Can an exploding snowman predict the summer season?

scatterplot

science

Author

Ana Luisa Bodevan

Published

December 2, 2025

This week challenge dataset in on Can an exploding snowman predict the summer season?. Check the TidyTuesday GitHub repo for the data.

1. SETUP

Code

library(pacman)

pacman::p_load(
  tidytuesdayR,
  tidyverse,
  dplyr,
  janitor,
  ggtext,
  showtext,
  scales,
  glue,
  skimr,
  ggbranding
)

tuesdata <- tidytuesdayR::tt_load('2025-12-02')

df <- tuesdata$sechselaeuten |>
  janitor :: clean_names()

rm(tuesdata)

2. DATA ANALYSIS

Code

skim(df)

Data summary
Name	df
Number of rows	67
Number of columns	9
_______________________
Column type frequency:
logical	1
numeric	8
________________________
Group variables	None

Variable type: logical

skim_variable	n_missing	complete_rate	mean	count
record	0	1	0.16	FAL: 56, TRU: 11

Variable type: numeric

skim_variable	n_missing	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
year	0	1.00	1990.19	21.72	1923.00	1974.50	1991.00	2007.50	2025.00	▁▂▇▇▇
duration	2	0.97	17.64	11.87	4.00	10.00	13.00	23.50	60.00	▇▅▁▁▁
tre200m0	0	1.00	17.61	1.34	15.07	16.69	17.50	18.31	21.67	▃▇▃▂▁
tre200mn	0	1.00	8.38	1.25	6.10	7.38	8.43	9.21	11.40	▅▃▇▃▁
tre200mx	0	1.00	30.64	1.62	27.50	29.46	30.53	31.69	34.57	▂▇▇▂▂
sre000m0	0	1.00	213.12	29.14	143.70	195.40	210.42	229.90	284.66	▂▅▇▃▂
sremaxmv	0	1.00	49.75	6.84	33.33	45.16	49.33	53.50	67.00	▂▃▇▃▁
rre150m0	0	1.00	126.80	32.17	52.83	103.87	126.03	147.38	186.90	▁▆▇▆▅

Code

skimr::skim(df) |> summary()

Data summary
Name	df
Number of rows	67
Number of columns	9
_______________________
Column type frequency:
logical	1
numeric	8
________________________
Group variables	None

Code

cor(df$duration, df$tre200m0, use = "complete.obs")

[1] 0.194294

So, the correlation between Boeoegg explosion duration and average summer temperature is positive but extremely weak (r = 0.19), offering no real predictive value for summer heat.

3. DATA TIDYING

Code

df_plot <- df |>
  select(year, duration, tre200m0, record) |>
  mutate(
    record = if_else(record, "Record-hot summer", "Normal summer"),
    record = factor(record, levels = c("Normal summer", "Record-hot summer"))
  )

corr_val <- cor(df_plot$duration, df_plot$tre200m0, use = "complete.obs")
corr_lab <- glue("Pearson r = {round(corr_val, 2)}")

4. PLOT

Code

col <- c(
  "Normal summer" = "#4A4A4A",
  "Record-hot summer" = "#D8432D"
)

Code

font_add_google("Open Sans", "opensans")
showtext_auto()

title <- "Can the exploding Boeoegg of Zurich predict the summer season?"
subtitle <- glue("Folklore has that the faster the snowman effigy blast the warmer the summer but hard data \nshows a very weak correlation (r = 0.19), offering no predictive value")

Code

base_theme <- function(base_size = 14, base_family = "opensans") {
  theme_minimal(base_size = base_size, base_family = base_family) +
    theme(
      # Backgrounds
      panel.background = element_rect(fill = "grey98", color = NA),
      plot.background = element_rect(fill = "white", color = NA),

      # Gridlines (Nature uses strong horizontal gridlines)
      panel.grid.major = element_line(color = "grey85", linewidth = 0.6),
      panel.grid.minor = element_blank(),

      # Axes
      axis.title = element_text(size = base_size * 1.1, face = "bold"),
      axis.text = element_text(size = base_size * 0.9, color = "grey20"),

      # Titles
      plot.title = element_text(size = base_size * 1.4, face = "bold"),
      plot.subtitle = element_text(size = base_size * 1.05, color = "grey30"),

      # Legend
      legend.position = c(0.02, 0.98),
      legend.justification = c("left", "top"),
      legend.background = element_rect(fill = alpha("white", 0.8), color = NA),
      legend.title = element_blank(),

      # Margins
      plot.margin = margin(t = 15, r = 20, b = 15, l = 20)
    )
}

Code

ggplot(df_plot, aes(duration, tre200m0, color = record)) +
  geom_point(size = 3, alpha = 0.9) +

  geom_smooth(
    method = "lm",
    linewidth = 0.6,
    color = "gray20",
    se = FALSE,
    alpha = 0.5
  ) +

  geom_hline(
    yintercept = 19,
    linetype = "dashed",
    linewidth = 0.6,
    color = "grey40"
  ) +

  scale_color_manual(values = col) +

  labs(
    title = title,
    subtitle = subtitle,
    x = "Boeoegg explosion duration (seconds)",
    y = "Average temperature (°C)"
  ) +
  base_theme() +
  add_branding(
    github = "anabodevan",
    bluesky = "1141bode",
    additional_text = "MeteoSwiss & Statistik Stadt Zürich | #TidyTuesday 2025 W48",
    line_spacing = 1L,
    icon_color = "gray40",
    text_color = "gray40",
    icon_size = "12pt", # larger icons
    text_size = "12pt", # larger text,
    text_family = "opensans", # custom font
    caption_margin = ggplot2::margin(t = 20, b = 5, unit = "pt") # custom margin
  )