CODEWITHSUNDEEP
rggplot2ggridges
Shannon Pileggi
Shannon Pileggi

Reputation: 101

ggplot in r: duplicate faceted geom_density and geom_density_ridges results

The default settings for ggplot2::geom_density() with a facet and ggridges::geom_density_ridges() yield slightly different curves. How can I modify the smoothing technique in one or the other to yield the same result?

library(tidyverse)
library(ggridges)

# standard density with facet by cyl ----
mtcars %>%
  ggplot(aes(x = mpg)) +
  geom_density(fill = "gray") +
  facet_wrap(. ~ factor(cyl, levels = c(8, 6, 4)), ncol = 1) +
  theme_minimal()


# density ridge with y = cyl ----
mtcars %>%
  ggplot(aes(x = mpg, y = factor(cyl))) +
  geom_density_ridges() +
  theme_minimal()
#> Picking joint bandwidth of 1.38

Created on 2019-04-04 by the reprex package (v0.2.1)

Upvotes: 3

Views: 794

Answers (1)

Claus Wilke
Claus Wilke

Reputation: 17790

You can use the same stat that geom_density() uses.

library(tidyverse)
library(ggridges)

# standard density with facet by cyl ----
mtcars %>%
  ggplot(aes(x = mpg)) +
  geom_density(fill = "gray") +
  facet_wrap(. ~ factor(cyl, levels = c(8, 6, 4)), ncol = 1) +
  theme_minimal()


# density ridge with y = cyl ----
mtcars %>%
  ggplot(aes(x = mpg, y = factor(cyl))) +
  geom_density_ridges(stat = "density", aes(height = stat(density))) +
  theme_minimal()

Created on 2019-04-04 by the reprex package (v0.2.1)

Alternatively, you can take the bandwidth that geom_density_ridges() reports and use it in geom_density() (here, bw = 1.38).

library(tidyverse)
library(ggridges)

# density ridge with y = cyl ----
mtcars %>%
  ggplot(aes(x = mpg, y = factor(cyl))) +
  geom_density_ridges() +
  theme_minimal()
#> Picking joint bandwidth of 1.38


# standard density with facet by cyl ----
mtcars %>%
  ggplot(aes(x = mpg)) +
  geom_density(fill = "gray", bw = 1.38) +
  facet_wrap(. ~ factor(cyl, levels = c(8, 6, 4)), ncol = 1) +
  theme_minimal()

Created on 2019-04-04 by the reprex package (v0.2.1)

The final two plots look slightly different because they have different x axis limits.

Upvotes: 1

Related Questions