About me:
Senior Librarian, Research & Data Services team, SMU Libraries.
Bachelor in Info Tech (IT), MSc in Info Studies from NTU.
Have been with SMU since the pandemic era (2021).
Have been doing this workshop since Aug 2023.
About this workshop:
Live-coding format; code along with me!
Goal of workshop: to give you enough fundamentals (at least to the point that ChatGPT can’t bluff you so easily) and confidence to explore R on your own.
Don’t be afraid to ask for help! We are all here to learn.
The workshops are structured to follow this workflow when dealing with data
R: The programming language and the software that interprets the R script
RStudio: An IDE (Integrated Development Environment) that we use to interact more easily with R language and scripts.
You will need to install both for this workshop. Go to https://posit.co/download/rstudio-desktop to download and install both if you have not done so.
Check out the course website for a step-by-step guide.
R Studio layout
Working directory -> where R will look for files (scripts, data, etc).
By default, it will be on your Desktop
Best practice is to use R Project to organize your files and data into projects.
When using R Project, the working directory = project folder.
Go to File
> New project
. Choose New directory
, then New project
Enter intro-r-socsci
as the name for this new folder (or “directory”) and choose where you want to put this folder, e.g. Desktop
or Documents
if you are on Windows. This will be your working directory for the rest of the workshop!
Next, let’s create 3 folders inside our working directory:
data
- we will save our raw data here. It’s best practice to keep the data here untouched.
data-output
- if we need to modify raw data, store the modified version here.
fig-output
- we will save all the graphics we created here!
Warning
Don’t put your R projects inside your OneDrive folder as that may cause issues sometimes.
Create a new R script - File
> New File
> R script
.
Note: RStudio does not autosave your progress, so remember to save from time to time!
In this line of code:
"Singapore"
is a value. This can be either a character, numeric, or boolean data type. (more on this soon)country_name
is the object where we store this value. This is so that we can keep this value to be used later.<-
is the assignment operator to assign the value to the object.
=
, but generally in R, <-
is the convention.Alt
+ -
in Windows (Option
+ -
in Mac)Non-Continuous Data
Nominal/Categorical: Non-ordered, non-numerical data, used to represent qualitative attribute.
Ordinal: Ordered non-numerical data.
Discrete: Numerical data that can only take specific value (usually integers)
Binary: Nominal data with only two possible outcome
Continuous Data
Interval: Numerical data that can take any value within a range. It does not have a “true zero”.
Ratio: Numerical data that can take any value within a range. it has a “true zero”.
The four basic data types are characters, numeric, boolean, and integer. Let’s look at examples using our WVS survey variables:
You can use str
or typeof
to check the data type of an R object.
You can do arithmetic operations in R. For example, let’s calculate average satisfaction scores:
Boolean operations in R are useful for filtering survey data. Before that, let’s look at how R evaluates simple TRUE/FALSE statements
Is life_satisfaction greater than 8?
[1] TRUE
Is the country Singapore?
Is the country NOT Singapore?
Sometimes, we may have multiple statements to evaluate. This is where the Boolean Operators will come handy.
AND operations (both conditions must be TRUE). In R, it is represented by ampersand &
Is the country New Zealand AND is the life satisfaction more than 8?
country_code == "NZL"
is FALSE while life_satisfaction > 8
is TRUE
The whole statement will return FALSE because not all conditions TRUE.
OR operations (at least one condition must be TRUE). In R, it is represented by pipe symbol |
Is the country New Zealand OR is the life satisfaction more than 8?
As long as one condition is met, this will be TRUE.
A function is like a recipe in cooking.
It takes some ingredients (inputs) and uses a set of instructions to produce a result (output).
In R, a function is a pre-written set of recipes/instructions that performs a specific task. Function name will always be followed by round brackets ()
Example: round()
function in R will round up numbers.
round()
is the “recipe”, while 3.1415926
is the “ingredients”Saving the result to an object:
Following the recipe analogy, arguments are the ingredients you provide to a function.
Some arguments are required, while others are optional (they have default values).
Each argument tells the function what to use or how to perform the task.
Example: Think of a bubble tea order as a function. The possible arguments/ingredients here are:
Tea - required ingredient
Milk - optional, the default is to include
Toppings - optional, the default choice is “pearls”
In R:
3.1415926
is the required argument (if this is not provided, the function will not run)
digits
is an optional argument specifying how many decimal places to round to (the default is 0)
You can call the help page / vignette in R by prepending ?
to the function name.
E.g. if you want to find out more about the round
function, you can run ?round
in your R console (bottom left panel)
Packages are a collections of R functions, datasets, etc. Packages extend the functionality of R.
Popular packages: tidyverse
, caret
, shiny
, etc.
Installation (you only need to do this once): install.packages("package name")
Loading packages (you need to run this everytime you restart RStudio): library(package name)
- let’s try to load tidyverse
!
In today’s session, we will explore 3 basic types of data structures in R:
Vector - can hold multiple values in a single variable/object.
Factor - Special data structure in R to handle categorical variables.
Data frame - De facto data structure for tabular data in R, and what we use for data processing, plotting, and statistics.
Basic objects in R can only contain one value. But quite often you may want to group a bunch of values together and save it in a single object.
A vector is a data structure that can do this. It is the most common and basic data structure in R. (pretty much the workhorse of R!)
Retrieve the first country in the vector
Retrieves the first three satisfaction scores
Update the first satisfaction score
Round brackets ()
are for running functions, like using a tool: mean()
or sum()
.
Square brackets []
are for accessing specific parts of your data, where we pass the index number(s) of the element(s) we want. For dataframes, we can use either index numbers or column names (more on this later!)
Let’s find high satisfaction scores (above 7)!
The code below will create a boolean vector called criteria
that basically keep tracks on whether each items inside satisfaction_scores
fulfil our condition.
The condition is “value must be > 7”. e.g. if item 1 fulfils our condition, then item 1 is ‘marked’ as TRUE
. Otherwise, it will be FALSE
[1] FALSE FALSE TRUE FALSE TRUE
criteria
to satisfaction_scores
, and only retrieve items that fulfils the condition. i.e. items whose position is marked as TRUE
by criteria
vectorNA values indicate null values, or the absence of a value (0 is still a value!)
Summary functions like mean
needs you to specify in the optional argument called na.rm
on how you want it to be handled.
Survey data often contains missing values (NA):
Several ways to add items to a vector
1satisfaction_scores <- c(satisfaction_scores, 7)
2satisfaction_scores <- c(satisfaction_scores, 8, 9, 10)
3satisfaction_scores <- c(8, satisfaction_scores)
4satisfaction_scores <- append(satisfaction_scores, 9, after = 2) # <2>
1satisfaction_scores <- satisfaction_scores[-c(2, 4)]
2satisfaction_scores <- satisfaction_scores[satisfaction_scores <= 7]
3satisfaction_scores <- na.omit(satisfaction_scores)
Special data structure in R to deal with categorical data.
Can be ordered (ordinal) or unordered (nominal).
May look like a normal vector at first glance, so use str()
to check.
Unordered (Nominal):
employment_factor <- factor(c("Full time", "Part time", "Student", "Retired", "Student"))
str(employment_factor)
Factor w/ 4 levels "Full time","Part time",..: 1 2 4 3 4
Ordered (Ordinal):
De facto data structure for tabular data in R, and what we use for data processing, plotting, and statistics.
Similar to spreadsheets!
You can create it by hand like so:
survey_data <- data.frame(
country = c("SGP", "CAN", "NZL", "SGP", "CAN"),
life_satisfaction = c(8, 7, 9, 6, 8),
employment = c("Full time", "Student", "Part time", "Retired", "Full time")
)
print(survey_data)
country life_satisfaction employment
1 SGP 8 Full time
2 CAN 7 Student
3 NZL 9 Part time
4 SGP 6 Retired
5 CAN 8 Full time
For this workshop, we will try loading a dataset from a file.
Go to the course website and go to the ‘Dataset’ tab to download the data file and information about this WVS data
Download this CSV and save it under your data
folder in your R project!
Let’s load our actual World Values Survey dataset:
Make sure to save the CSV file in your data folder!
Some basic dataframe functions before we move on to data wrangling next week:
Next Session: Data wrangling with dplyr
and tidyr
packages - we’ll learn how to:
Filter survey responses by country
Calculate average satisfaction scores by demographic groups
Create new variables from existing ones
Handle missing values in survey data
And much more!