

Recognise the building blocks of the R environment: objects, values, and functions.
Use the R console to load a dataset, calculate a statistic and produce a plot
Begin your Zettelkästen with notes on the key pieces of R syntax you have learned

Prerequisites

Andy Matuschak, ‘Evergreen notes’
Install R on your computer

Getting started with R

R is a ‘statistical computing environment’. It combines a programming language (R) with an interactive console for exploring and visualising data, as well as thousands of extensions (packages) that can be installed to perform specialised tasks.

On Windows and macOS, the usual way to install R is to download a binary installer from CRAN. For Linux, it can be found in the official repositories of most distributions. Windows users should make sure to install the Rtools package too, which is available from the same page. Once installed, you should be able to launch RGui, which gives you access to the R console and a collection of other tools (e.g. for installing add-on packages). Alternatively, you can launch the R console directly from your operating system’s terminal.

R IDEs

As you become more proficient in R, you may want to switch to a more complex ‘integrated development environment’ such as RStudio or VSCode, which provide a range of advanced tools for editing and running R code. However I recommend sticking with the basic R console to begin with, to remove distractions and help you focus on how the R environment changes as you input commands.

The R environment

R’s engine is its console, the command-line interface through which you interact with the R environment. When you first start the console, you will be presented with a welcome message like this:

R version 4.2.2 (2022-10-31) -- "Innocent and Trusting"
Copyright (C) 2022 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

  Natural language support but running in an English locale

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

>

The > character at the bottom is R’s command prompt. It indicates that R is waiting for you to tell it to do something! Try typing in a simple arithmetic expression such as 2 + 2:

2 + 2

R responds with the answer: 4 (we will get back to what the [1] means later). The numbers in this expression represent the first building block of the R environment: values. These simply represent what they are, e.g. the value 2 is the number 2.

Values

If we enter a value into the prompt on its own, R simply prints it back for us:

Values don’t have to be numbers. Text can be entered as a string or character value, which is represented by text surrounded by "double quotes":

"Hello world!"

Again, presented with a value alone, R simply prints it. We can’t do arithmetic with string values (that wouldn’t make sense), but we can manipulate them in other ways:

toupper("Hello world!")

In total, there are six basic data types that can be represented as values in R:

Type	Examples	Description
`double`	`1.2`	A decimal number, known for historical reasons as a double precision value.
`integer`	`1L`	A whole number; an integer.
`character`	`"abc"`	Text, also known as a string. Character values can also be entered with single quotes, e.g. `'abc'`, but double quotes is the norm.
`logical`	`TRUE`, `FALSE`	A binary or boolean value, representing ‘yes’ or ‘no’, ‘true’ or ‘false’, etc. Logical values can also be entered with the abbreviations `T` and `F`, but this should be avoided because it is possible to do `T <- FALSE` and wreak havoc.
`complex`	`1i`	A complex number
`raw`	`as.raw(255)`	The raw bytes underlying other values

In most cases R converts freely between doubles and integers, so they are often grouped together and called numerics. Complex and raw values are rarely encountered in practice and are included here only for completeness.

Logical values give us the possibility of using R to evaluate logical statements. For example, we can test if two values are equal to one another using the == operator:

2 == 2
1 == 2
"Hello" == "HELLO"

The result is a logical value: TRUE or FALSE. We will look at logical operations in more detail in the next section. For now, be aware that R sometimes try to be clever and interpret one data type as another:

3L == 3
3 == "3"
TRUE == 1

But it is not that clever:

3 == "three"
"false" == FALSE

Vectors

R doesn’t limit us to single values. We can combine them together into a vector by entering a list of values separated with a comma and enclosed in c( ). For example, a sequence of numbers:

c(1, 2, 3, 4, 5)

Or a series of strings:

c("red", "orange", "yellow", "green", "blue")

There is also a shortcut syntax for creating vectors with a sequence of consecutive numbers:

1:10

Vectors enable vector operations. For example, if we do arithmetic with two vectors, R will repeat the operation on each pair of values (note that this is not what a mathematician would do if asked to multiply vectors):

c(1, 2, 3) * c(3, 2, 1)
c("A", "B", "C") == c("A", "b", "c")

In fact, R considers all values to be vectors: a value like 2 is simply a vector of one number. This is why when we print even a single value it is proceeded by a [1] (indicating the initial position in a vector). Vector operations are fundamental to R and by default anything that you can do with a single value, you can also do with a vector:

c(1, 2, 3) * 2
toupper(c("red", "orange", "yellow", "green", "blue"))

Note that vectors can only contain one data type. If you try to combine more than one data type, R will try to find the ‘common denominator’:

c(1, 2, "3")

`NA` and null values

In addition to the six basic data types, R recognises a number of special values that represent the absence of a value.

The most common of these is NA, which in R means “missing data” or “unknown”. When importing tables of data into R, empty cells will be read as NA. NA values can occur in any of the six basic data types, for example:

c(1, NA, 3)
c("red", NA, "yellow")

Importantly, NA is not equivalent to 0 or FALSE; those are known values. Generally speaking, introducing NA into an expression ‘propagates’ the unknown value:

NA == 5
NA * 5

This is because R doesn’t know if NA is 5. It could be; it could not. It also doesn’t know what the result of multiplying an unknown value by 5 is.

The strictness with which R treats NAs can be frustrating at times, but it is mathematically well-founded and generally protects you from doing silly things (as we saw last week!)

Exercises

Use R to calculate the mean of these ten random numbers: 70, 34, 92, 63, 5, 77, 24, 5, 41, 17
What is the result of c(FALSE, TRUE, 2, 3) == 0:3? Why?
What is the result of NA == NA? Why?

Objects

Apart from values, R interprets everything you type into the console as the name of an object. This is why we have to surround character values with quotes, to distinguish them from objects:

"version"
version

Why do these two commands produce such different output? "version" is a value – the word ‘version’. version is an object, storing information on the version of R you have installed.

Errors

Most words aren’t the name of an object, so if you forget to enclose a character value with quotes (which is easily done), you are likely to encounter an error like this:

hello

“Error: object ‘hello’ not found” tells us fairly straightforwardly what the problem is: there is no object called hello.

This is the first of many errors you are going to see on this course! What if, for example, we try to multiply two strings, an operation that doesn’t make sense?

"Hello world!" * "Goodbye, world..."

This time the message (“non-numeric argument to binary operator”) is pretty cryptic, which is unfortunately quite common with R. Essentially what it means is that R does not know how to add (a “binary operation”) something that is not a number (a “non-numeric argument”). But however cryptic they are, it’s important that you pay attention to errors when they occur. R never raises errors without reason and it is unlikely that you will be able to continue with your work until you have resolved them.

Deciphering R error messages

As you get used to the jargon used in R’s error and warning messages, you will find that they do actually explain what the problem is. In the meantime, apart from asking your instructor, you can always try just pasting the error into a search engine. Unless you’re doing something truly exotic, it is very likely that someone else has encountered it before and has explained what it means in ordinary language.

When you run more than one line of R code at a time, an error in one line is likely to produce errors in subsequent code. In these circumstances, it’s important to start investigating the problem with the first error encountered. If you encountered an error when trying to create an object, for example, you might subsequently get the “object not found” error when trying to use it – but this doesn’t tell you the root of the problem.

Assignment

The R environment is initialised with a certain number of in-built objects, like version. Much more useful, however, is creating objects yourself. For this, we use the assignment operator, <-:

x <- 2
x

What is happening here? In the first line, we assign the value 2 to the object x. In the second line, we ask R to print the object x, which is 2.

Assignment is how we can start to build up routines that are more complex than those we could do on a calculator. For example, we can break down the exercise of calculating the mean into several steps:

total <- 10 + 20 + 30 + 40 + 50
n <- 5
total / n

As you populate the R environment with objects, it can be easy to get lost. For example, what if I forget that I’ve already used the name n and reassign it to something else? Then if I try to calculate the mean again:

n <- "N"
total / n

Entering ls() lists the objects currently in the R environment, so we can keep track:

ls()

If you need to clean things up, you can remove objects with rm():

rm(n)
ls()

Or to start completely afresh, simply restart R: objects are not saved between sessions.

Naming objects

As the work you do in R becomes more complicated, choosing good names for objects is crucial for keeping track. There are some hard rules about object names:

Object names can’t contain spaces
Object names can’t start with a number
Object names can only contain letters, numbers, _ and .
Object names are case sensitive (e.g. X is different from x)

And some good tips:

Pick a convention for separating words (e.g. snake_case or camelCase) and stick to it
It is better to use a long and descriptive name (e.g. switzerland_sites) than a short, abstract one (e.g. data)
Don’t reassign object names unless you have a good reason to

Exercises

Create objects containing a) a vector of numbers, b) a single character value and c) the sum of a set of numbers
Why does 3sites <- c("Stonehenge", "Giza", "Knossos") produce an error?
What would you call an object containing the mean site size of Neolithic sites in Germany?

Functions

Functions are a special type of object that do something with other objects or values (called arguments). You have already met several functions in this tutorial: arithmetic operators like +, logical operators like ==, and regular functions like ls().

These show two different function syntaxes. The arithmetic and logical operators use infix syntax, with two arguments and the function name (the operator) in the middle:

2 + 2

Most functions, and any that can have more than two arguments, use regular function syntax. This looks something like this:

function(argument1, argument2, argument3)

ls() is a function of this style, which takes no arguments (hence the empty brackets). An example of a function that does take arguments is sum():

sum(10, 20, 30, 40, 50)

Function arguments can also be objects:

x <- 1:100
sum(x)

Exercises

Calculate the mean of these ten random numbers using the sum() function: 90, 8, 95, 74, 26, 51, 14, 13, 83, 73
Why is it a bad idea to assign the mean of a variable to a new object called mean?

Exploratory data analysis with R

Now you’re familiar with the fundamental building blocks of the R environment, we’re ready to put them into action in exploring some data. For this, follow sections 2.1 and 2.2 in R for Data Science (2nd edition), including the exercises.

Extension exercises

Review the highlighted concepts in this tutorial and make sure that your notes cover them.
Identify and fix the errors in the following code:

libary(todyverse)

ggplot(dTA = mpg) + 
  geom_point(maping = aes(x = displ y = hwy)) +
  geom_smooth(method = "lm)

Objectives

Prerequisites

Getting started with R

The R environment

Values

Vectors

NA and null values

Exercises

Objects

Errors

Assignment

Exercises

Functions

Exercises

Exploratory data analysis with R

Extension exercises

`NA` and null values