Univariate distribution dynamics in R:

Using the ggridges Package

Suggested Citation:

Mendez Carlos (2020). Univariate distribution dynamics in R: Using the ggridges package. R Studio/RPubs. Available at https://rpubs.com/quarcs-lab/univariate-distribution-dynamics

This work is licensed under the Creative Commons Attribution-Share Alike 4.0 International License.

1 Load the Data

Let us use a dataset containing the per-capita GDP of 152 Countries over the 1970-2003 period. The data is from the package ConvergenceClubs

data(GDP)
GDP

2 Data transformations

2.1 Pivot from wide to long

dat_long <- GDP %>% 
  pivot_longer(
    cols = starts_with("Y"),
    names_prefix = "Y",
    names_to = "year",
    values_to = "GDPpc"
  )
dat_long

2.2 Take the log of GPDpc

dat_long <- dat_long %>% 
  mutate(
    ln_GPDpc = log(GDPpc)
  )

2.3 Focus on the last the years

dat_long <- dat_long %>% 
  filter(
    year >= 1998
  )

3 Plot distribution dynamics

dat_long %>% 
  ggplot(aes(x = ln_GPDpc, y = year)) +
  geom_density_ridges() +
  theme_ridges()

3.1 Scale parameter

A setting of scale=1 means the tallest density curve just touches the baseline of the next higher one.

dat_long %>% 
  ggplot(aes(x = ln_GPDpc, y = year)) +
  geom_density_ridges(scale = 1) + 
  theme_ridges()

Larger scale values create more overlap.

dat_long %>% 
  ggplot(aes(x = ln_GPDpc, y = year)) +
  geom_density_ridges(scale = 10) + 
  theme_ridges()

3.2 Fill colors

dat_long %>% 
  ggplot(aes(x = ln_GPDpc, y = year, fill = stat(x))) +
  geom_density_ridges_gradient(scale = 10) +
  scale_fill_viridis_c(name = "Ln of GPDpc") +
  theme_ridges()

3.3 Map the probabilities directly onto color.

dat_long %>% 
  ggplot(aes(
    x = ln_GPDpc,
    y = year,
    fill =  0.5 - abs(0.5 - stat(ecdf))
            )
        ) +
stat_density_ridges(
  geom = "density_ridges_gradient",
  calc_ecdf = TRUE,
  scale = 10
                    ) +
  scale_fill_viridis_c(name = "Tail probability",
                       direction = -1) +
  theme_ridges()

3.4 Indicate quantiles

3.4.1 Median

dat_long %>% 
  ggplot(aes(x = ln_GPDpc, y = year)) +
  stat_density_ridges(quantile_lines = TRUE, 
                      quantiles = 2,
                      scale = 10) +
  theme_ridges()

3.4.2 Any quantile

dat_long %>% 
  ggplot(aes(x = ln_GPDpc, y = year)) +
  stat_density_ridges(quantile_lines = TRUE, 
                      quantiles = c(0.025, 0.975),
                      scale = 10,
                      alpha = 0.7) +
  theme_ridges()

3.4.3 Color by Quartile

dat_long %>% 
  ggplot(aes(x = ln_GPDpc,
             y = year,
             fill = factor(stat(quantile))
             )
         ) +
  stat_density_ridges(
    geom = "density_ridges_gradient", 
    calc_ecdf = TRUE,
    quantiles = 4, 
    quantile_lines = TRUE,
    scale = 10
    ) +
  scale_fill_viridis_d(name = "Quartiles") +
  theme_ridges()

3.5 Further modifications

dat_long %>% 
  ggplot(aes(x = ln_GPDpc,
             y = year,
             fill = factor(stat(quantile))
             )
         ) +
  stat_density_ridges(
    geom = "density_ridges_gradient", 
    calc_ecdf = TRUE,
    quantiles = 4, 
    quantile_lines = TRUE,
    scale = 10
    ) +
  scale_fill_viridis_d(name = "Quartiles") +
  theme_ridges() +
  scale_x_continuous(expand = c(0, 0)) +
  scale_y_discrete(expand = c(0, 0)) +
  coord_cartesian(clip = "off") 

dat_long %>% 
  ggplot(aes(x = ln_GPDpc,
             y = year,
             fill = factor(stat(quantile))
             )
         ) +
  stat_density_ridges(
    geom = "density_ridges_gradient", 
    calc_ecdf = TRUE,
    quantiles = 4, 
    quantile_lines = TRUE,
    scale = 10
    ) +
  scale_fill_viridis_d(name = "Quartiles") +
  scale_x_continuous(expand = c(0, 0)) +
  scale_y_discrete(expand = c(0, 0)) +
  coord_cartesian(clip = "off") + 
  theme_ridges(grid = FALSE, center_axis_labels = TRUE)

4 References

• https://wilkelab.org/ggridges/
• https://wilkelab.org/ggridges/articles/introduction.html

END