plotcorr: Compact Visualization of a Correlation Matrix

In felix-hof/biostatUZH: Misc Tools of the Department of Biostatistics, EBPI, University of Zurich

Compact Visualization of a Correlation Matrix

Description

This function plots a correlation matrix using ellipse-shaped glyphs for each entry. The ellipse represents a level curve of the density of a bivariate normal with the matching correlation.

Usage

plotcorr(
  corr,
  outline = TRUE,
  col = TRUE,
  lower.panel = "ellipse",
  upper.panel = "number",
  diag.panel = NULL,
  bty = "n",
  axes = FALSE,
  xlab = "",
  ylab = "",
  asp = 1,
  cex.lab = par("cex.lab"),
  cex = 0.75 * par("cex"),
  mar = 0.1 + c(2, 2, 4, 2),
  ...
)

Arguments

corr	A matrix containing entries between -1 and 1 to be plotted as correlations.
outline	Whether the ellipses should be outlined in the default colour.
col	Which colour(s) to use to fill the ellipses (recycled to length(corr)). The special setting col = TRUE will fill the ellipses according to the color range colorRampPalette(c("blue", "white", "red"))(11) mapped to the correlations as shown in the example below.
lower.panel, upper.panel, diag.panel	each panel can be either NULL, "ellipse", or "number". NULL disables the respective part, and "number" plots numerical correlations in place of ellipses. Correlations will be rounded to a single decimal place. The ellipse package is required for "ellipse" panels.
bty, axes, xlab, ylab, asp, mar, cex.lab, ...	Graphical parameters which will be passed to plot when plotting.
cex	Graphical parameter which will be passed to text when plotting.

Details

This is a fork of the original plotcorr function from the ellipse package as at version 0.3-8. The arguments numbers, type, and diag have been replaced by lower.panel, upper.panel, and diag.panel similar to pairs. This enables displaying numbers in one triangle and ellipses in the other. However, there is no support for diag = FALSE in the original sense of the function.

The ellipses being plotted will be tangent to a unit character square, with the shape chosen to match the required correlation.

Author(s)

Sebastian Meyer, Duncan Murdoch

References

Murdoch, D.J. and Chow, E.D. (1996). A graphical display of large correlation matrices. The American Statistician 50, 178-180.

See Also

plotcorr

Examples

if (requireNamespace("ellipse")) {
  ## Plot the correlation matrix for the mtcars data full model fit 
  data("mtcars")
  fit <- lm(mpg ~ ., mtcars)
  corr.fit <- summary(fit, correlation = TRUE)$correlation
  plotcorr(corr.fit, col = "gray")

  ## with default color coding
  plotcorr(corr.fit, col = TRUE)

  ## Colour the ellipses and order by correlations with miles/gallon
  corr.mtcars <- cor(mtcars)
  ord <- order(corr.mtcars[1,])
  xc <- corr.mtcars[ord, ord]
  colors <- colorRampPalette(c("blue", "white", "red"))(11)
  plotcorr(xc, col = colors[5*xc + 6])
}

felix-hof/biostatUZH documentation built on Sept. 27, 2024, 1:48 p.m.