Plot a Heatmap for a Correlation Matrix

Generates a heatmap of the upper triangle of a correlation matrix.

plot_correlation_heatmap(
  x,
  ttl = "Correlation matrix",
  labels = NULL,
  lyt = NULL,
  gradient = c("#E69F00", "#56B4E9"),
  txtSz = 1.25,
  mc_adjust = "BH",
  cut_off = 0.05,
  new = T,
  H = 20/3,
  W = 25/3,
  abbr_labels = TRUE,
  status = FALSE
)

Arguments

x

A data frame (all variables will be used when generating the correlation matrix).

ttl

An optional title for the figure.

labels

The labels for the rows/columns. Users can pass a list with two character vectors of matching length to provide separate labels for rows and columns.

lyt

An optional matrix specifying the layout of the main panel (1) versus the side panel (2) with the color gradient.

gradient

The final end colors for the negative and positive correlations, respectively.

txtSz

The size of the text in the figure (a second value can be provided to adjust variable labels separately).

mc_adjust

The method to use when correcting for multiple comparisons (see p.adjust).

cut_off

Cut-off for statistical significance.

new

Logical; if TRUE generates a new plotting window via x11.

H

The height in inches of the figure if a new plotting window is generated.

W

The width in inches of the figure if a new plotting window is generated.

abbr_labels

Logical; if TRUE abbreviates labels to 4 characters. A second value can be provided to abbreviate labels on the top separately.

status

Logical; if TRUE displays the progress of the function for debugging purposes.

Value

A heatmap for the upper-triangle portion of the correlation matrix.

Examples

# Load data
data("mtcars")
x <- mtcars[, c(1,3,4,5,6,7)]
plot_correlation_heatmap( x, new = FALSE )


# Simulate a correlation matrix

# 5 x 5 matrix of random values
rand_mat <- matrix( rnorm(25), 5, 5 )
# Create covariance matrix by
# multiplying matrix by its transpose
cov_mat <- rand_mat %*% t( rand_mat )
corr_mat <- cov_mat/sqrt(diag(cov_mat)%*%t(diag(cov_mat)))

# Simulate data
x <- MASS::mvrnorm( 100, rep( 0, nrow( corr_mat ) ), corr_mat )
colnames( x ) <- paste0( 'V', 1:ncol( x ) )
x <- data.frame(x)
plot_correlation_heatmap( x, new = FALSE )