Examples of lattice corrgrams"
In corrgram: Plot a Correlogram

Package overview

The corrgram package provides functions for creating corrgrams using three different graphics systems, base, grid, and lattice.

Base R graphics + single function corrgram() for dataframes or matrices. + Enables most features found in the paper by @friendly2002corrgrams. - No automatic legend. - Not easily combined with other graphics.

lattice graphics + Separate panel functions for lattice::levelplot() for dataframes and lattice::splom() for correlation matrices. + Enables automatic legend. + Enables corrgrams conditioned on other variables. + Can be combined with other lattice graphics for complex figures. - Not feature complete compared to base R.

grid graphics + single function corrgram2() for either dataframes or correlation matrices. + Enables automatic legend. + Can be combined with other grid graphics for complex figures. - Not feature complete compared to base R. + Faster than base R when evaluated inside Positron.

This vignette

This vignette demonstrates how to create corrgrams using lattice graphics, you can use some custom panel functions provided in the corrgram package along with the lattice::splom() or lattice::levelplot() functions. An example of each type of corrgram is shown below.

Correlation matrix corrgram in lattice

The levelplot() function in lattice only has a single plotting region, so does not (by default) suppot upper and lower panels. However, you can write a custom panel function with different glyphs above and below the diagonal. See the panel.ellipse() example below.

Using splom() makes it easy to include a color scale next to the corrgram.

library("lattice")
library("corrgram")

# The easiest way to have an automatic color key is to set the theme
opar <- trellis.par.get()
trellis.par.set(
  regions=list(col=colorRampPalette(c("red","salmon","white","royalblue","navy")))
)

# Create a correlation matrix
library(MASS) # foor Cars93
cor.Cars93 <-   cor(Cars93[, !sapply(Cars93, is.factor)], use = "pair")
ord <- order.dendrogram(as.dendrogram(hclust(dist(cor.Cars93))))
cars93 <- cor.Cars93[ord,ord]
head(cars93)

# lattice corrgram using pie-shaped glyphs
levelplot(cars93, xlab = NULL, ylab = NULL,
          at = do.breaks(c(-1.01, 1.01), 101), panel = levelplot_panel.pie,
          scales = list(x = list(rot = 90)), colorkey = list(space = "top") )

# lattice corrgram using ellipse-shaped glyphs above the diagonal
# and value labels below the diagonal
levelplot(cars93, xlab = NULL, ylab = NULL,
          at = do.breaks(c(-1.01, 1.01), 101), panel = levelplot_panel.ellipse,
          label=TRUE,
          scales = list(x = list(rot = 90)), colorkey = list(space = "top") )

Scatterplot matrix corrgram in lattice

Since the lattice::splom() function supports conditioning on a factor, we can use it to create corrgrams that are conditioned on a factor.

The penguins data provides a nice example of Simpson's paradox, where the overall correlation between two variables can be negative, but the correlation within each group (species) can be positive.

pengvars <- c("bill_len", "bill_dep", "flipper_len", "body_mass")
library(lattice)
splom(~penguins[ , pengvars], upper.panel=splom_panel.pie, pscales=0)
splom(~penguins[ , pengvars]|penguins$species, upper.panel=splom_panel.pie, pscales=0)
splom(~penguins[ , pengvars], upper.panel=splom_panel.shade, pscales=0)
splom(~penguins[ , pengvars]|penguins$species, upper.panel=splom_panel.shade, pscales=0)
splom(~penguins[ , pengvars], upper.panel=splom_panel.ellipse, pscales=0)
splom(~penguins[ , pengvars]|penguins$species, upper.panel=splom_panel.ellipse, pscales=0)

You can also use the hexbin package to create hexagonal binning plots in the upper panels of the scatterplot matrix. This is useful when you have a large number of points and want to visualize the density of points in different regions of the plot.

# Hexbin
library(lattice)
library(hexbin)
splom(~penguins[ , pengvars], upper.panel=hexbin::panel.hexbinplot, pscales=0)
splom(~penguins[ , pengvars]|penguins$species, upper.panel=hexbin::panel.hexbinplot, pscales=0)