Correlation Plot — correlation

Plots a correlation matrix across all numeric columns of data using corrplot.

Usage

correlation_plot(
  data,
  cor.use = "everything",
  cor.method = "pearson",
  method = "ellipse",
  type = "upper",
  order = "FPC",
  title = "Correlation Coefficient Plot",
  na_omit = TRUE,
  tl.cex = 0.5,
  number.cex = 0.75,
  tl.srt = 45,
  mar = c(0, 0, 2, 0),
  show_coef = TRUE,
  ...
)

Arguments

data: A data frame or tibble.
cor.use: Passed to cor: "everything" (default), "all.obs", "complete.obs", "na.or.complete", or "pairwise.complete.obs".
cor.method: Passed to cor: "pearson" (default), "kendall", or "spearman".
method: Visualization method passed to corrplot (default "ellipse").
type: "upper" (default), "full", or "lower".
order: Ordering method for the correlation matrix (default "FPC", first principal component order).
title: Plot title.
na_omit: Logical (default TRUE); drop rows with any NA among the numeric columns before computing correlations.
tl.cex, number.cex, tl.srt: Label/number sizing and rotation, passed to corrplot.
mar: Plot margin, passed to corrplot in the standard par("mar") form c(bottom, left, top, right) (default c(0, 0, 2, 0)). corrplot() does not otherwise reserve any extra space above the matrix for title, so with the default (zero) margin the title text collides with the 45-degree diagonal column labels sitting just below it; the default here adds two lines of top margin so the title clears the labels. Increase further if using a long title or a larger tl.cex.
show_coef: Logical (default TRUE); whether to draw the correlation coefficient inside each cell. On a small matrix the numbers add useful precision on top of the visual encoding, but on a large matrix (many variables) they quickly overlap each other and the diagonal labels. Set to FALSE to omit them entirely – this is the clean way to declutter a large matrix; shrinking tl.cex and number.cex down to near-zero to visually "hide" the numbers (as earlier callers did) also erases the variable name labels and should be avoided. See correlation_plot_split for an alternative that keeps coefficients readable by splitting a large matrix into several smaller plots instead of hiding them.
...: Additional arguments passed to corrplot.

Value

Invisibly, the correlation matrix (called for its plot side effect).

Examples

correlation_plot(mtcars, tl.cex = .7)

correlation_plot(mtcars, tl.cex = .7, show_coef = FALSE)