Tidy Tuesday: 420 Data

Published

April 21, 2025

Week 16 of Tidy Tuesday looks at data on fatal car accidents in the USA and whether more deaths occur on 20th of April (the 420 Cannabis holiday). I made a line plot showing the change in number of fatalities over time, and decided to emphasise the seasonality by adding a colour gradient to the trend line.

Line plot showing the seasonal change in car accident fatalities in the USA from 1992 to 2017. There is a clear trend of more fatalities in the warmer months. Over the years, there was a slight reduction in fatalities from around 2007 to 2011, followed by a gradual increase back to near-1995 levels.

I discovered that geom_smooth() does not support gradients, but you can get around this by manually creating the loess (locally estimated scatterplot smoothing) using l <- loess(fatalities_count ~ as.numeric(date), data = daily_accidents, span = 0.02). This can be added as a variable in the data frame using mutate(smooth = predict(l)) and then plotted as a normal geom_line() in ggplot2.

To change the colour of the line, I specified geom_line(aes(y = smooth, color = months, group = 1)). Having group = 1 tells ggplot2 that I only want one line with multiple colours, rather than multiple lines each with their own colour. Then, scale_color_gradientn() is used to add the colours.

The full code is below.

Code 
tuesdata <- tidytuesdayR::tt_load(2025, week = 16)
daily_accidents <- tuesdata$daily_accidents
daily_accidents_420 <- tuesdata$daily_accidents_420

library(ggplot2)
library(dplyr)
library(lubridate)

l <- loess(fatalities_count ~ as.numeric(date), data = daily_accidents, span = 0.02)

accidents2 <- daily_accidents %>%
  mutate(smooth = predict(l),
         months = lubridate::month(date)) %>%
  arrange(date)

t <- "Car Accident Fatalities in the USA: 1992–2017"
st <- glue::glue("The grey line shows the daily number of fatalities, with the most deaths occuring on 3 November 2007, and the fewest on 2 February 2011.
                 The trend is displayed as a gradient-colored line showing the seasonality of fatal car crashes.")
cptn <- "Created by: jessimoore.bsky.social    Source: 420 Data, Harper & Palayew (2018)"

p2 <- ggplot(accidents2, 
             aes(x = date)) +
  geom_line(aes(y = fatalities_count),
            alpha = 0.2, linetype = "dotted",
            color = "grey40") +
  geom_line(aes(y = smooth, 
                color = months,
                group = 1), 
            linewidth = 1.5) +
  scale_color_gradientn(colors = c("#90caf9","#f3722c","#ffc400","#90be6d", "#90caf9"),
                        values = scales::rescale(c(1,3.5,6,9,12)),
                        labels = c("January", "March", "June", "September", "December"),
                        transform = "reverse") +
  scale_x_date(date_breaks = "3 years", date_minor_breaks = "6 months",
               date_labels = "%Y") +
  labs(x = NULL, y = "Fatalities",
       color = NULL,
       title = t, subtitle = st, caption = cptn) +
  theme_minimal() +
  theme(panel.grid = element_line(linewidth = 0.15),
        panel.grid.major.x = element_line(linewidth = 0.2, 
                                          color = "grey60"),
        plot.margin = margin(30,30,30,30),
        plot.caption = element_text(size = 8, hjust = 0.5, vjust = -15),
        plot.caption.position = "plot")