Sherlock Holmes

This weeks TidyTuesday data explored the complete line-by-line texts of the Sherlock Holmes stories and novels. The dataset includes the full version of texts, organized by book and line number.

As this is my first TidyTuesday, I wanted to try something simple. Therefore I evaluated the most common words across all texts and plotted them in a word cloud.

To make things a little more interesting, I shaped the word cloud in the form of Sherlock’s head, which I took from a book cover of The hound of Baskerville.

#!/bin/Rscript

#######################
##     LIBRARIES     ##
#######################

# data processing
library(tidyverse)
library(tidytext)

# word cloud
library(ggwordcloud)
library(ggtext)
library(svgtools)

# output
library(patchwork)

##################
##     DATA     ##
##################

holmes <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/main/data/2025/2025-11-18/holmes.csv')

tokens <- holmes %>%
    unnest_tokens(word, text) %>%
    filter(!is.na(word))

tokens_clean <- tokens %>%
    anti_join(stop_words, by="word")

word_count <- tokens_clean %>%
    count(word, sort=TRUE) %>%
    slice(1:150) %>%
    mutate(angle = 90 * sample(c(0, 1), n(), replace = TRUE, prob = c(60, 40)))

####################
##     COLORS     ##
####################

fg_0 <- "bisque1"
fg_1 <- "bisque4"

bg <- "darkslategray"

#########################
##     PLOT PARAMS     ##
#########################

sherlock.png <- "assets/sherlock.png"
sherlock.svg <- "assets/sherlock.svg"

mask <- png::readPNG(sherlock.png)
image <- svgparser::read_svg(sherlock.svg, xoffset = 0, yoffset = 0)

plot_theme <- theme_void() +
    theme(
        plot.margin=unit(c(0,0,0,0), 'cm'),
        panel.background = element_rect(fill='transparent'),
        plot.background = element_rect(fill='transparent', color=NA)
    )

pdf_theme <- theme_void() +
    theme(
        plot.margin=unit(c(1,1,1,1), 'cm'),
        panel.background = element_rect(fill = bg, color = bg),
        plot.background  = element_rect(fill = bg, color = bg)
    )

###################
##     PLOTS     ##
###################

set.seed(123)

# header and footer
df <- data.frame(x = 1:10, y = 1:10)
header <- ggplot(df, aes(x = x, y = y)) +
  geom_blank() +
  labs(
        title = "Sherlock Holmes",
        subtitle = "Most common words across all books"
        ) +
  plot_theme +
  theme(
    plot.title    = element_text(color = fg_0, size = 69, face = "bold"),
    plot.subtitle = element_text(color = fg_1, size = 42, face = "bold")
  )

# Word Cloud
cloud <- ggplot(
        word_count,
        aes(label = word, size = n,
            color = n, angle = angle
        )
    ) +
    geom_text_wordcloud_area(
        mask = mask,
        rm_outside = TRUE
    ) +
    scale_size_area(max_size = 69, trans = power_trans(1/.7)) +
    scale_color_gradient(low = fg_1, high = fg_0) +
    labs(caption = "#TidyTuesday") +
    plot_theme +
    theme(
        plot.caption  = element_markdown(color = fg_1, size = 25, face = "bold", lineheight = 1.2)
    )

# Background image
bg_img <- ggplot(df, aes(x = x, y = y)) +
  geom_blank() +
  annotation_custom(image, xmin = -Inf, xmax = Inf, ymin = -Inf, ymax = Inf) +
  plot_theme

###########################
##     OUTPUT PARAMS     ##
###########################

# text_height = 1
text_height = 1
th <- text_height

# bg_x = 0.12
bg_x = 0.11
bg_y = 0.005

layout <- c(
  area(t = 1,      l = 1, b = 2,      r = 6),
  area(t = 1 + th, l = 1, b = 6 + th, r = 6),
  area(t = 1 + th, l = 1, b = 6 + th, r = 6)
)

####################
##     OUTPUT     ##
####################

word_count

pdf("sherlock-cloud.pdf", height = 12 + ( 2 * th ), width = 12)

header /
    ( cloud + inset_element(bg_img,
                            left = bg_x, bottom = bg_y, right = 1 + bg_x, top = 1 + bg_y,
                            on_top = FALSE)
    ) /
    plot_layout(design = layout, widths = 6, heights = c(th, 6, 6)) +
    plot_annotation(theme = pdf_theme)

dev.off()