In edwindj/cbsodataR: Statistics Netherlands (CBS) Open Data API Client

library(dplyr)

Statistics Netherlands (CBS) is the office that produces all official statistics of the Netherlands.

Besides statistical data, SN also publishes cartographic maps, i.e. administrative areas such as "gemeenten" (municipalities), "buurten" (neighborhoods), "wijken" (districts) and many more. The main outlet for publishing these CBS specific geospatial data is via Publieke Diensten op de Kaart (PDOK), which offers a wide range of geospatial API's for accessing map material using Geographical Information System software.

A simple use case for using maps is thematic cartography: turning regional statistical data into thematic cartographic map. For this purpose the map with the boundaries of administrative regions, which typically has a precision of meters is turned into a picture in which a pixel is (several) 100 meters by 100 meters: the map has more precision than that is needed for a cartographic picture for the Netherlands. The website https://cartomap.github.io/nl/ therefore offers simplified, and thus smaller , maps of the CBS map material, that are useful for making plots of the Netherlands.

cbsodataR allows easy access to these maps using the following functions:

cbs_get_maps, returns a data.frame with all available simplified maps.
cbs_get_sf, returns a geospatial sf object that can be used for plotting.
cbs_join_sf_with_data, returns a geospatial sf object combined with the data downloaded with cbs_get_data.

R offers a range of specialized packages for plotting data on maps such as ggplot2, tmap, leaflet, mapview and others, and these are to be used when plotting maps. For this example we stick to ggplot2 to keep things simple and refer the user to the help of the other packages.

Available maps

cbsodataR gives an easy way to retrieve cartographic maps that can be used with data that is downloaded with cbsodataR.

To see a list of available maps use, the function cbs_get_maps()

library(cbsodataR)
cbs_maps <- cbs_get_maps()
# the layout of the data.frame is:
str(cbs_maps)

with:

$region: the type of region
$year: the year in which this region map is valid, regional classifications change yearly in NL...
$wgs84: link to geojson with wgs84 projection (epsg:4326).
$rd: link to geojson with rijksdriehoeksstelsel projection (epsg:28992).

At the moment of generating this document, the following regions are available:

cat("> ")
cbs_maps$region |> 
  unique() |> 
  paste0(collapse = ", ") |> 
  cat()

Map retrieval: `cbs_get_sf`

To download a map, without any data, the function cbs_get_sf can be used.

library(sf)
gemeente_2023 <- cbs_get_sf("gemeente", 2023)

gemeente_2023 is a map file with the boundaries of the Dutch municipalities of 2023. Note that is just contains the codes ($statcode) and names ($statnaam) of the municipalities. The names are useful for displaying purposes and the codes are useful for joining/connecting the map with data on municipalities in 2023.

str(gemeente_2023) # sf object
plot(gemeente_2023, max.plot = 1) # just plot the statcode kolom.

Creating a map with data: `cbs_join_sf_with_data`

cbsodataR contains a utility function cbs_join_sf_with_data that allows for creating a map object with the data you downloaded from Statistics Netherlands/CBS.

Suppose we have downloaded the following dataset:

youth_care_data <- cbs_get_data( "85098NED" # youth care table
                               , Perioden="2022JJ00" # only figures of 2022
                               , RegioS=has_substring("GM") # only "gemeente" figures
                               )

Since we have data on municipalities ("gemeente") in 2022, we can create a map object containing the downloaded data for the "gemeente"s in 2022

map_with_data <- 
  cbs_join_sf_with_data("gemeente", 2022, youth_care_data) |> 
  transform( natura = JongerenMetJeugdhulpInNatura_1 )

print(map_with_data)

And this map can be plotted e.g. with ggplot2

library(sf)
library(ggplot2)
map_with_data |> 
  ggplot() + 
  geom_sf(aes(fill=natura), color="#FFFFFF99") +
  scale_fill_viridis_c(direction = -1)+ 
  labs(fill="% youth with care in natura") + 
  theme_void()

Manual process

cbs_join_sf_with_data is handy, but if you want more control on the process, or want to combine your own map material, the following steps can be taken.

library(dplyr) # not necessary, but familiar for tidyverse code

# retrieve map
gemeente_2022 <- cbs_get_sf(region="gemeente", year="2022")

# add a statcode column to the dataset so it can be joined to the map.
youth_care_data <- youth_care_data |> 
  cbs_add_statcode_column() |> 
  mutate(natura = JongerenMetJeugdhulpInNatura_1)

# the map is left joined with the data (so the result is again an sf object)
gemeente_2022 |>
  left_join(youth_care_data, by="statcode") |> 
  ggplot() + 
  geom_sf(aes(fill=natura), color="#FFFFFF99") +
  scale_fill_viridis_c(direction = -1, option = "A")+ 
  labs(fill="% youth with care in natura") + 
  theme_void()