In wfmackey/absmaps: Download and Use Maps Data from the Australia Bureau of Statistics

library(knitr)
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

The absmaps package exists to make it easier to download, compress and convert ABS shapefile data to sf objects to be used in R.

Installation

You can install absmaps from github with:

# install.packages("devtools")
devtools::install_github("wfmackey/absmaps")

Additionally, the sf package is required to handle the sf objects:

install.packages("sf")
library(sf)

And we will use the tidyverse packages in this example:

library(tidyverse)

Maps loaded with this package

Available maps are listed below. These will be added to over time. If you would like to request a map to be added, let me know via an issue on this Github repo. (Or send me an email: [e-mail])

More details about ASGS structrues can be found on the ABS website.

ASGS Main Structures

• Statistical Area 1 2011: area = sa1, year = 2011
• Statistical Area 1 2016: area = sa1, year = 2016
• Statistical Area 2 2011: area = sa2, year = 2011
• Statistical Area 2 2016: area = sa2, year = 2016
• Statistical Area 3 2011: area = sa3, year = 2011
• Statistical Area 3 2016: area = sa3, year = 2016
• Statistical Area 4 2011: area = sa4, year = 2011
• Statistical Area 4 2016: area = sa4, year = 2016
• Greater Capital Cities 2011: area = gcc, year = 2011
• Greater Capital Cities 2016: area = gcc, year = 2016
• State 2011: area = state, year = 2011
• State 2016: area = state, year = 2016

Remoteness areas

• Remoteness Areas 2011: area = ra, year = 2011
• Remoteness Areas 2016: area = ra, year = 2016

Mesh blocks

• NSW mesh blocks 2011: area = mesh_nsw, year = 2011
• NSW mesh blocks 2016: area = mesh_nsw, year = 2016
• Victoria mesh blocks 2011: area = mesh_vic, year = 2011
• Victoria mesh blocks 2016: area = mesh_vic, year = 2016
• Queensland mesh blocks 2011: area = mesh_qld, year = 2011
• Queensland mesh blocks 2016: area = mesh_qld, year = 2016
• SA mesh blocks 2011: area = mesh_sa, year = 2011
• SA mesh blocks 2016: area = mesh_sa, year = 2016
• WA mesh blocks 2011: area = mesh_wa, year = 2011
• WA mesh blocks 2016: area = mesh_wa, year = 2016
• Tasmania mesh blocks 2011: area = mesh_tas, year = 2011
• Tasmania mesh blocks 2016: area = mesh_tas, year = 2016
• NT mesh blocks 2011: area = mesh_nt, year = 2011
• NT mesh blocks 2016: area = mesh_nt, year = 2016
• ACT mesh blocks 2011: area = mesh_act, year = 2011
• ACT mesh blocks 2016: area = mesh_act, year = 2016
• Other mesh blocks 2011: area = mesh_other, year = 2011
• Other mesh blocks 2016: area = mesh_other, year = 2016

ASGS Indigenous areas

• Indigenous locations 2011: area = indigenous_locations, year = 2011
• Indigenous locations 2016: area = indigenous_locations, year = 2016
• Indigenous area 2011: area = indigenous_area, year = 2011
• Indigenous area 2016: area = indigenous_area, year = 2016
• Indigenous regions 2011: area = indigenous_regions, year = 2011
• Indigenous regions 2016: area = indigenous_regions, year = 2016

ASGS Non-ABS Structures

• Commonwealth Electoral Divisions 2018: area = ced, year = 2018
• State Electoral Divisions 2018: area = sed, year = 2018
• Local Government Areas 2016: area = lga, year = 2016
• Local Government Areas 2018: area = lga, year = 2018

Just show me how to make a map with this package

Downloading, compressing and converting an ABS shapefile

The absmaps package uses the load_absmaps to download and load a particular geospatial object:

library(absmaps)

mapdata1 <- load_absmaps(area = "sa3", year = 2011)

glimpse(mapdata1)
#> Observations: 351
#> Variables: 12
#> $ sa3_code_2011   <chr> "10101", "10102", "10103", "10104", "10201", "10…
#> $ sa3_name_2011   <chr> "Goulburn - Yass", "Queanbeyan", "Snowy Mountain…
#> $ sa4_code_2011   <chr> "101", "101", "101", "101", "102", "102", "103",…
#> $ sa4_name_2011   <chr> "Capital Region", "Capital Region", "Capital Reg…
#> $ gcc_code_2011   <chr> "1RNSW", "1RNSW", "1RNSW", "1RNSW", "1GSYD", "1G…
#> $ gcc_name_2011   <chr> "Rest of NSW", "Rest of NSW", "Rest of NSW", "Re…
#> $ state_code_2011 <chr> "1", "1", "1", "1", "1", "1", "1", "1", "1", "1"…
#> $ state_name_2011 <chr> "New South Wales", "New South Wales", "New South…
#> $ albers_sqkm     <dbl> 21236.6140, 6511.1214, 14281.8301, 9864.9397, 98…
#> $ cent_lat        <dbl> 149.0763, 149.6013, 148.9416, 149.8063, 151.2182…
#> $ cent_long       <dbl> -34.55399, -35.44940, -36.43958, -36.49934, -33.…
#> $ geometry        <MULTIPOLYGON [°]> MULTIPOLYGON (((149.1198 -3..., MUL…

Or

mapdata2 <- load_absmaps(area = "sa2", year = 2016)

glimpse(mapdata2)
#> Observations: 2,310
#> Variables: 15
#> $ sa2_main_2016   <chr> "101021007", "101021008", "101021009", "10102101…
#> $ sa2_5dig_2016   <chr> "11007", "11008", "11009", "11010", "11011", "11…
#> $ sa2_name_2016   <chr> "Braidwood", "Karabar", "Queanbeyan", "Queanbeya…
#> $ sa3_code_2016   <chr> "10102", "10102", "10102", "10102", "10102", "10…
#> $ sa3_name_2016   <chr> "Queanbeyan", "Queanbeyan", "Queanbeyan", "Quean…
#> $ sa4_code_2016   <chr> "101", "101", "101", "101", "101", "101", "101",…
#> $ sa4_name_2016   <chr> "Capital Region", "Capital Region", "Capital Reg…
#> $ gcc_code_2016   <chr> "1RNSW", "1RNSW", "1RNSW", "1RNSW", "1RNSW", "1R…
#> $ gcc_name_2016   <chr> "Rest of NSW", "Rest of NSW", "Rest of NSW", "Re…
#> $ state_code_2016 <chr> "1", "1", "1", "1", "1", "1", "1", "1", "1", "1"…
#> $ state_name_2016 <chr> "New South Wales", "New South Wales", "New South…
#> $ areasqkm_2016   <dbl> 3418.3525, 6.9825, 4.7634, 13.0034, 3054.4099, 1…
#> $ cent_lat        <dbl> 149.7932, 149.2328, 149.2255, 149.2524, 149.3911…
#> $ cent_long       <dbl> -35.45508, -35.37590, -35.35103, -35.35520, -35.…
#> $ geometry        <MULTIPOLYGON [°]> MULTIPOLYGON (((149.7606 -3..., MUL…

The resulting sf object contains one observation per area (in the following examples, one observation per sa3). It stores the geometry information in the geometry variable, which is a nested list describing the area’s polygon. The object can be joined to a standard data.frame or tibble and can be used with dplyr functions.

Creating maps with your sf object

We do all this so we can create gorgeous maps. And with the sf object in hand, plotting a map via ggplot and geom_sf is simple.

map <-
sa32016 %>%
  filter(gcc_name_2016 == "Greater Melbourne") %>%   # let's just look Melbourne
  ggplot() +
  geom_sf(aes(geometry = geometry))  # use the geometry variable

map

include_graphics("VIGNETTE-1.png")

The data also include centorids of each area, and we can add these points to the map with the cent_lat and cent_long variables using geom_point.

map <-
sa32016 %>%
  filter(gcc_name_2016 == "Greater Melbourne") %>%   # let's just look Melbourne
  ggplot() +
  geom_sf(aes(geometry = geometry)) +   # use the geometry variable
  geom_point(aes(cent_lat, cent_long))  # use the centroid lat and longs

map

include_graphics("VIGNETTE-2.png")

Cool. But, sidenote, this all looks a bit ugly. We can pretty it up using ggplot tweaks. See the comments on each line for its objective. Also note that we’re filling the areas by their areasqkm size, another variable included in the sf object (we’ll replace this with more interesting data in the next section).

map <-
sa32016 %>%
  filter(gcc_name_2016 == "Greater Melbourne") %>%   # let's just look Melbourne
  ggplot() +
  geom_sf(aes(geometry = geometry,  # use the geometry variable
              fill = areasqkm_2016),     # fill by area size
          lwd = 0,                  # remove borders
          show.legend = FALSE) +    # remove legend
  geom_point(aes(cent_lat,
                 cent_long),        # use the centroid lat and longs
             colour = "white") +    # make the points white
  theme_void() +                    # clears other plot elements
  coord_sf(datum = NA)              # fixes a gridline bug in theme_void()

map

include_graphics("VIGNETTE-3.png")

Joining with other datasets

At some point, we’ll want to join our spatial data with data-of-interest. The variables in our mapping data—stating the numeric code and name of each area and parent area—will make this relatively easy.

For example: suppose we had a simple dataset of median income by SA3 over time.

# Read data in some data
income <- read_csv("data/median_income_sa3.csv")

#> Parsed with column specification:
#> cols(
#>   sa3_name_2016 = col_character(),
#>   year = col_character(),
#>   median_income = col_double()
#> )

This income data contains a variable sa3_name_2016, and we can use dplyr::left_join() to combine with our mapping data.

combined_data <- left_join(income, sa32016, by = "sa3_name_2016")

Now that we have a tidy dataset with 1) the income data we want to plot, and 2) the geometry of the areas, we can plot income by area:

map <-
combined_data %>%
  filter(gcc_name_2016 == "Greater Melbourne") %>%   # let's just look Melbourne
  ggplot() +
  geom_sf(aes(geometry = geometry,  # use the geometry variable
              fill = median_income),        # fill by unemployment rate
          lwd = 0) +                # remove borders
  theme_void() +                    # clears other plot elements
  coord_sf(datum = NA) +            # fixes a gridline bug in theme_void()
  labs(fill = "Median income")

map

include_graphics("VIGNETTE-4.png")

Why does this package exist?

The motivation for this package is that maps are cool and fun and are, sometimes, the best way to communicate data. And making maps is R with ggplot is relatively easy when you have the right object.

Getting the right object is not technically difficult, but requires research into the best-thing-to-do at each of the following steps:

• Find the ASGS ABS spatial-data page and determine the right file to download.
• Read the shapefile into R using one-of-many import tools.
• Convert the object into something usable.
• Clean up any inconsistencies and apply consistent variable naming/values across areas and years.
• Find an appropriate compression function and level to optimise output.

For me, at least, finding the correct information and developing the best set of steps was a little bit interesting but mostly tedious and annoying. The absmaps package holds this data for you, so you can spend more time making maps, and less time on Stack Overflow, the ABS website, and lovely-people’s wonderful blogs.

How does this package do the-things-it-does

The absmaps package uses four key functions to do-the-things-it-does:

• utils::download.file to download shapefile data from the ABS.
• sf::st_read to read the shapefile into an sf object.
• rmapshaper::ms_simplify to nicely compress the sf object.
• readr::write_rds to write, and readr::read_rds to read our nice sf objects.

There are a bunch of other fiddley-things on top of that, but those four steps are the 'workhorse' functions. A big thanks to people who built them.

If you're interested, check out load_absmaps.R code in the R/ file. And, of course, if you have a suggestion to improve, please do let me know via an issue at https://github.com/wfmackey/absmaps.

I want to complain about this package

Fair enough! The best avenue is via a Github issue at (wfmackey/absmaps). This is also the best place to request data that isn’t yet available in the package. You can also email me at [e-mail].

wfmackey/absmaps documentation built on Oct. 4, 2021, 8:23 p.m.