Graphic Design with ggplot2

Concepts of the {ggplot2} Package Pt. 1:
Solution Exercise 2

Cédric Scherer // rstudio::conf // July 2022

Exercise 2

  • Explore the TfL bike sharing data visually:
    create a boxplot of counts per weather type
    • Turn the plot into a jitter strips plot (random noise across the x axis)
    • Combine both chart types (jittered points on top of the boxplots)
    • Bonus: Sort the boxplot-jitter hybrid by median counts
    • Apply your favorite theme to the plot.
    • Add meaningful labels.
    • Bonus: Explore other chart types to visualize the distributions.
  • Save the plot as a vector graphic with a decent plot size.

Import Data (if not yet)

bikes <- readr::read_csv(
  here::here("data", "london-bikes-custom.csv"),
  col_types = "Dcfffilllddddc"
)

bikes$season <- forcats::fct_inorder(bikes$season)

library(tidyverse)

Boxplot of Counts vs. Weather Type

ggplot(
    bikes,
    aes(x = weather_type, y = count)
  ) +
  geom_boxplot()

Avoid Overlapping Axis Labels

ggplot(
    bikes,
    aes(x = count, y = weather_type)
  ) +
  geom_boxplot()

Avoid Overlapping Axis Labels

ggplot(
    bikes,
    aes(x = stringr::str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot()

Apply a Theme

theme_set(theme_minimal(
  base_size = 14,
  base_family = "Roboto Condensed"
))

ggplot(
    bikes,
    aes(x = stringr::str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot()

Customize the Theme

theme_set(theme_minimal(
  base_size = 14,
  base_family = "Roboto Condensed"
))

theme_update(
  panel.grid.major.x = element_blank(),
  panel.grid.minor = element_blank()
)

ggplot(
    bikes,
    aes(x = stringr::str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot()

Add Meaningful Labels

ggplot(
    bikes,
    aes(x = stringr::str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot() +
  labs(x = NULL, y = "Reported bike shares")

Add Meaningful Labels

ggplot(
    bikes,
    aes(x = stringr::str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot() +
  labs(
    x = NULL, y = NULL,
    title = "Reported bike shares by weather type"
  )

Add Meaningful Labels

theme_update(
  plot.title.position = "plot",
  axis.title = element_blank()
)

ggplot(
    bikes,
    aes(x = stringr::str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot() +
  ggtitle("Reported bike shares by weather type")

Jitter Strips of Counts per Weather Type

ggplot(
    bikes,
    aes(x = stringr::str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_jitter(
    alpha = .2
  ) +
  ggtitle("Reported bike shares by weather type")

Jitter Strips of Counts per Weather Type

ggplot(
    bikes,
    aes(x = str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_point(
    position = "jitter",
    alpha = .2
  )

ggplot(
    bikes,
    aes(x = str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_point(
    position = position_jitter(),
    alpha = .2
  )

Jitter Strips of Counts vs. Weather Type

ggplot(
    bikes,
    aes(x = str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_point(
    position = position_jitter(
      seed = 2022,
      width = .2,
      height = 0
    ),
    alpha = .2
  ) +
  ggtitle("Reported bike shares by weather type")

Boxplot + Jitter Strip Hybrid

ggplot(
    bikes,
    aes(x = str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot() +
  geom_point(
    position = position_jitter(
      seed = 2022,
      width = .2,
      height = 0
    ),
    alpha = .2
  ) +
  ggtitle("Reported bike shares by weather type")

Boxplot + Jitter Strip Hybrid

ggplot(
    bikes,
    aes(x = str_wrap(weather_type, 6),
        y = count)
  ) +
  geom_boxplot(
     outlier.shape = NA
     # outlier.color = "transparent"
     # outlier.alpha = 0
  ) +
  geom_point(
    position = position_jitter(
      seed = 2022,
      width = .2,
      height = 0
    ),
    alpha = .2
  ) +
  ggtitle("Reported bike shares by weather type")

Bonus: Sort Weather Types

ggplot(
    bikes,
    aes(
      x = forcats::fct_reorder(
        str_wrap(weather_type, 6), -count
      ),
      y = count)
  ) +
  geom_boxplot(
     outlier.shape = NA
     # outlier.color = "transparent"
     # outlier.alpha = 0
  ) +
  geom_point(
    position = position_jitter(
      seed = 2022,
      width = .2,
      height = 0
    ),
    alpha = .2
  ) +
  ggtitle("Reported bike shares by weather type")

Save the Plot

ggsave(here::here("exercises", "plots", "02_concepts_pt1_ex2.pdf"),
       width = 5, height = 6.5, device = cairo_pdf)
The final plot with an aspect ratio of 5 x 7 inches.

Alternative Chart Types
to Visualize Distributions

Let’s Update our Data Set

bikes <-
  bikes %>%
  mutate(
    weather_type_fct = stringr::str_wrap(
      weather_type, 6
    ),
    weather_type_fct = forcats::fct_reorder(
      weather_type_fct, -count
    )
  )

levels(bikes$weather_type_fct)
[1] "scattered\nclouds" "broken\nclouds"    "clear"            
[4] "cloudy"            "rain"              "snowfall"

Let’s Store Our ggplot Setup

g <-
  ggplot(
    bikes,
    aes(x = weather_type_fct,
        y = count)
  ) +
  ggtitle(
    "Reported bike shares by weather type"
  )

Beeswarm Plots with {ggbeeswarm}

g +
  geom_boxplot(
     outlier.shape = NA
  ) +
  ggbeeswarm::geom_beeswarm(
    size = .3,
    alpha = .2,
    cex = .6
  )

Beeswarm Plots with {ggbeeswarm}

g +
  geom_boxplot(
     outlier.shape = NA
  ) +
  ggbeeswarm::geom_quasirandom(
    size = .3,
    alpha = .2,
    width = .3,
    varwidth = TRUE
  )

Sina Plots with {ggforce}

g +
  geom_boxplot(
     outlier.shape = NA
  ) +
  ggforce::geom_sina(
    size = .5,
    alpha = .2,
    maxwidth = 1.2
  )

Barcode Strips

g +
  geom_boxplot(
     position = position_nudge(x = .15),
     width = .35
  ) +
  geom_point(
    shape = "-",
    size = 8,
    alpha = .1,
    position = position_nudge(x = -.15)
  )

Violin Plots

g +
  geom_violin()

Violin Plots

g +
  geom_violin(
    scale = "count",
    draw_quantiles = c(.5),
    fill = "grey80"
  )

Violin Plots

g +
  geom_violin(
    scale = "width",
    draw_quantiles = c(.5),
    trim = FALSE,
    bw = 250,
    fill = "grey80"
  )

Violin Plots with {ggdist}

g +
  ggdist::stat_eye()

Violin Plots with {ggdist}

g +
  ggdist::stat_eye(
    .width = c(0.5, 0.95),
    width = 1.5,
    adjust = .33
  )

Violin Plots with {ggdist}

g +
  ggdist::stat_eye(
    aes(thickness = stat(f*n)),
    .width = c(0.5, 0.95),
    adjust = .33
  )

Violin Plots with {ggdist}

g +
  ggdist::stat_eye(
    aes(thickness = stat(f*n)),
    .width = c(0.5, 0.95),
    adjust = .5,
    interval_size_range = c(.4, 2.5),
    point_size = 1.2,
    point_color = "white"
  )

Raincloud Plots with {ggdist}

g +
  ggdist::stat_halfeye(
    aes(thickness = stat(f*n)),
    .width = 0,
    width = .5,
    position = position_nudge(x = .2)
  ) +
  geom_jitter(
    width = .1,
    size = .5,
    alpha = .1
  )

Raincloud Plots with {ggdist}

g +
  ggdist::stat_halfeye(
    aes(thickness = stat(f*n)),
    color = NA,
    width = .5,
    position = position_nudge(x = .2)
  ) +
  geom_boxplot(
    width = .3,
    outlier.color = NA
  ) +
  geom_jitter(
    width = .1,
    size = .5,
    alpha = .1
  )

Raincloud Plots with {gghalves}

g +
  ggdist::stat_halfeye(
    aes(thickness = stat(f*n)),
    color = NA,
    width = .5,
    position = position_nudge(x = .1)
  ) +
  geom_boxplot(
    width = .1,
    outlier.size = .2
  ) +
  gghalves::geom_half_point(
    side = "l",
    range_scale = .4,
    size = .3,
    alpha = .1
  )

Raincloud Plots with {gghalves}

g +
  ggdist::stat_halfeye(
    aes(thickness = stat(f*n)),
    color = NA,
    width = .5,
    position = position_nudge(x = .1)
  ) +
  geom_boxplot(
    width = .1,
    outlier.size = .2
  ) +
  gghalves::geom_half_point(
    side = "l",
    range_scale = .4,
    size = .3,
    alpha = .1
  ) +
  coord_flip()

Cédric Scherer // rstudio::conf // July 2022