ggheat2: ggheat2

In raredd/plotr: More plotting functions and examples

ggheat2

Description

Function to plot a heat map using ggplot.

Usage

ggheat2(
  data,
  corr = cor(data, use = "pairwise.complete"),
  cluster = TRUE,
  nbreaks = NULL,
  palette = if (is.null(nbreaks)) c("blue", "white", "red") else 1,
  legend_name = expression(rho),
  pch,
  cex = c(2, 6),
  label = FALSE,
  label_alpha = FALSE,
  label_color = "black",
  label_digits = 2,
  midpoint = 0,
  clim = c(-1, 1),
  ...
)

Arguments

data	a data frame or matrix (observations x variables) of numeric values
corr	a correlation matrix
cluster	logical or function; if TRUE, the variables will be clustered and reordered; if FALSE, no reordering will be done; otherwise, a custom clustering function may be given; see details
nbreaks	number of breaks to categorize the correlations (default is NULL, ie, a continuous scale)
palette	for a continuous scale, a vector of length three giving the low, mid, and high colors of the gradient (default is c('blue','white',''red), see scale_colour_gradient2); for a discrete scale, a character string of the palette name or an integer giving the index of the palette, see scale_colour_brewer
legend_name	the legend name; see discrete_scale
pch	(optional) plotting character; if missing, geom_tile is used instead of plotting characters
cex	size of pch; a vector of length one (all pch will be the same size) or two (size will be proportional to the correlation); default is c(2,6); seescale_size_identity
label	logical; if TRUE, adds correlation coefficients on top of each pch
label_alpha	logical, if TRUE, adds alpha transparency when label is TRUE (correlations closer to 0 will be less visible)
label_color	color of correlations (default is 'black')
label_digits	number of digits in correlation labels
midpoint	the midpoint value for continuous scaling of correlations (default is 0)
clim	vector of length two giving the limits of correlation coefficients (default is -1,1)
...	additional arguments passed to geom_text for the diagonal labels

Details

Default cluster method is stats::hclust(dist(x), method = 'average') which will return a list containing a named vector, "order", which is used to reorder the variables.

In order to pass a custom clustering function to cluster, the function must take a single input (a correlation matrix) and return either a vector or a list with a named vector, "order".

See Also

cor, ggheat, icorr, corrplot, https://github.com/briatte/ggcorr

Examples

library('ggplot2')
ggheat2(mtcars)

ggheat2(mtcars, label = TRUE, label_alpha = TRUE, cluster = FALSE,
        ## additional args passed to diagonal labels
        colour = 'red', angle = 45, size = 7)

ggheat2(mtcars, pch = 19, nbreaks = 6, cex = c(2,10),
        palette = 'PuOr',         ## colorblind palette
        size = 5, hjust = 0.75) + ## passed to diag text
    labs(title = 'Correlation Matrix')

## custom clustering function
ggheat2(data = NULL, corr = cor(mtcars, use = 'pairwise'),
        nbreaks = 5, palette = 'Blues',
        cluster = function(...) sample(ncol(mtcars)))

raredd/plotr documentation built on Nov. 19, 2023, 4:09 a.m.