R Bootcamp - Problem Set 5 (Extreme aRt)

Extreme aRt — Objective

Your objective is to create some new extreme art using R, something either ugly or sublime. What you cannot do is be boring; your goal is to stay on the fringe.

Your submission (a Quarto .qmd file) is due Tues Sept 2 by 5pm. If you submit an entry, you get full credit on the problem set. Entries will be anonymized and winners will selected by popular vote.

Favorite Rtists

Here are some of my favorite generative artists who use R. Be inspired!

• Danielle Navarro [Art] [Github]
• Ijeamaka Anyene [Github] and this study in particular.
• Claus Wilke [Art] [Github], a biologist at UT Austin who also wrote the book on data visualization (it’s excellent).
• Thomas Lin Pederesen [Art] [Github]. I have some of his pieces in my office.
• inconvergent [Art] [Github]. It’s lisp, not R. But it’s so good.

There are several resources for color palettes, an important component of any hideous or beautiful creation.

• The section in Data Viz for R on color is worth a read.
• The colors in e.g. scale_color_brewer come from Cynthia Brewer, a cartographer who makes visually informative maps.
• color-hex has collections of complementary color palettes.

There are also several R packages that might help you build Rtistic plots.

• gganimate provides tools to bring your plots to life.
• ggforce provides interesting geoms that build on ggplot2.
• patchwork provides layout functions for plots.

Informative, but boring.

This is an informative but relatively boring plot. NOT THE GOAL HERE.

library(tidyverse)
library(palmerpenguins)
library(cowplot)

penguins_clean <- drop_na(penguins)

ggplot(
  penguins_clean,
  aes(
    x = body_mass_g / 1000,
    y = bill_length_mm
  )
) +
  geom_point(
    aes(
      shape = sex,
      color = species
    )
  ) +
  facet_grid(~island) +
  theme_minimal_grid() +
  scale_color_brewer(palette = "Dark2") +
  labs(
    title = "Analysis of geographic isolation on penguin phenotypes",
    x = "Body mass (kg)",
    y = "Bill length (mm)"
  )

Ugly Plots

Yikes. We can thank Yunus Ozekin for this abomination.

library(tidyverse)
titanic_tbl <- as_tibble(Titanic)

ggplot(
  titanic_tbl,
  aes(
    x = Survived,
    y = n,
    color = Class,
    shape = Sex,
    size = 6
  )
) +
  geom_jitter() +
  scale_y_sqrt() +
  labs(
    x = "Not Dead?",
    y = "How many? (ppl)",
    title = "WhO dIEd In titaNic?",
    caption = "Some lived, some died."
  ) +
  scale_x_discrete(position = "top") +
  theme(
    axis.text.x = element_text(
      face = "bold.italic",
      color = "#993333",
      size = 18,
      angle = 180
    ),
    axis.text.y = element_text(
      face = "bold",
      color = "orange",
      size = 18,
      angle = 135
    ),
    plot.background = element_rect(fill = "darkblue"),
    plot.title = element_text(
      face = "italic",
      color = "green",
      size = 48,
      angle = 183
    ),
    plot.caption = element_text(color = "white", size = 22),
    axis.title.x = element_text(size = 22, color = "pink", angle = 12),
    axis.title.y = element_text(color = "yellow", angle = 273, size = 17),
    legend.background = element_rect(fill = "yellow"),
    legend.title = element_text(
      angle = 71,
      face = "bold",
      color = "purple",
      size = 12
    ),
    legend.key = element_rect(color = "green", fill = "orange"),
    legend.text = element_text(color = "red", size = 14),
    panel.background = element_rect(fill = "yellow"),
    panel.grid.major.y = element_line(
      color = "green",
      linetype = "dotdash",
      linewidth = 1.2
    ),
    panel.grid.major.x = element_line(
      color = "purple",
      linewidth = 3,
      linetype = "twodash"
    ),
    panel.grid.minor = element_line(
      color = "red",
      linewidth = 2,
      linetype = "dashed"
    ),
    legend.position = "bottom"
  )

Beautiful Plots

This is a piece from Ijeamaka Anyene’s ode to coord_polar() (link above). Reminds me of Miro.

library(tidyverse)

apply_pattern_theme <- function(bg_hex, caption_hex) {
  theme(
    plot.background = element_rect(fill = bg_hex),
    panel.background = element_rect(fill = bg_hex),
    panel.grid = element_blank(),
    plot.caption = element_text(
      family = "Open Sans",
      size = 6,
      color = caption_hex
    ),
    legend.position = "none",
    axis.title = element_blank(),
    axis.text = element_blank(),
    axis.ticks = element_blank()
  )
}

outline <- tibble(
  x = 1,
  xend = 7,
  y = 15,
  yend = 15
)
segment_line <- tibble(
  x = c(1, 7),
  xend = c(1, 7),
  y = c(0, 2),
  yend = 15
)
area <- tibble(
  x = c(3, 5, 6),
  y = c(5, 7.5, 2),
  type = LETTERS[1:3]
)
palette_values <- c("#2a2640", "#a64e46", "#f29544")
ggplot() +
  geom_col(
    data = area,
    aes(x = x, y = y, fill = type),
    alpha = 0.75,
    width = 4
  ) +
  geom_segment(
    data = outline,
    aes(
      x = x,
      y = y,
      xend = xend,
      yend = yend
    ),
    size = 0.5
  ) +
  geom_segment(
    data = segment_line,
    aes(
      x = x,
      xend = xend,
      y = y,
      yend = yend
    ),
    size = 0.5
  ) +
  geom_point(aes(x = 5, y = 0)) +
  scale_fill_manual(values = palette_values) +
  scale_y_continuous(limits = c(0, 15)) +
  scale_x_continuous(limits = c(1, 10)) +
  coord_polar() +
  labs(caption = "Ijeamaka Anyene | @ijeamaka_a") +
  apply_pattern_theme(
    bg_hex = "#ded5c9",
    caption_hex = "black"
  )

Here’s another more complex geometric creation, again using coord_polar(). This will take a few seconds to render.

# https://twitter.com/aschinchon/status/1095057262744387587
library(tidyverse)

xy <- seq(-2, 2, by = .005)
expand.grid(x = xy, y = xy) |>
  ggplot(
    aes(
      x = (cos(x)^2 + sin(y^2)),
      y = (sin(y)^3 - cos(x^2))
    )
  ) +
  geom_point(alpha = .01, shape = 20, size = 0) +
  theme_void() +
  coord_polar()