Grundbegriffe

German authorities record road accident with number as key for factor value. For example, in many cases 0 means FALSE and 1 means TRUE. To understand the statistic, the Federal Statistical Office (Statistisches Bundesamt − Destatis) publishes a booklet explaining the basic concepts of the database about traffic accident. Today, I was curious to look at the variable accident class (Unfallart), describing the direction of movement of the vehicles involved. Accidents are categorised into 10 classes:

1. Collision with another vehicle that starts, stops or is in stationary traffic

2. Collision with another vehicle driving ahead or waiting

3. Collision with another vehicle driving sideways in the same direction.

4. Collision with another vehicle traveling in the opposite direction.

5. Collision with another vehicle turning or crossing.

6. Collision between vehicle and pedestrian.

7. Impact on an obstacle on the road.

8. Leaving the lane to the right.

9. Leaving the lane to the left.

10. Accidents of a different class.

In your opinion, what is the main class of accident involving bikes?

Picture of the day

A graph of the accidents involving a bike in Berlin and Brandenburg split by class.

library(sf)

# Select Berlin and Brandenburg
# ULAND = 11 (Berlin) | 12 (Brandenburg)
berbrand <- read.csv2(pipe("awk 'BEGIN {FS=\";\"} {if ($3 == 11 || $3 == 12 ) print $0}' Unfallorte2020_LinRef.csv"),
header=FALSE,
colClasses = c(rep("character", 2),
               rep("factor", 18),
               rep("character", 4),
               rep("factor", 1)
))
readLines("Unfallorte2020_LinRef.csv", n=1) |>
  strsplit(";") |> unlist()  |> tolower() -> colnames(berbrand)
colnames(berbrand)
levels(berbrand$uwochentag) <- c("So", "Mo", "Di", "Mi", "Do",
                                 "Fr", "Sa")
levels(berbrand$ulichtverh) <-
  c("Tageslicht","Dämmerung","Dunkelheit")
berbrand$uwochentag <-
  factor(berbrand$uwochentag,levels(berbrand$uwochentag)[c(2:7,1)])

# Select only when bike are involved
 berbrand <- berbrand[berbrand$istrad == 1,]

# Colour palette
retro <- c( "#9239F6", "#903495"  , "#6F3460"  , "#4A354F"  ,
"#D20076"  , "#FF0076"  , "#FF4373"  , "#FF6B58", "#F8B660",
"red")

library(ggplot2)
library(dplyr)

# Bike accident by class
jpeg("bike_unfaelle_class.jpg", width=800, quality=90)
ggplot(berbrand, aes(x = uart)) +
  geom_bar(aes(fill = uart)) +
    theme_minimal() +
    guides(fill="none") +
                      scale_fill_manual( values = retro ) +
 labs(x = "Type")
dev.off()

Number of bike accident by class. Exhibit made with R (code provided in a comment of this webpage) Data Copyright: DL-DE BY 2.0 Statistische Ämter des Bundes und der Länder, Deutschland, 2021.

The answer is unambiguous: 49.5 % of the accidents are of class 5 (i.e. when turning, and I suspect that this is mostly the case when cars are turning on the right and did not check if a bike was already there). Then we have the case [1]0 (almost 20%) and the case 1 (8%), but I am not sure what that concretely means and I will have to look closer tomorrow.