getACS

The goal of getACS is to make it easier to work with American Community Survey data from the tidycensus package by Kyle Walker and others.

This package includes:

• Functions that extend tidycensus::get_acs() to support multiple tables, geographies, or years
• Functions for creating formatted tables from ACS data using the gt package

As of April 2024, this package uses a development version of {tigris}, available at https://github.com/elipousson/tigris.

Installation

You can install the development version of getACS from GitHub with:

# install.packages("pak")
pak::pkg_install("elipousson/getACS")

Usage

library(getACS)
library(gt)
library(ggplot2)

The main feature of {getACS} is support for returning multiple tables, geographies, and years.

acs_data <- get_acs_geographies(
  geography = c("county", "state"),
  county = "Baltimore city",
  state = "MD",
  table = "B08134",
  year = 2022,
  quiet = TRUE
)

The package also includes utility functions for filtering data and selecting columns to support the creation of tables using the {gt} package:

tbl_data <- filter_acs(acs_data, indent == 1, line_number <= 10)
tbl_data <- select_acs(tbl_data)

commute_tbl <- gt_acs(
  tbl_data,
  groupname_col = "NAME",
  column_title_label = "Commute time",
  table = "B08134"
)

as_raw_html(commute_tbl)

Commute time	Est.	% share
Baltimore city, Maryland
Less than 10 minutes	16,140 ± 1,096	7% ± 0%
10 to 14 minutes	20,798 ± 1,312	9% ± 1%
15 to 19 minutes	36,667 ± 1,740	16% ± 1%
20 to 24 minutes	39,803 ± 1,834	17% ± 1%
25 to 29 minutes	16,404 ± 970	7% ± 0%
30 to 34 minutes	40,744 ± 1,695	17% ± 1%
35 to 44 minutes	16,880 ± 1,160	7% ± 0%
45 to 59 minutes	20,318 ± 1,296	9% ± 1%
60 or more minutes	26,662 ± 1,207	11% ± 0%
Maryland
Less than 10 minutes	203,738 ± 3,511	8% ± 0%
10 to 14 minutes	255,052 ± 4,240	10% ± 0%
15 to 19 minutes	333,717 ± 5,269	13% ± 0%
20 to 24 minutes	342,189 ± 4,777	13% ± 0%
25 to 29 minutes	177,597 ± 3,129	7% ± 0%
30 to 34 minutes	400,919 ± 5,980	15% ± 0%
35 to 44 minutes	249,413 ± 4,443	9% ± 0%
45 to 59 minutes	312,390 ± 4,394	12% ± 0%
60 or more minutes	371,252 ± 4,828	14% ± 0%
Source: 2018-2022 ACS 5-year Estimates, Table B08134.

The gt_acs_compare() function also allows side-by-side comparison of geographies:

commute_tbl_compare <- gt_acs_compare(
  data = tbl_data,
  id_cols = "column_title",
  column_title_label = "Commute time",
  table = "B08134"
)

as_raw_html(commute_tbl_compare)

Commute time	Baltimore city, Maryland		Maryland
	Est.	% share	Est.	% share
Less than 10 minutes	16,140 ± 1,096	7% ± 0%	203,738 ± 3,511	8% ± 0%
10 to 14 minutes	20,798 ± 1,312	9% ± 1%	255,052 ± 4,240	10% ± 0%
15 to 19 minutes	36,667 ± 1,740	16% ± 1%	333,717 ± 5,269	13% ± 0%
20 to 24 minutes	39,803 ± 1,834	17% ± 1%	342,189 ± 4,777	13% ± 0%
25 to 29 minutes	16,404 ± 970	7% ± 0%	177,597 ± 3,129	7% ± 0%
30 to 34 minutes	40,744 ± 1,695	17% ± 1%	400,919 ± 5,980	15% ± 0%
35 to 44 minutes	16,880 ± 1,160	7% ± 0%	249,413 ± 4,443	9% ± 0%
45 to 59 minutes	20,318 ± 1,296	9% ± 1%	312,390 ± 4,394	12% ± 0%
60 or more minutes	26,662 ± 1,207	11% ± 0%	371,252 ± 4,828	14% ± 0%
Source: 2018-2022 ACS 5-year Estimates, Table B08134.

gt_acs_compare_vars() is a variant on gt_acs_compare() where the default values support comparisons with values in columns and geographical areas in rows:

commute_tbl_compare_vars <- acs_data |>
  filter_acs(indent == 1, line_number > 10) |>
  gt_acs_compare_vars(
    table = acs_data$table_id
  )

as_raw_html(commute_tbl_compare_vars)

NAME	Car, truck, or van	Public transportation (excluding taxicab)	Walked	Taxicab, motorcycle, bicycle, or other means
Baltimore city, Maryland	176,543 ± 2,817	34,640 ± 1,637	14,954 ± 1,007	8,279 ± 901
Maryland	2,357,924 ± 11,085	171,785 ± 3,655	59,507 ± 1,858	57,051 ± 2,213
Source: 2018-2022 ACS 5-year Estimates, Table B08134.

The package also includes several functions to support creating plots with the {ggplot2} package including geom_acs_col() and labs_acs_survey():

plot_data <- acs_data |>
  filter_acs(indent == 1, line_number > 10) |>
  select_acs() |>
  fmt_acs_county(state = "Maryland")

plot_data |>
  ggplot() +
  geom_acs_col(
    fill = "NAME",
    position = "dodge",
    color = NA,
    alpha = 0.75,
    perc = TRUE,
    errorbar_params = list(position = "dodge", linewidth = 0.25)
  ) +
  scale_y_discrete("Means of transportation to work") +
  scale_fill_viridis_d("Geography") +
  labs_acs_survey(
    .data = acs_data
  ) +
  theme_minimal()

The geom_acs_col() function calls geom_acs_errorbar() (passing the errorbar_params argument as additional parameters) and scale_x_acs() or scale_y_acs() (depending on whether orientation = "y" or the default value of NA).

For more information on working with Census data in R read the book Analyzing US Census Data: Methods, Maps, and Models in R (February 2023).

Related projects

Related R packages and analysis projects

• {easycensus}: Quickly Extract and Marginalize U.S. Census Tables
• {cwi}: Functions to speed up and standardize Census ACS data analysis for multiple staff people at DataHaven, preview trends and patterns, and get data in more layperson-friendly
• {camiller}: A set of convenience functions, functions for working with ACS data via {tidycensus}
• {psrccensus}: A set of tools developed for PSRC (Puget Sound Regional Council) staff to pull, process, and visualize Census Data for geographies in the Central Puget Sound Region.
• {CTPPr}: A R package for loading and working with the US Census CTPP survey data.
• {lehdr}: a package to grab LEHD data in support of city and regional planning economic and transportation analysis
• {mapreliability}: A R package for map classification reliability calculator
• Studying Neighborhoods With Uncertain Census Data: Code to create and visualize demographic clusters for the US with data from the American Community Survey

Related Python libraries

• census-data-aggregator: A Python library from the L.A. Times data desk to help “combine U.S. census data responsibly”
• census-table-metadata: Tools for generating metadata about tables and fields in a Census release based on sequence lookup and table shell files. (Note: the pre-computed data from this repository is used to label ACS data by label_acs_metadata())