add confidence interval values to GGally:ggpairs plot

Question

Is it possible to add confidence interval values to GGally:ggpairs plot? Thought that it would be possible via GGally::ggpairs(data, upper = list(continuous = ggcorr)) with an adjusted version of ggcorr(), but I have been unable to understand the internal code. Ideally, a GGally::ggpairs version of [this][1] plot would be ideal.

[Plot from http://handlesman.blogspot.com/2011/03/matrix-plot-with-confidence-intervals.html][2]

EDIT:

This is my closest attempt, but can't work out adjusting the placement of the annotation for each cell in the matrix.

my_high <- function(data, mapping, ...){
  
  p <- ggally_cor(data, mapping = mapping)+annotate("text",x=0,y=0,label=paste0("[",round(cor.test(data[,1],data[,2],method = "pearson",conf.int=0.95)$conf.int[1],3),",",round(cor.test(data[,1],data[,2],method = "pearson",conf.int=0.95)$conf.int[2],3),"]"))
  return(p)
}

GGally::ggpairs(data, upper = list(continuous = my_high))```


  [1]: https://i.stack.imgur.com/XWNyb.png
  [2]: https://i.stack.imgur.com/7j2Qb.png

score 0 · Answer 1 · answered Aug 19 '21 at 15:30

Solved it!

You need to manually adjust the GGally::ggally_cor() function.

ggally_cor_New<-
function (data, mapping, ..., stars = TRUE, method = "pearson", 
          use = "complete.obs", display_grid = FALSE, digits = 3, title_args = list(...), 
          group_args = list(...), justify_labels = "right", align_percent = 0.5, 
          title = "Corr", alignPercent = warning("deprecated. Use `align_percent`"), 
          displayGrid = warning("deprecated. Use `display_grid`")) 
{
  if (!missing(alignPercent)) {
    warning("`alignPercent` is deprecated. Please use `align_percent` if alignment still needs to be adjusted")
    align_percent <- alignPercent
  }
  if (!missing(displayGrid)) {
    warning("`displayGrid` is deprecated. Please use `display_grid`")
    display_grid <- displayGrid
  }
  na.rm <- if (missing(use)) {
    NA
  }
  else {
    (use %in% c("complete.obs", "pairwise.complete.obs", 
                "na.or.complete"))
  }
  ggally_statistic(data = data, mapping = mapping, na.rm = na.rm, 
                   align_percent = align_percent, display_grid = display_grid, 
                   title_args = title_args, group_args = group_args, justify_labels = justify_labels, 
                   justify_text = "left", sep = if ("colour" %in% names(mapping)) 
                     ": "
                   else ":\n", title = title, text_fn = function(x, y) {
                     if (GGally:::is_date(x)) {
                       x <- as.numeric(x)
                     }
                     if (GGally:::is_date(y)) {
                       y <- as.numeric(y)
                     }
                     corObj <- stats::cor.test(x, y, method = method, 
                                               use = use)
                     cor_est <- as.numeric(corObj$estimate)
                     cor_txt <- formatC(cor_est, digits = digits, format = "f")
                     if (isTRUE(stars)) {
                       cor_txt <- str_c(cor_txt, signif_stars(corObj$p.value))
                       cor_CI <- as.numeric(corObj$conf.int)
                       cor_txt2 <- formatC(cor_CI, digits = digits, format = "f")
                       cor_txt <- str_c(cor_txt, paste0("[",cor_txt2[1],',',cor_txt2[2],"]"),sep="\n")
                     }
                     cor_txt
                   })
}


GGally::ggpairs(data, upper = list(continuous = ggally_cor_New))

add confidence interval values to GGally:ggpairs plot

1 Answers1