R script to produce the plot shown here: https://twitter.com/matteodefelice/status/1224426544803827713

eurostat_cars.R

library(tidyverse)
library(eurostat)
library(lubridate)

n_veih <- get_eurostat("tran_r_vehst") %>% # https://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=tran_r_vehst&lang=en
  dplyr::filter(
    str_length(geo) == 2,
    unit == "P_THAB", # NR: Number, P_THAB: Per thousand inhabitants
    vehicle == "CAR" # Passenger cars
  )

accid <- get_eurostat("tran_r_acci") %>% # https://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=tran_r_acci&lang=en
  dplyr::filter(
    str_length(geo) == 2,
    unit == "P_MHAB", # NR: Number, P_THAB: Per million inhabitants
    victim == "KIL" # KIL: killed, INJ: injured
  )
pop <- get_eurostat("tgs00096") # https://ec.europa.eu/eurostat/databrowser/view/tgs00096/default/table?lang=en

###
toplot <- inner_join(
  n_veih %>% select(geo, time, vehicles = values),
  accid %>% select(geo, time, kills = values),
  by = c("geo", "time")
) %>%
  dplyr::filter(
    geo %in% c("ES", "IT", "DE", "FR", "UK", "BE")
  ) %>%
  group_by(geo) %>%
  arrange(time) %>%
  mutate(
    kills_smooth = zoo::rollmean(kills, k = 5, align = "right", fill = NA)
  ) %>%
  dplyr::filter(
    !is.na(kills_smooth)
  ) %>%
  ungroup() %>%
  group_by(geo) %>%
  arrange(time) %>%
  ungroup()

g <- ggplot(toplot, aes(x = vehicles, y = kills_smooth, color = geo, group = geo)) +
  geom_point(aes(size = year(time)), alpha = 0.2) +
  geom_point(
    data = toplot %>%
      dplyr::filter(year(time) == 2017),
    aes(size = year(time)), alpha = 0.7
  ) +
  geom_path() +
  theme_light() +
  scale_color_brewer(palette = "Set2", name = "country") +
  scale_size_continuous(breaks = c(1994, 2017), name = "year") +
  labs(
    x = "Number of cars (per 1000 inhabitants)",
    y = "Kills due to car accidents (per million inhabitants)",
    title = "Cars in Europe: density and death rates",
    subtitle = "Time evolution for six European countries (from 1994 to 2017)",
    caption = "Source EUROSTAT data (datasets `tran_r_acci` and `tran_r_vehst`)"
  ) +
  theme(
    text = element_text(size = 12),
    plot.title = element_text(size = 16, face = "bold")
  )