Combine boxplots + ridgeplots using patchwork

boxridge.R

################################################################################
## Example code for generating side-by-side boxplots + raw data and ridgeplots
## NOTE: Uses Thomas Lin Pedersen's patchwork package, not yet on CRAN @ writing
################################################################################

## devtools::install_github("thomasp85/patchwork")
library(patchwork)
library(ggridges)
library(tidyverse)
  ## for data generation & management; could do this with base, of course, but I
  ## know my target audience is OK with tidyverse functions :)

## Generate example data, ~100 obs each at 5 time points; median inc across time
df <- data.frame(
  time = rep(1:5, each = 100),
  x = purrr::map(1:5, rnorm, n = 100) %>% purrr::flatten_dbl()
)

## Format data for plotting: we want to show entire distribution at the top in
## one color, plus each time point separately; concatenate original data and add
## an overall/not indicator
df_plot <- bind_rows(
  df %>% mutate(time = 0, overall = TRUE),
  df %>% mutate(overall = FALSE)
) %>%
  ## Make a factor version of time, reversing the order so that 0, 1, 2... goes
  ## top to bottom
  mutate(
    time_f = forcats::fct_rev(
      factor(time, levels = 0:5, labels = c("Overall", 1:5))
    )
  )

## -- Create boxplots ----------------------------------------------------------
bp <- ggplot(data = df_plot, aes(x = time_f, y = x, group = time_f)) +
  ## Include raw data behind boxplots
  geom_point(aes(colour = overall),
             alpha = 0.2, position = position_jitter(width = 0.1)) +
  ## Add boxplots, with no fill or outliers so raw data is visible
  ## Line size is thicker for overall than for individual times
  geom_boxplot(aes(size = overall), fill = NA, outlier.shape = NA) +
  scale_size_manual(values = c(0.5, 1.0), guide = FALSE) +

  ## Flip it so time is vertical
  coord_flip() +
  
  ## Visual tweaks to go with our usual look
  scale_colour_manual(values = c("#B40F20", "#003D74"), guide = FALSE) +
  theme_bw() +
  labs(y = "Example Boxplot", x = NULL) +
  
  theme(
    axis.ticks.y = element_blank(), ## I don't think ticks are helpful here
    axis.text.y = element_text(face = "bold", size = 14)
  )

## -- Create ridgeplots --------------------------------------------------------
## Using Claus Wilke's ggridges package
rp <- ggplot(data = df_plot, aes(y = time_f, x = x)) +
  ggridges::geom_density_ridges(aes(fill = overall), alpha = 0.8) +
  
  ## Visual tweaks
  scale_fill_manual(values = c("#B40F20", "#003D74"), guide = FALSE) +
  theme_bw() +
  
  ## Clean up that Y axis
  theme(
    axis.title.y = element_blank(),
    axis.text.y = element_blank(),
    axis.ticks.y = element_blank()
  ) +
  labs(x = "Example ridgeplot")

## -- Combine boxplots + ridgeplots to display together ------------------------
bp + rp + plot_layout(ncol = 2, widths = c(2, 1))

Result:

Note: If using a Github-only package is an issue, Winston Chang's multiplot function would also work for this!

Thank you! :D I’m more than ok with your using of the tidyverse ;)