CODEWITHSUNDEEP
rggally
Barun
Barun

Reputation: 197

Is there a way to add robust correlations and significance value/stars using WRS2 in GGally: ggpairs plots?

ggpairs plot computes correlations using base cor method only. How can we supply the pbcor or wincor values/p-values (from WRS2 package) to the plot?

Upvotes: 0

Views: 25

Answers (1)

jkd
jkd

Reputation: 1654

ggpairs allows you to provide custom functions for the different facets of your matrix (as arguments to upper, lower and diag). There is also a ggally_statistic function that allows you to print results from other functions/tests into those facets. By wrapping this function (using the wrapping mechanism provided by the GGally package) with the correct parameters, we can easily create a custom function using the required correlation measure.

library(GGally)

d <- data.frame(A = rnorm(100),
                B = rnorm(100),
                C = rnorm(100))

mcor <- wrap(ggally_statistic,
             title = "pbcor", # name of the metric to be displayed
             text_fn = function(x, y) {
               corObj <- WRS2::pbcor(x, y) # use the desired function here or add arguments
               cor_txt <- formatC(corObj$cor, digits = 3, format = "f")
               cor_txt <- stringr::str_c(cor_txt, signif_stars(corObj$p.value))
               cor_txt
             })
ggpairs(d, upper = list(continuous = mcor))

Example of a ggpairs matrix with pbcor

Upvotes: 1

Related Questions