Skip to contents

Getting started with census2020download

Source: vignettes/census2020download.Rmd
census2020download.Rmd

What this package does

census2020download is a small set of helper functions used by the EJAM package. It downloads the 2020 Census redistricting (PL94-171) files for US states, DC, and Puerto Rico — plus the Demographic and Housing Characteristics (DHC) files for the Island Areas (VI, GU, MP, AS) — then unzips, reads, and reshapes them into compact data.tables.

The full pipeline has four steps, wrapped by the single function census2020_get_data():

  1. census2020_download() — fetch the zip files from the Census Bureau
  2. census2020_unzip() — unzip the data and geographic header files
  3. census2020_read() — read and merge the pipe-delimited files
  4. census2020_clean() — rename columns, compute total area, subset

The one-call pipeline

census2020_get_data() runs all four steps. Try it with a couple of small states so the download is quick:

blocks <- census2020_get_data(c("DE", "DC"))
dim(blocks)
head(blocks)

By default it returns the columns blockfips, lat, lon, pop, and area (area in square meters). To keep every available column — including the race/ethnicity breakdown — pass cols_to_keep = "all":

blocks_all <- census2020_get_data(c("DE", "DC"), cols_to_keep = "all")

# The race/ethnicity subgroups sum exactly to the total block population:
groups <- c("hisp", "nhwa", "nhba", "nhaiana", "nhaa",
            "nhnhpia", "nhotheralone", "nhmulti")
all.equal(blocks_all$pop, rowSums(blocks_all[, ..groups]))

Controlling downloads

  • folder — where zip files are saved (default: a temporary directory).
  • overwrite = FALSE — skip downloading any zip already present in folder.
  • timeout — seconds before an individual download times out (default 180).
blocks <- census2020_get_data(
  c("DE", "DC"),
  folder = "~/census2020zip",
  overwrite = FALSE,
  timeout = 300
)

Building the EJAM data tables

census2020_save_datasets() splits the cleaned blocks into the five tables EJAM uses, and computes each block’s population weight within its parent block group:

tables <- census2020_save_datasets(blocks)
names(tables)
#> "bgid2fips" "blockid2fips" "blockpoints" "blockwts" "quaddata"

Island Areas

Block-level data are not published for the Island Areas, so these come at block-group resolution via a dedicated helper:

islands <- census2020_get_data_islandareas()  # VI, GU, MP, AS

census2020_get_data() can also accept a mix of mainland states and Island Areas and will dispatch each to the appropriate reader.

Technical references

See the help for ?census2020_read for links to the Census Bureau technical documentation describing the files, tables, and variables.