1. Descriptive Statistics in R
Descriptive statistics give you a first look at your data. Before running any model, you need to understand the shape, centre, and spread of your variables.
Setup Link to heading
library(tidyverse)
# We'll use the built-in mtcars dataset
data(mtcars)
head(mtcars)
Measures of Central Tendency Link to heading
Mean, Median, Mode Link to heading
# Mean
mean(mtcars$mpg)
# Median
median(mtcars$mpg)
# Mode (no built-in function)
get_mode <- function(x) {
ux <- unique(x)
ux[which.max(tabulate(match(x, ux)))]
}
get_mode(mtcars$cyl)
When to use which? Link to heading
| Measure | Best for | Sensitive to outliers? |
|---|---|---|
| Mean | Symmetric distributions | Yes |
| Median | Skewed distributions | No |
| Mode | Categorical data | No |
Measures of Dispersion Link to heading
# Standard deviation
sd(mtcars$mpg)
# Variance
var(mtcars$mpg)
# Range
range(mtcars$mpg)
diff(range(mtcars$mpg))
# Interquartile range
IQR(mtcars$mpg)
# Quantiles
quantile(mtcars$mpg, probs = c(0.25, 0.5, 0.75))
Summary Statistics at a Glance Link to heading
# Base R summary
summary(mtcars$mpg)
# More detailed summary with psych package
# install.packages("psych")
library(psych)
describe(mtcars$mpg)
Frequency Tables Link to heading
# For categorical/discrete variables
table(mtcars$cyl)
# Proportions
prop.table(table(mtcars$cyl))
# Cross-tabulation
table(mtcars$cyl, mtcars$gear)
Group-wise Summaries Link to heading
mtcars %>%
group_by(cyl) %>%
summarise(
n = n(),
mean_mpg = mean(mpg),
sd_mpg = sd(mpg),
median_mpg = median(mpg),
min_mpg = min(mpg),
max_mpg = max(mpg)
)
Checking Normality Link to heading
# Shapiro-Wilk test
shapiro.test(mtcars$mpg)
# Q-Q plot
qqnorm(mtcars$mpg)
qqline(mtcars$mpg, col = "red")
Skewness and Kurtosis Link to heading
library(moments)
# install.packages("moments")
skewness(mtcars$mpg)
kurtosis(mtcars$mpg)
A skewness near 0 indicates symmetry. Kurtosis near 3 indicates a normal-like tail weight (excess kurtosis near 0).
Key Takeaways Link to heading
- Always start with descriptive statistics before modelling.
- Use
summary()anddescribe()for a quick overview. - Check normality with Shapiro-Wilk and Q-Q plots.
- Group-wise summaries using
dplyrreveal differences across subgroups.
Next: Data Visualisation →