The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.

Researching vehicles involved in collisions with STATS19 data

library(stats19)
library(dplyr)

Vehicle level variables in the STATS19 datasets

Of the three dataset types in STATS19, the vehicle tables are perhaps the most revealing yet under-explored. They look like this:

v = get_stats19(year = 2022, type = "vehicle")
names(v)
v

We will categorise the vehicle types to simplify subsequent results:

v = v %>% mutate(vehicle_type2 = case_when(
  grepl(pattern = "motorcycle", vehicle_type, ignore.case = TRUE) ~ "Motorbike",
  grepl(pattern = "Car", vehicle_type, ignore.case = TRUE) ~ "Car",
  grepl(pattern = "Bus", vehicle_type, ignore.case = TRUE) ~ "Bus",
  grepl(pattern = "cycle", vehicle_type, ignore.case = TRUE) ~ "Cycle",
  # grepl(pattern = "Van", vehicle_type, ignore.case = TRUE) ~ "Van",
  grepl(pattern = "Goods", vehicle_type, ignore.case = TRUE) ~ "Goods",
  
  TRUE ~ "Other"
))
# barplot(table(v$vehicle_type2))

All of these variables are of potential interest to road safety researchers. Let’s take a look at summaries of a few of them:

table(v$vehicle_type2)
summary(v$age_of_driver)
summary(v$engine_capacity_cc)
table(v$propulsion_code)
summary(v$age_of_vehicle)

The output shows vehicle type (a wide range of vehicles are represented), age of driver (with young and elderly drivers often seen as more risky), engine capacity and populsion (related to vehicle type and size) and age of vehicle. In addition to these factors appearing in prior road safety research and debate, they are also things that policy makers can influence, e.g by:

Relationships between vehicle type and crash severity

To explore the relationship between vehicles and crash severity, we must first join on the ‘accidents’ table:

a = get_stats19(year = 2022, type = collision)
va = dplyr::inner_join(v, a)

Now we have additional variables available to us:

dim(v)
dim(va)
names(va)

Let’s see how crash severity relates to the variables of interest mentioned above:

xtabs(~vehicle_type2 + accident_severity, data = va) %>% prop.table()
xtabs(~vehicle_type2 + accident_severity, data = va) %>% prop.table() %>% plot()

As expected, crashes involving large vehicles such as buses and good vehicles tend to be more serious (involve proportionally more deaths) than crashes involving smaller vehicles.

To focus only on cars, we can filter the va table as follows:

vac = va %>% filter(vehicle_type2 == "Car")

The best proxy we have for car type in the open STATS19 data (there are non-open versions of the data with additional columns) is engine capacity, measured in cubic centimetres (cc). The distribution of engine cc’s in the cars dataset created above is shown below.

summary(vac$engine_capacity_cc)

The output shows that there are some impossible values in the data, likely due to recording error. Very few cars have an engine capacity above 5 litres (5000 cc) and we can be confident that none have an engine capacity below 300 cc. We’ll identify these records and remove them as follows:

We have set the anomolous vehicle size data to NA meaning it will not be used in the subsequent analysis.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.