Visualization of a Correlation Matrix using 'ggplot2'

The 'ggcorrplot' package can be used to visualize easily a correlation matrix using 'ggplot2'. It provides a solution for reordering the correlation matrix and displays the significance level on the plot. It also includes a function for computing a matrix of correlation p-values.

ggcorrplot: Visualization of a correlation matrix using ggplot2

ggcorrplot draws a correlation matrix as a ggplot2 plot. Because the result is a plain ggplot object, you can restyle it, annotate it, and combine it with other layers using the usual + syntax.

It can:

reorder the matrix by hierarchical clustering and outline the clusters,
show only the lower or upper triangle, or a mixed layout with a different glyph per triangle,
overlay the correlation coefficients and mark the statistically significant cells (including a standalone significance map), and
compute the matrix of correlation p-values with cor_pmat().

Learn more in ggcorrplot: Correlation Matrix Heatmap in R with ggplot2.

Installation

Install the released version from CRAN:

install.packages("ggcorrplot")

Or the development version from GitHub:

if (!require(devtools)) install.packages("devtools")
devtools::install_github("kassambara/ggcorrplot")

library(ggcorrplot)

Getting started

The examples below use the mtcars data set. cor() builds the correlation matrix and cor_pmat() [in ggcorrplot] computes the matrix of correlation p-values.

data(mtcars)
corr <- round(cor(mtcars), 1)
corr[1:4, 1:4]
#>       mpg  cyl disp   hp
#> mpg   1.0 -0.9 -0.8 -0.8
#> cyl  -0.9  1.0  0.9  0.8
#> disp -0.8  0.9  1.0  0.8
#> hp   -0.8  0.8  0.8  1.0

# Matrix of correlation p-values
p.mat <- cor_pmat(mtcars)
p.mat[1:4, 1:4]
#>               mpg          cyl         disp           hp
#> mpg  0.000000e+00 6.112687e-10 9.380327e-10 1.787835e-07
#> cyl  6.112687e-10 0.000000e+00 1.802838e-12 3.477861e-09
#> disp 9.380327e-10 1.802838e-12 0.000000e+00 7.142679e-08
#> hp   1.787835e-07 3.477861e-09 7.142679e-08 0.000000e+00

Correlation matrix visualization

The default draws each correlation as a colored square; method = "circle" encodes the value with the circle area instead.

ggcorrplot(corr)
ggcorrplot(corr, method = "circle")

Sized glyphs in boxed cells

scale.square = TRUE sizes the squares by the absolute correlation, so strong correlations dominate; cell.grid = TRUE draws a light box around every cell so the glyphs sit inside a grid instead of floating on the axis lines.

ggcorrplot(corr, scale.square = TRUE, cell.grid = TRUE, outline.color = "white")
ggcorrplot(corr, method = "circle", cell.grid = TRUE)

Reorder by clustering, and outline the clusters

hc.order = TRUE reorders the variables by hierarchical clustering so that correlated variables sit together. hc.rect then draws rectangles around the clusters obtained by cutting the tree.

ggcorrplot(corr, hc.order = TRUE, outline.color = "white")
ggcorrplot(corr, hc.order = TRUE, hc.rect = 3, outline.color = "white")

Lower / upper triangle

For a symmetric matrix the two triangles are redundant, so you can keep just one.

ggcorrplot(corr, hc.order = TRUE, type = "lower", outline.color = "white")
ggcorrplot(corr, hc.order = TRUE, type = "upper", outline.color = "white")

Mixed layout

lower.method and upper.method draw a different glyph in each triangle — here the coefficients as numbers below the diagonal and circles above it, with the variable names on the diagonal. Adding cell.grid = TRUE boxes every cell for the tidy corrplot look.

ggcorrplot(corr,
  lower.method = "number", upper.method = "circle",
  cell.grid = TRUE, show.legend = FALSE
)

Add the coefficients

ggcorrplot(corr, hc.order = TRUE, type = "lower", lab = TRUE)

Highlighting significance

Passing p.mat marks the cells whose correlation is not significant at sig.level (default 0.05). By default a cross is drawn over them (insig = "pch"); insig = "blank" hides them instead.

# Cross out the non-significant coefficients
ggcorrplot(corr, hc.order = TRUE, type = "lower", p.mat = p.mat)
# Leave them blank
ggcorrplot(corr, hc.order = TRUE, type = "lower", p.mat = p.mat, insig = "blank")

Significance map

insig = "stars" flips the emphasis: instead of crossing out the non-significant cells, it marks the significant ones with significance stars (***, **, * for p <= 0.001, 0.01, 0.05 – fixed thresholds, not sig.level). With the default lab = FALSE this is a standalone significance map; with lab = TRUE the stars are appended to the coefficients (e.g. -0.85***).

ggcorrplot(corr, p.mat = p.mat, insig = "stars")

Colors and theme

ggcorrplot() returns a ggplot object, so any ggplot2 theme applies. colors sets the low / mid / high gradient. For the ggplot2 side of this, see ggplot2 Colours in R: Change Colours by Group and ggplot2 Themes in R: Customize the Look.

ggcorrplot(corr,
  hc.order = TRUE, type = "lower", outline.color = "white",
  ggtheme = ggplot2::theme_gray,
  colors = c("#6D9EC1", "white", "#E46726")
)