Joy-Bird or Bird-Division?

So, if yo’re at all sensible you’ll be fed-up already with statistics from my Birdnet-Pi, but roll with me - this one is more like artwork.

Inspired by listening to Joy Division earlier I thought a ridgeline plot might be the best way to show some of the time of day stuff. Here’s a comparison of when in the day all the different birds have been detected, coloured by the average confidence of the identification of the bird. (not so useful that one but it makes it pretty)…

# Load packages
library(DBI)
library(RSQLite)
library(dplyr)
library(ggplot2)
library(ggridges)
library(forcats)
library(lubridate)

# Connect to the database
con <- dbConnect(RSQLite::SQLite(), "birds.db")

# Read detections
detections <- dbGetQuery(con, "SELECT * FROM detections")

# Prepare data
detections <- detections %>%
  mutate(
    DateTime = as.POSIXct(paste(Date, Time), format = "%Y-%m-%d %H:%M:%S"),
    Hour = hour(DateTime) + minute(DateTime) / 60
  )

# (Optional) Top 100 species
top_species <- detections %>%
  count(Com_Name, sort = TRUE) %>%
  slice_head(n = 100) %>%
  pull(Com_Name)

detections_filtered <- detections %>%
  filter(Com_Name %in% top_species)

# 🛠️ Calculate overall average confidence per species
species_confidence <- detections_filtered %>%
  group_by(Com_Name) %>%
  summarise(AvgConfidence = mean(Confidence, na.rm = TRUE), .groups = "drop")

# 🛠️ Calculate peak hour for each species (only where n >= 2)
species_peak <- detections_filtered %>%
  group_by(Com_Name) %>%
  filter(n() >= 2) %>%
  summarise(
    PeakHour = {
      dens <- density(Hour, from = 0, to = 24)
      dens$x[which.max(dens$y)]
    },
    .groups = "drop"
  )

# Join confidence and peak time into main data
detections_final <- detections_filtered %>%
  left_join(species_confidence, by = "Com_Name") %>%
  left_join(species_peak, by = "Com_Name") %>%
  filter(!is.na(PeakHour)) %>%
  mutate(Com_Name = fct_reorder(Com_Name, PeakHour, .fun = min))

# --- RIDGELINE PLOT with Fill and Labels ---
ggplot(detections_final, aes(x = Hour, y = fct_rev(Com_Name), fill = AvgConfidence)) +
  geom_density_ridges(
    scale = 3,
    rel_min_height = 0.1,
    color = "white",
    size = 0.3,
    alpha = 1
  ) +
  scale_fill_viridis_c(name = "Avg % Confidence", option = "plasma") +
  theme_minimal(base_family = "sans") +
  labs(
    title = "Bird Detection Patterns by Time of Day",
    subtitle = "Ridges sorted by earliest peak; Fill = Avg Confidence",
    x = "Hour of Day",
    y = "Species"
  ) +
  theme(
    plot.background = element_rect(fill = "white", color = NA),
    panel.background = element_rect(fill = "white", color = NA),
    axis.text.y = element_text(size = 3, color = "black"),
    axis.text.x = element_text(color = "black"),
    axis.title = element_text(color = "black"),
    plot.title = element_text(size = 12, face = "bold", color = "black"),
    plot.subtitle = element_text(size = 8, color = "black"),
    legend.title = element_text(color = "black"),
    legend.text = element_text(color = "black")
  )

To those of us of a certain age or musical persuassion there’s more than gentle echos of the cover of Joy Division’s Unknown Pleasures (1979). That image, one of the most iconic album artworks ever, is much more minimalist, stark and mysterious.

Interestingly the cover is a plot of successive radio pulses from the first pulsar discovered, CP 1919, superimposed vertically from Radio Observations of the Pulse Profiles and Dispersion Measures of Twelve Pulsars (Craft, 1970).

I’m not going to go into much more here as Scientific American has already done such a good job.

Thanks to Scientific American, there is a complete explanation of the dataset and its origin and we can plot it here to recreate the album:

# Load necessary libraries
library(ggplot2)
library(ggridges)
library(dplyr)

# Read the uploaded CSV
pulsar <- read.csv("joydivision.csv")

library(ggplot2)
library(ggridges)

col1 <- "black"
col2 <- "white"

ggplot(pulsar, aes(x = x, y = y, height = z, group = y)) +
  geom_ridgeline(
    min_height = min(pulsar$z),
    scale = 0.2,
    linewidth = 0.5,
    fill = col1,
    colour = col2
  ) +
  scale_y_reverse() +
  theme_void() +
  theme(
    panel.background = element_rect(fill = col1),
    plot.background = element_rect(fill = col1, color = col1),
  )

So having recreated that one - lets see what we can do with my birdnet data…

# --- (Optional) Top 20 species ---
top_species <- detections %>%
  count(Com_Name, sort = TRUE) %>%
  slice_head(n = 70) %>%    # now only 20 species
  pull(Com_Name)

detections_filtered <- detections %>%
  filter(Com_Name %in% top_species)

# 🛠️ Calculate overall average confidence per species
species_confidence <- detections_filtered %>%
  group_by(Com_Name) %>%
  summarise(AvgConfidence = mean(Confidence, na.rm = TRUE), .groups = "drop")

# Join back into main data
detections_final <- detections_filtered %>%
  left_join(species_confidence, by = "Com_Name")

# --- JOY DIVISION STYLE RIDGELINE PLOT ---
ggplot(detections_final, aes(x = Hour, y = fct_rev(factor(Com_Name)), group = Com_Name)) +
  geom_density_ridges(
    fill = "black",
    color = "white",
    scale = 5,
    size = 0.1,
    rel_min_height = 0
  ) +
  theme_void() +
  theme(
    plot.background = element_rect(fill = "black", color = NA),
    panel.background = element_rect(fill = "black", color = NA),
    axis.text = element_blank(),
    axis.title = element_blank(),
    plot.margin = margin(20, 20, 20, 20)
  )

Not far off!