Clip {ggdist} ccdf-barplot-CodePudding

I'm making a complementary cumulative distribution function barplot with {ggdist}. When I export the plot to svg (or other vector representation), I notice that there is a zero-width stripe protruding from the polygon (see attached image). I rather not have this protruding stripe.

library(ggplot2)
library(ggdist)

df <- data.frame(
  x = rep(c("A", "B"), each = 10),
  y = c(rnorm(20, mean = rep(c(5, 7), each = 10)))
)

p <- ggplot(df, aes(x, y))  
  stat_ccdfinterval(geom = "slab")  
  lims(y = c(0, NA))

ggsave("test.svg", plot = p, device = svglite::svglite)

The code above results in an svg file below, wherein I've highlighted the outline of the polygon with the stripe.

I'd like to get rid of that zero-width stripe. I tried setting the relevant aesthetic to NA at these points, but that also deletes one of the corners that is not part of the stripe.

ggplot(df, aes(x, y))  
  stat_ccdfinterval(
    geom = "slab", 
    aes(thickness = after_stat(ifelse(f == 0, NA, f)))
  )  
  lims(y = c(0, NA))

^{Created on 2022-03-10 by the reprex package (v2.0.1)}

CodePudding user response：

I found a solution specific to this problem, but it might pan out differently if the orientation is horizontal or the cdf instead of the ccdf is used. In brief, we're still setting 0-thickness datapoints to NA, but we now do this only where the y aesthetic exceeds the groupwise maximum.

library(ggplot2)
library(ggdist)

df <- data.frame(
  x = rep(c("A", "B"), each = 10),
  y = c(rnorm(20, mean = rep(c(5, 7), each = 10)))
)

helper <- function(f, y, group) {
  split(f, group) <- Map(
    function(value, y) {
      f <- value
      max_y <- max(y[f != 0])
      f[f == 0 & y > max_y] <- NA
      f
    },
    value = split(f, group),
    y     = split(y, group)
  )
  f
}

ggplot(df, aes(x, y))  
  stat_ccdfinterval(
    geom = "slab", 
    aes(thickness = after_stat(helper(f, y, group)))
  )  
  lims(y = c(0, NA))

^{Created on 2022-03-13 by the reprex package (v2.0.1)}