Tidytuesday: Mario Kart World Records. How to build a scatter plot in ggplot2
Tidy Tuesday
This tutorial is inspired by the great article The Evolution of a ggplot (Ep. 1) of Cédric Scherer.
Tidy Tuesday is a weekly data visualization challenge for the R community. I will base this tutorial on Mario Kart World Records challenge. I want to analyze if records in the game grow over time as in 100 metres running. A general idea of the chart: scatter plot with the year of a race and the time in seconds.
Data preparation
First, I will add needed libraries, download the data and extract a year from the date.
library(dplyr)
library(lubridate)
library(ggplot2)
library(ggtext) # Markdown format for text in ggplot2
library(showtext) # Fonts
font_add_google('Merriweather', 'Merriweather')
font_add_google('Montserrat', 'Montserrat')
showtext_auto()
tuesdata <- tidytuesdayR::tt_load('2021-05-25')
records <- tuesdata$records
records$date_year <- year(records$date)
Building a plot
Now let’s build a basic scatter plot with color parameter ‘type’ meaning single or three lap record. I chose this chart because I wanted to see destribution of records each year.
records %>%
ggplot(aes(x = date_year, y = time, col = type)) +
geom_point(size = 1.5, alpha = 0.5)
On this stage I also do simple cleaning like adjusting axis names and title or making Y axis start from zero.
records %>%
ggplot(aes(x = date_year, y = time, col = type)) +
geom_point(size = 1.5, alpha = 0.5) +
scale_y_continuous(limits = c(0, 400), expand = c(0.005, 0.005)) + # Expansion so Y axis starts from 0
labs(
title = 'Yearly Time records',
x = '',
y = ''
)
We see from the chart that most of the records for both laps are below 150 seconds and don’t change much over time. Although I want to be sure, so let’s add bigger dots for each year median.
records %>%
ggplot(aes(x = date_year, y = time, col = type)) +
geom_point(size = 1.5, alpha = 0.5) +
geom_point(
data = median_time,
aes(x = date_year, y = time),
size = 7,
alpha = 0.3
) +
scale_y_continuous(limits = c(0, 400), expand = c(0.005, 0.005)) +
labs(
title = 'Yearly Time records',
x = '',
y = ''
)
Indeed, median records don’t change much except for years 2008-2010 that can be because of low number of drivers. I will add annotation and arrow to show it explicitly.
records %>%
ggplot(aes(x = date_year, y = time, col = type)) +
geom_point(size = 1.5, alpha = 0.5) +
geom_point(
data = median_time,
aes(x = date_year, y = time),
size = 7,
alpha = 0.3
) +
scale_y_continuous(limits = c(0, 400), expand = c(0.005, 0.005)) +
labs(
title = 'Yearly Time records',
x = '',
y = ''
) +
annotate(
'text',
x = 2008,
y = 225,
size = 4,
label = 'Lowest\nnumber of players'
) +
geom_curve(
x = 2008, y = 195, xend = 2009.5, yend = 120,
arrow = arrow(length = unit(0.07, 'inch')),
size = 0.4,
curvature = 0
)
The analytics part is ready!
Legend & Colors
I chose different colors for a legend and plot lines, I use Color Palettes service for this.
legend_colors <- c(`Single Lap` = '#bf3f27', `Three Lap` = '#72a2ac')
line_color <- 'gray50'
text_color <- 'gray30'
title_color <- 'gray20'
I decided to remove legend and include it in the title with proper color instead. To color the text I used ggtext library and element_markdown() in theme() settings. I also adjusted annotation and line colors.
records %>%
ggplot(aes(x = date_year, y = time, col = type)) +
geom_point(size = 1.5, alpha = 0.5) +
geom_point(
data = median_time,
aes(x = date_year, y = time),
size = 7,
alpha = 0.3
) +
annotate(
'text',
x = 2008,
y = 225,
size = 4,
label = 'Lowest\nnumber of players',
color = title_color
) +
geom_curve(
x = 2008, y = 195, xend = 2009.5, yend = 120,
arrow = arrow(length = unit(0.07, 'inch')),
size = 0.4,
curvature = 0,
color = line_color
) +
scale_y_continuous(limits = c(0, 400), expand = c(0.005, 0.005)) +
labs(
title = glue::glue("Yearly Time records:
<span style='color:{legend_colors['Single Lap']};'>Single Lap</span>
and <span style='color:{legend_colors['Three Lap']};'>Three Lap</span>"),
x = '',
y = ''
) +
scale_color_manual(values = legend_colors) +
guides(col = FALSE)
As you can see, the title doesn’t look like what we expect because we didn’t adjust theme() yet. We will go back to it later.
Fonts
I will use two fonts: Merriweather for title and annotation and Montserrat for general text. I decided to use serif and sans-serif families for contrast. Both are Google fonts, install them locally and add those lines to your code.
font_add_google('Merriweather', 'Merriweather')
font_add_google('Montserrat', 'Montserrat')
showtext_auto()
Theme
ggplot2 has theme() component which allows to format your plot, i.e. clean it from extra grids, lines and ticks, change fonts and colors. Let’s add final touch to our plot and that’s it!
records %>%
ggplot(aes(x = date_year, y = time, col = type)) +
geom_point(size = 1.5, alpha = 0.5) +
geom_point(
data = median_time,
aes(x = date_year, y = time),
size = 7,
alpha = 0.3
) +
annotate(
'text',
x = 2008,
y = 225,
size = 4,
label = 'Lowest\nnumber of players',
color = title_color
) +
geom_curve(
x = 2008, y = 195, xend = 2009.5, yend = 120,
arrow = arrow(length = unit(0.07, 'inch')),
size = 0.4,
curvature = 0,
color = line_color
) +
scale_y_continuous(limits = c(0, 400), expand = c(0.005, 0.005)) +
labs(
title = glue::glue("Yearly Time records:
<span style='color:{legend_colors['Single Lap']};'>Single Lap</span>
and <span style='color:{legend_colors['Three Lap']};'>Three Lap</span>"),
x = '',
y = ''
) +
scale_color_manual(values = legend_colors) +
guides(col = FALSE) +
theme(
panel.background = element_blank(),
axis.ticks = element_blank(),
panel.grid = element_blank(),
axis.line.x.bottom = element_line(color = line_color),
axis.line.y.left = element_line(color = line_color),
plot.title = element_markdown(color = title_color, size = 14, family = 'Merriweather'),
text = element_text(color = text_color, size = 11, family = 'Montserrat')
)
Thanks for reading! Find the full code here :)