6 Summarising Data
In this tutorial we will use the pirates dataset from the yarrr package as our example. If you want to follow along, make sure library(yarrr) is in your library code chunk, then run:
data("pirates")Summarising vs EDA — what’s the difference? This tutorial is about quality control — checking your data arrived correctly, understanding its structure, fixing column types, and dealing with missing values. You’re making sure everything is sensible before any analysis begins.
EDA (Exploratory Data Analysis) comes next — that’s about looking for patterns and relationships using plots and grouped summaries. The two overlap, but a useful rule of thumb is: summarising first (is my data clean?), EDA second (what’s interesting?).
6.1 First look at your data
6.1.0.1 View your data — View()
The quickest way to look at your data is to click its name in the Environment pane, or type this directly into the console (not your code chunk):
View(pirates)This opens a spreadsheet-style viewer. Run this in the console rather than in a code chunk — it’s an interactive command that won’t work when you knit your document.
6.1.0.2 Check structure — str() and glimpse()
str() (structure) is usually the first command to run on any new dataset. It shows how many rows and columns you have, each column’s name, its data type, and the first few values.
str(pirates)## 'data.frame': 1000 obs. of 17 variables:
## $ id : int 1 2 3 4 5 6 7 8 9 10 ...
## $ sex : chr "male" "male" "male" "female" ...
## $ age : num 28 31 26 31 41 26 31 31 28 30 ...
## $ height : num 173 209 170 144 158 ...
## $ weight : num 70.5 105.6 77.1 58.5 58.4 ...
## $ headband : chr "yes" "yes" "yes" "no" ...
## $ college : chr "JSSFP" "JSSFP" "CCCC" "JSSFP" ...
## $ tattoos : num 9 9 10 2 9 7 9 5 12 12 ...
## $ tchests : num 0 11 10 0 6 19 1 13 37 69 ...
## $ parrots : num 0 0 1 2 4 0 7 7 2 4 ...
## $ favorite.pirate: chr "Jack Sparrow" "Jack Sparrow" "Jack Sparrow" "Jack Sparrow" ...
## $ sword.type : chr "cutlass" "cutlass" "cutlass" "scimitar" ...
## $ eyepatch : num 1 0 1 1 1 1 0 1 0 1 ...
## $ sword.time : num 0.58 1.11 1.44 36.11 0.11 ...
## $ beard.length : num 16 21 19 2 0 17 1 1 1 25 ...
## $ fav.pixar : chr "Monsters, Inc." "WALL-E" "Inside Out" "Inside Out" ...
## $ grogg : num 11 9 7 9 14 7 9 12 16 9 ...glimpse() from the tidyverse gives the same information in a slightly cleaner layout:
glimpse(pirates)## Rows: 1,000
## Columns: 17
## $ id <int> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,…
## $ sex <chr> "male", "male", "male", "female", "female", "male", "f…
## $ age <dbl> 28, 31, 26, 31, 41, 26, 31, 31, 28, 30, 25, 20, 24, 26…
## $ height <dbl> 173.11, 209.25, 169.95, 144.29, 157.85, 190.20, 158.05…
## $ weight <dbl> 70.5, 105.6, 77.1, 58.5, 58.4, 85.4, 59.6, 74.5, 68.7,…
## $ headband <chr> "yes", "yes", "yes", "no", "yes", "yes", "yes", "yes",…
## $ college <chr> "JSSFP", "JSSFP", "CCCC", "JSSFP", "JSSFP", "CCCC", "J…
## $ tattoos <dbl> 9, 9, 10, 2, 9, 7, 9, 5, 12, 12, 10, 14, 8, 9, 14, 8, …
## $ tchests <dbl> 0, 11, 10, 0, 6, 19, 1, 13, 37, 69, 1, 5, 6, 12, 70, 3…
## $ parrots <dbl> 0, 0, 1, 2, 4, 0, 7, 7, 2, 4, 3, 3, 0, 3, 0, 1, 0, 3, …
## $ favorite.pirate <chr> "Jack Sparrow", "Jack Sparrow", "Jack Sparrow", "Jack …
## $ sword.type <chr> "cutlass", "cutlass", "cutlass", "scimitar", "cutlass"…
## $ eyepatch <dbl> 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, …
## $ sword.time <dbl> 0.58, 1.11, 1.44, 36.11, 0.11, 0.59, 3.01, 0.06, 0.74,…
## $ beard.length <dbl> 16, 21, 19, 2, 0, 17, 1, 1, 1, 25, 1, 27, 0, 19, 0, 1,…
## $ fav.pixar <chr> "Monsters, Inc.", "WALL-E", "Inside Out", "Inside Out"…
## $ grogg <dbl> 11, 9, 7, 9, 14, 7, 9, 12, 16, 9, 7, 8, 12, 7, 9, 10, …Both are useful for spotting problems immediately — for example, a column of numbers that R has accidentally read in as text (chr instead of num), or a categorical column that should be a factor but isn’t.
6.2 Checking and fixing column types
When R reads in data, it makes a guess at each column’s type. It usually gets this right, but not always. Before running any analysis, it’s worth checking that every column is the type you expect.
The main data types you’ll encounter are:
| Type | What it means | Example |
|---|---|---|
num or dbl
|
Numeric (continuous) | height, weight, temperature |
int |
Integer (whole numbers) | count of tattoos, age in years |
chr |
Character (text) | names, free-text responses |
factor |
Categorical with defined levels | sex, college, treatment group |
logi |
Logical (TRUE/FALSE) | passed/failed, yes/no |
6.2.0.1 Checking the type of a single column — class()
To check one column at a time, use class() with the $ operator (datasetname$columnname):
class(pirates$sex)## [1] "character"
class(pirates$age)## [1] "numeric"
class(pirates$college)## [1] "character"If a column should be a factor but shows as chr, or should be numeric but shows as chr, you need to fix it before your analysis will work correctly.
6.2.1 Factors
A factor is R’s way of storing a categorical variable — one that takes a fixed set of values (called levels), like sex, treatment group, or college attended. Factors matter because:
- Many statistical functions (like
lm()) treat factors differently from plain text — they automatically create the correct dummy variables for you - Plots will group and order categories correctly
- R will warn you if you try to use a value that isn’t one of the defined levels
If a categorical column is stored as chr (character/text) rather than a factor, R will often still produce output — but it may be wrong, or the output may not behave as expected.
6.2.1.1 Converting a column to a factor — as.factor()
Looking at str(pirates), several columns are stored as chr (character/text) that would be better as factors for analysis: sex, college, headband, and sword.type. The eyepatch column is stored as num (0/1) but is also categorical — it should be a factor too.
pirates$sex <- as.factor(pirates$sex)
pirates$college <- as.factor(pirates$college)
pirates$headband <- as.factor(pirates$headband)
pirates$sword.type <- as.factor(pirates$sword.type)
pirates$eyepatch <- as.factor(pirates$eyepatch)After converting, run str() again to confirm the changes:
str(pirates)## 'data.frame': 1000 obs. of 17 variables:
## $ id : int 1 2 3 4 5 6 7 8 9 10 ...
## $ sex : Factor w/ 3 levels "female","male",..: 2 2 2 1 1 2 1 1 1 2 ...
## $ age : num 28 31 26 31 41 26 31 31 28 30 ...
## $ height : num 173 209 170 144 158 ...
## $ weight : num 70.5 105.6 77.1 58.5 58.4 ...
## $ headband : Factor w/ 2 levels "no","yes": 2 2 2 1 2 2 2 2 2 2 ...
## $ college : Factor w/ 2 levels "CCCC","JSSFP": 2 2 1 2 2 1 2 2 2 2 ...
## $ tattoos : num 9 9 10 2 9 7 9 5 12 12 ...
## $ tchests : num 0 11 10 0 6 19 1 13 37 69 ...
## $ parrots : num 0 0 1 2 4 0 7 7 2 4 ...
## $ favorite.pirate: chr "Jack Sparrow" "Jack Sparrow" "Jack Sparrow" "Jack Sparrow" ...
## $ sword.type : Factor w/ 4 levels "banana","cutlass",..: 2 2 2 4 2 2 2 2 2 2 ...
## $ eyepatch : Factor w/ 2 levels "0","1": 2 1 2 2 2 2 1 2 1 2 ...
## $ sword.time : num 0.58 1.11 1.44 36.11 0.11 ...
## $ beard.length : num 16 21 19 2 0 17 1 1 1 25 ...
## $ fav.pixar : chr "Monsters, Inc." "WALL-E" "Inside Out" "Inside Out" ...
## $ grogg : num 11 9 7 9 14 7 9 12 16 9 ...
6.2.1.2 Checking factor levels — levels()
levels() shows you all the categories R knows about, in the order it will use them:
levels(pirates$sex)## [1] "female" "male" "other"The first level is always the reference category in regression models — R will compare all other levels against it. By default levels are ordered alphabetically, so female would come before male.
6.2.1.3 Changing the reference level — relevel()
If you want a different reference category (for example, you want "male" to be the baseline in a regression rather than "female", which comes first alphabetically):
levels(pirates$sex) # check current order first — female is the default reference## [1] "female" "male" "other"## [1] "male" "female" "other"
6.2.1.4 Renaming factor levels — levels() assignment
To rename the categories themselves (for example, to capitalise labels for a plot):
levels(pirates$sex) # check current labels and order first## [1] "male" "female" "other"## [1] "Male" "Female" "Other"Warning: The new names must be listed in the same order as the existing levels. Always run
levels()first to check the current order before renaming.
6.2.1.5 Ordered factors
Some categorical variables have a natural order — for example, education level (low, medium, high) or survey responses (never, sometimes, always). A plain factor treats all levels as equal; an ordered factor tells R that the categories have a meaningful sequence.
The pirates dataset doesn’t have a natural ordinal column, but we can create one by grouping grogg consumption into low, medium, and high:
# Create an ordinal grogg consumption variable
pirates$grogg.level <- cut(pirates$grogg,
breaks = c(0, 7, 12, Inf),
labels = c("low", "medium", "high"))Here’s the actual code you need to create an ordered factor where you control the order.
pirates$grogg.level <- factor(pirates$grogg.level,
levels = c("low", "medium", "high"),
ordered = TRUE)
levels(pirates$grogg.level)## [1] "low" "medium" "high"
is.ordered(pirates$grogg.level)## [1] TRUEOrdered factors matter in regression — R fits them differently from unordered factors (using polynomial contrasts rather than dummy variables), so only use ordered = TRUE when the order genuinely reflects a meaningful progression.
6.3 Dealing with missing data
Missing values in R are stored as NA. It’s important to understand where your missing data is and deal with it before running any analysis, because most statistical functions will either return NA or silently drop rows if you don’t.
6.3.0.1 Checking for missing values
To count missing values in every column at once:
## id sex age height weight
## 0 0 0 0 0
## headband college tattoos tchests parrots
## 0 0 0 0 0
## favorite.pirate sword.type eyepatch sword.time beard.length
## 0 0 0 0 0
## fav.pixar grogg grogg.level
## 0 0 2skim() (covered in the next section) also shows missing value counts automatically, which is one reason it’s useful as a first QC step.
6.3.0.2 Ignoring NAs in calculations — na.rm = TRUE
Most summary functions have a built-in argument to skip NAs rather than returning NA:
mean(pirates$age, na.rm = TRUE)na.rm = TRUE works in mean(), median(), sd(), sum(), min(), max(), and most other summary functions.
6.3.0.3 Removing rows where a specific column is NA — filter()
If you want to remove rows where a particular column is missing, use filter(). This is the most precise approach:
!is.na(age) means “keep rows where age is NOT missing”. You can chain multiple columns:
6.3.0.4 Removing all incomplete rows — na.omit()
To remove any row that has an NA in any column:
pirates_clean <- na.omit(pirates)Note:
na.omit()can remove a lot of rows if your dataset has many columns. Always check how many rows you lose:
If you’ve lost more than you expected, use
filter(!is.na(columnname))instead to target only the columns that matter for your analysis.
Once you’ve cleaned your data, USE THE CLEAN VERSION IN YOUR ANALYSIS
6.4 Summarising your data
Once you’re happy that column types are correct and missing data is handled, you’re ready to summarise.
6.4.1 The whole dataset
6.4.1.1 summary()
summary() gives a statistical overview of every column — min, max, mean, and quartiles for numeric columns, and level counts for factors.
summary(pirates)Tip:
summary()output can wrap awkwardly in a knitted document. Run it in the console for easier reading.
6.4.1.2 skim()
skim() from the skimr package gives a more detailed summary, including missing value counts and a small histogram for each numeric column. It’s the most useful single command for a thorough QC check.
skim(pirates)Like summary(), skim() is often easier to read in the console than in a knitted document.
6.5 Frequency tables
For categorical columns, table() counts how many observations fall into each level:
table(pirates$sex)##
## Male Female Other
## 490 464 46For proportions instead of counts:
proportions(table(pirates$sex))##
## Male Female Other
## 0.490 0.464 0.0466.6 Troubleshooting
Got NA as your answer?
Your column probably contains missing values. Add na.rm = TRUE:
mean(pirates$age, na.rm = TRUE)Error: object 'columnname' not found
You probably forgot the $. R doesn’t know which dataset to look in for a bare column name. Use datasetname$columnname.
Factor conversion seems to have done nothing
Make sure you saved the result back to the column with <-:
pirates$sex <- as.factor(pirates$sex) # correct — saves the result
as.factor(pirates$sex) # wrong — converts but doesn't savelevels() returns NULL
The column isn’t a factor yet. Convert it with as.factor() first, then check levels.
Relabelling levels changed the wrong categories
levels() assignment replaces labels in the current level order, not alphabetically. Always run levels(columnname) first to check the order before reassigning.
Lost more rows than expected after na.omit()
na.omit() removes any row with any NA across all columns. Use filter(!is.na(columnname)) to target only the columns you care about.
summary() or skim() output looks broken in my knitted document
Run these in the console — they’re wide commands that don’t always render neatly in knitted HTML.