Below is a reprex of the data I'm working with. The geom_segment
calls make the rendering very sluggish. Is there an alternate way to achieve the same result more quickly?
library(ggplot2)
library(ggridges)
n <- 5000; l <- c(2, 5, 7, 9); sd_27 <- c(5.9, 11, 14, 17)
df <- data.frame(name = c(rep("A", n), rep("B", n),
rep("C", n), rep("D", n)),
value = c(rpois(n, l[1]), rpois(n, l[2]),
rpois(n, l[3]), rpois(n, l[4])))
ggplot(df, aes(x = value, y = name, fill = name)) + geom_density_ridges(alpha = 0.8) +
geom_segment(aes(x = l[[1]], y = "A", xend = l[[1]], yend = 2, color = "mean")) +
geom_segment(aes(x = l[[2]], y = "B", xend = l[[2]], yend = 3, color = "mean")) +
geom_segment(aes(x = l[[3]], y = "C", xend = l[[3]], yend = 4, color = "mean")) +
geom_segment(aes(x = l[[4]], y = "D", xend = l[[4]], yend = 5, color = "mean")) +
geom_segment(aes(x = sd_27[[1]], y = "A", xend = sd_27[[1]], yend = 2, color = "sd_27")) +
geom_segment(aes(x = sd_27[[2]], y = "B", xend = sd_27[[2]], yend = 3, color = "sd_27")) +
geom_segment(aes(x = sd_27[[3]], y = "C", xend = sd_27[[3]], yend = 4, color = "sd_27")) +
geom_segment(aes(x = sd_27[[4]], y = "D", xend = sd_27[[4]], yend = 5, color = "sd_27"))
Instead of adding each segment via a separate geom_segment
layer you could put all the data for the segments in one data frame and add the segments via one geom_segment
which according to microbenchmark
reduces the rendering time to about one fifth:
geom_segment
library(ggplot2)
library(ggridges)
set.seed(42)
n <- 5000; l <- c(2, 5, 7, 9); sd_27 <- c(5.9, 11, 14, 17)
df <- data.frame(name = c(rep("A", n), rep("B", n),
rep("C", n), rep("D", n)),
value = c(rpois(n, l[1]), rpois(n, l[2]),
rpois(n, l[3]), rpois(n, l[4])))
dl <- data.frame(x = l, y = LETTERS[1:4], yend = 2:5, color = "mean")
dsd <- data.frame(x = sd_27, y = LETTERS[1:4], yend = 2:5, color = "sd_27")
d <- do.call(rbind, list(dl, dsd))
p1 <- function() {
ggplot(df, aes(x = value, y = name, fill = name)) +
geom_density_ridges(alpha = 0.8) +
geom_segment(data = d, aes(x = x, y = y, xend = x, yend = yend, color = color), inherit.aes = FALSE)
}
p2 <- function() {
ggplot(df, aes(x = value, y = name, fill = name)) + geom_density_ridges(alpha = 0.8) +
geom_segment(aes(x = l[[1]], y = "A", xend = l[[1]], yend = 2, color = "mean")) +
geom_segment(aes(x = l[[2]], y = "B", xend = l[[2]], yend = 3, color = "mean")) +
geom_segment(aes(x = l[[3]], y = "C", xend = l[[3]], yend = 4, color = "mean")) +
geom_segment(aes(x = l[[4]], y = "D", xend = l[[4]], yend = 5, color = "mean")) +
geom_segment(aes(x = sd_27[[1]], y = "A", xend = sd_27[[1]], yend = 2, color = "sd_27")) +
geom_segment(aes(x = sd_27[[2]], y = "B", xend = sd_27[[2]], yend = 3, color = "sd_27")) +
geom_segment(aes(x = sd_27[[3]], y = "C", xend = sd_27[[3]], yend = 4, color = "sd_27")) +
geom_segment(aes(x = sd_27[[4]], y = "D", xend = sd_27[[4]], yend = 5, color = "sd_27"))
}
# Check plot
p1()
#> Picking joint bandwidth of 0.381
# Compare running time
microbenchmark::microbenchmark(p1())
#> Unit: milliseconds
#> expr min lq mean median uq max neval
#> p1() 1.859514 1.917135 2.162416 1.936781 2.42122 5.056147 100
microbenchmark::microbenchmark(p2())
#> Unit: milliseconds
#> expr min lq mean median uq max neval
#> p2() 9.37298 9.669749 10.20821 9.774624 10.17852 22.42459 100