broom (version 0.4.1)

rcorr_tidiers: Tidying methods for rcorr objects

Description

Tidies a correlation matrix from the rcorr function in the "Hmisc" package, including correlation estimates, p-values, and the number of observations in each pairwise correlation. Note that it returns these in "long", or "melted", format, with one row for each pair of columns being compared.

Usage

"tidy"(x, diagonal = FALSE, ...)

Arguments

x

An object of class "rcorr"

diagonal

Whether to include diagonal elements (where estimate is 1 and p.value is NA), default FALSE

...

extra arguments (not used)

Value

A data.frame with one row for each pairing in the correlation matrix. Columns are:

column1

Name or index of the first column being described

column2

Name or index of the second column being described

estimate

Estimate of Pearson's r or Spearman's rho

n

Number of observations used to compute the correlation

p.value

P-value of correlation

Details

Only half the symmetric matrix is shown.

Examples

if (require("Hmisc", quietly = TRUE)) {
    mat <- replicate(52, rnorm(100))
    # add some NAs
    mat[sample(length(mat), 2000)] <- NA
    # also column names
    colnames(mat) <- c(LETTERS, letters)
    
    rc <- rcorr(mat)
    
    td <- tidy(rc)
    head(td)
    
    library(ggplot2)
    ggplot(td, aes(p.value)) +
        geom_histogram(binwidth = .1)
 
    ggplot(td, aes(estimate, p.value)) +
        geom_point() +
        scale_y_log10()
}