R Graphics · Pairs Plot

Comparing Many Variables with Pairs Plots in R

Use matrix scatterplots, groups, correlations, custom panels, transformations, and selected variables.

6 examplesReproducible R code

Before you begin

Choose the variation that serves the analysis

Pairs plots show every pairwise relationship in a numeric data set. They support rapid multivariate screening for association, nonlinearity, clusters, and unusual observations.

Examples: 06
Levels: Beginner + Intermediate

Example 01Base RBeginner

Create a Basic Pairs Plot

A pairs plot creates a scatterplot for every pair of numeric variables.

Start with a small, relevant variable set; too many columns produce tiny panels and cognitive overload.

RRun this code

pairs(iris[, 1:4], pch = 19,
      col = adjustcolor("#2563eb", .45), main = "Iris measurements")

Example 02Base RBeginner

Color Observations by Group

Group colors reveal clusters and differences across every pairwise relationship.

Map factor levels to a restrained palette and add a legend outside or within an unused panel.

RRun this code

cols <- c("#2563eb", "#0891b2", "#f59e0b")[iris$Species]
pairs(iris[, 1:4], pch = 19, col = adjustcolor(cols, .6))

Example 03Base RBeginner

Display Correlations Above the Diagonal

Correlation coefficients summarize linear association alongside the raw scatterplots.

A custom upper panel can print coefficients while the lower panels preserve the observations.

RRun this code

panel.cor <- function(x, y, ...) {
  usr <- par("usr"); on.exit(par(usr)); par(usr = c(0, 1, 0, 1))
  text(.5, .5, format(cor(x, y), digits = 2), cex = 1.4)
}
pairs(iris[, 1:4], upper.panel = panel.cor, pch = 19, col = "#2563eb")

Example 04Base RIntermediate

Add Smooth Relationships

Smooth curves reveal nonlinear pairwise patterns that a coefficient may miss.

panel.smooth adds a lowess curve to each scatterplot while retaining the original points.

RRun this code

pairs(mtcars[, c("mpg", "disp", "hp", "wt")],
      lower.panel = panel.smooth, pch = 19,
      col = adjustcolor("#2563eb", .5))

Example 05Base RIntermediate

Place Histograms on the Diagonal

Diagonal histograms add the marginal distribution of every variable.

Custom diagonal panels make the display more informative without changing the pairwise panels.

RRun this code

panel.hist <- function(x, ...) {
  usr <- par("usr"); on.exit(par(usr)); par(usr = c(usr[1:2], 0, 1.5))
  h <- hist(x, plot = FALSE); rect(h$breaks[-length(h$breaks)], 0, h$breaks[-1], h$counts/max(h$counts), col = "#93c5fd")
}
pairs(iris[, 1:4], diag.panel = panel.hist, pch = 19)

Example 06Base RIntermediate

Focus on Selected Variables

A focused matrix keeps only variables relevant to the analytical question.

Selection is often the strongest design improvement: fewer panels allow larger marks, labels, and patterns.

RRun this code

vars <- mtcars[, c("mpg", "hp", "wt", "qsec")]
pairs(vars, pch = 21, bg = "#60a5fa", col = "white",
      main = "Performance and efficiency")