Engaging and Beautiful Data Visualizations with ggplot2

Fundamentals & Workflows
— Exercise Solutions —

Cédric Scherer // posit::conf // September 2023

Preparation

library(readr)
library(dplyr)
library(ggplot2)

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

## plot style
theme_set(theme_minimal(base_family = "Asap SemiCondensed", base_size = 18))
theme_update(
  panel.grid.minor = element_blank()
)

Exercise 1

Discuss / investigate with your neighbor:
- What are the differences between geom_line() and geom_path()?
- Why can you use geom_smooth() and stat_smooth() interchangeably?
- What are the three ways to remove a legend from a ggplot?

geom_line() versus geom_path()

?geom_line

Description

geom_path() connects the observations in the order in which they appear in the data. geom_line() connects them in order of the variable on the x axis.

geom_line() versus geom_path()

bikes_filtered <- 
  bikes |> 
  filter(year == 2015, month == 2, day_night == "day")

geom_line() versus geom_path()

ggplot(bikes_filtered,
       aes(x = humidity, y = temp)) +
  geom_line() +
  geom_point(aes(color = date), size = 4)

ggplot(bikes_filtered,
       aes(x = humidity, y = temp)) +
  geom_path() +
  geom_point(aes(color = date), size = 4)

geom_smooth() and stat_smooth()

?geom_smooth

Description

geom_smooth() and stat_smooth() are effectively aliases: they both use the same arguments. Use stat_smooth() if you want to display the results with a non-standard geom.

layer(data, mapping, geom, stat, position)

layer()

ggplot() +
  layer(
    data = bikes, 
    mapping = aes(x = temp, y = count),
    geom = "point", 
    stat = "identity", 
    position = "identity"
  )

layer()

ggplot(
    data = bikes, 
    mapping = aes(x = temp, y = count),
  ) +
  layer(
    geom = "point", 
    stat = "identity", 
    position = "identity"
  )

geom_smooth() and stat_smooth()

geom_smooth(
  stat = "smooth",
  position = "identity"
)

stat_smooth(
  geom = "smooth",
  position = "identity"
)

geom_smooth() and stat_smooth()

ggplot(bikes,
       aes(x = temp, y = count)) +
  geom_point(alpha = .1) +
  geom_smooth()

ggplot(bikes,
       aes(x = temp, y = count)) +
  geom_point(alpha = .1) +
  stat_smooth()

“Non-Standard Geom”

ggplot(bikes, aes(x = temp, y = count)) +
  geom_point(alpha = .1) +
  stat_smooth(geom = "pointrange")

“Non-Standard Geom”

ggplot(bikes, aes(x = temp, y = count)) +
  geom_point(alpha = .1) +
  stat_smooth(geom = "col")

“Non-Standard Geom”

ggplot(bikes, aes(x = temp, y = count)) +
  geom_point(alpha = .1) +
  stat_smooth(geom = "text", aes(label = "🤓"), size = 4)

geom_* versus stat_*

ggplot(bikes, aes(x = weather_type, fill = year)) + 
  geom_bar() # with `stat = "count", position = "stack"`

geom_* versus stat_*

ggplot(bikes, aes(x = weather_type, fill = year)) + 
  stat_count() # with `geom = "bar", position = "stack"`

Remove Legends

ggplot(filter(bikes, year == "2016"), aes(x = date, y = count)) +
  geom_line(aes(color = day_night, linetype = day_night)) +
  geom_point(aes(fill = day_night), shape = 21)

Remove Legends: Layer

ggplot(filter(bikes, year == "2016"), aes(x = date, y = count)) +
  geom_line(aes(color = day_night, linetype = day_night), show.legend = FALSE) +
  geom_point(aes(fill = day_night), shape = 21)

Remove Legends: Aesthetic

ggplot(filter(bikes, year == "2016"), aes(x = date, y = count)) +
  geom_line(aes(color = day_night, linetype = day_night)) +
  geom_point(aes(fill = day_night), shape = 21) +
  scale_color_discrete(guide = "none")

Remove Legends: Aesthetic

ggplot(filter(bikes, year == "2016"), aes(x = date, y = count)) +
  geom_line(aes(color = day_night, linetype = day_night)) +
  geom_point(aes(fill = day_night), shape = 21) +
  guides(color = "none", linetype = guide_none())

Remove Legends: All

ggplot(filter(bikes, year == "2016"), aes(x = date, y = count)) +
  geom_line(aes(color = day_night, linetype = day_night)) +
  geom_point(aes(fill = day_night), shape = 21) +
  theme(legend.position = "none")

Exercise 2

Explore the TfL bike share data visually:
- Create a time series of counts per day and night.
- Draw box and whisker plots of average temperatures per month.
- Visualize bike counts per weather type and period as bar chart.
Combine the three plots with patchwork.
Export the final graphic in a format of your choice.

Time Series

ggplot(bikes, aes(x = date, y = count)) +
  geom_line(aes(color = day_night))

Time Series

g1 <- 
  ggplot(bikes, aes(x = date, y = count)) +
  geom_line(aes(color = day_night)) +
  scale_x_date(date_breaks = "4 months", date_labels = "%m/'%y") +
  scale_y_continuous(labels = scales::label_comma()) +
  scale_color_manual(values = c(day = "#FFA200", night = "#757BC7")) +
  labs(x = NULL, y = "Rented bikes", color = NULL)

g1

Time Series

Box and Whisker Plots

ggplot(bikes, aes(x = month, y = temp)) +
  geom_boxplot()

Box and Whisker Plots

ggplot(bikes, aes(x = forcats::fct_reorder(month, -temp), y = temp)) +
  geom_boxplot()

Box and Whisker Plots

g2 <- 
  ggplot(bikes, aes(x = lubridate::month(date, label = TRUE), y = temp)) +
  geom_boxplot() +
  scale_y_continuous(labels = scales::label_number(suffix = "°C")) +
  labs(x = NULL, y = "Temperature")

g2

Box and Whisker Plots

Bar Chart

ggplot(bikes, aes(x = weather_type, y = count, fill = day_night)) + 
  geom_col()

Bar Chart

ggplot(bikes, aes(x = weather_type, y = count, fill = day_night)) + 
  stat_summary(geom = "col")

Bar Chart

ggplot(bikes, aes(x = weather_type, y = count, fill = day_night)) + 
  stat_summary(geom = "col", fun = "sum", position = "stack")

Bar Chart

bikes |>
  group_by(weather_type, day_night) |>
  summarize(count = sum(count)) |>
  ggplot(aes(x = weather_type, y = count, fill = day_night)) + 
  geom_col()

Bar Chart

ggplot(bikes, aes(x = weather_type, y = count, fill = day_night)) + 
  stat_summary(geom = "col", fun = "sum", position = "dodge")

Bar Chart

ggplot(bikes, aes(x = weather_type, y = count, fill = day_night)) + 
  stat_summary(
    geom = "col", fun = "sum", 
    width = .7,
    position = position_dodge(
      width = .8, preserve = "single"
    )
  )

Bar Chart

ggplot(bikes, aes(x = weather_type, y = count, fill = day_night)) + 
  stat_summary(
    geom = "col", fun = "sum", 
    width = .7,
    position = position_dodge2(
      padding = .2, preserve = "single"
    )
  )

Bar Chart

g3 <- ggplot(bikes, aes(x = weather_type, y = count, fill = day_night)) + 
  stat_summary(
    geom = "col", fun = "sum", 
    width = .7,
    position = position_dodge2(
      padding = .2, preserve = "single"
    )
  ) +
  scale_y_continuous(labels = scales::label_comma(scale = 1/10^6, suffix = "M")) +
  scale_fill_manual(
    values = c(day = "#FFA200", night = "#757bc7")
  ) +
  labs(x = NULL, y = "Rented bikes", fill = NULL) +
  theme(
    panel.grid.major.x = element_blank()
  )

g3

Bar Chart

Combine Plots

library(patchwork)
g2 + (g1 / g3) + plot_layout(guides = "collect", widths = c(.7, 1))

Combine Plots

library(patchwork)
(g2 + g1) / g3 + plot_layout(guides = "collect")

Combine Plots

library(patchwork)
(g2 + g1 + plot_layout(widths = c(.7, 1))) / g3 + plot_layout(guides = "collect")

Save Multipanel Plot

ggsave("bikes-panel.png", width = 14, height = 9, dpi = 600, bg = "white")

The final PNG graphic.