# load libraries
library(tidyverse)
library(sf)
library(tidycensus)
library(tigris)
library(gt)
# set options
options(tigris_use_cache = TRUE)Part 1: Racial composition of Chicago, by tract
Get demographic data
Use the CENSUS API to download racial data from the 2018-2022 5-year ACS for Cook County.
B03002_003= White (non-Hispanic)B03002_004= Black (non-Hispanic)B03002_012= Hispanic
race_vars <- c(
totpop = "B03002_001",
white = "B03002_003",
black = "B03002_004",
hispanic = "B03002_012"
)
# download data via API
acs_2022 <- get_acs(
geography = "tract",
state = "IL", county = "Cook",
variables = race_vars,
output = "wide", geometry = TRUE,
year = 2022,
survey = "acs5"
)
# remove margin of error columns and trailing "E" from estimate
acs_2022 <- acs_2022 |>
select(-ends_with("M")) |>
rename_with(~ str_remove(., "E$"))Determine which tracts are in the City of Chicago
Download the boundary of the City of Chicago
# Download the boundary of Chicago
chi_bnd <-
places(state = "IL") |>
filter(NAME == "Chicago")
# make sure the CRS of the two datasets match (3435)
acs_2022 <- st_transform(acs_2022, 3435)
chi_bnd <- st_transform(chi_bnd, 3435)
# compute the intersection of the tracts with the boundary of Chicago
acs_2022_chi <- acs_2022 |>
st_intersection(chi_bnd) |>
select(GEOID:hispanic)Calculate the proportion of white, black, and hispanic in each tact
acs_2022_chi <- acs_2022_chi |>
# pivot the data from wide to long format
pivot_longer(
cols = c(white, black, hispanic),
names_to = "racial_group", values_to = "group_pop"
) |>
# compute proportion of each group in each tract
mutate(
group_proportion = group_pop / totpop
)Create map and analyze
Create faceted map showing the percentage distribution of the non-Hispanic White, non-Hispanic Black, and Hispanic populations across census tracts in the City of Chicago
# create faceted map by racial group
acs_2022_chi |>
ggplot() +
geom_sf(aes(fill = group_proportion * 100)) +
facet_wrap(~racial_group) +
scale_fill_viridis_c() +
labs(fill = "Percent of Total Population") +
theme_void()
The racial segregation of the city is quite striking. In the south, the majority of the population in Black; in the north, the majority of the population is White; and in the west, the majority of the population is Hispanic.
We see a similar partition in Map 2, which highlights population change. The south of Chicago (majority Black) tends to have a shrinking population, while the north and west tend to have a growing population. The far north has low change. This might suggest the racial composition of the city is changing, to be less Black and more White and Hispanic.
Part 2 - Changing populations in Cook County
Load LTDB data
Load and clean two years of historical LTDB population data (1970 and 2020)
# load data
F_1970 <- read_csv("data/LTDB_Std_1970_fullcount.csv") |>
filter(county == "Cook County" & state == "IL")
F_2020 <-
read_csv("data/ltdb_std_2020_fullcount.csv") |>
filter(substr(TRTID2010, 1, 5) == "17031")
# select and rename cols, standardize datatypes
S_1970 <- F_1970 |>
select(GEOID = TRTID10, POP1970 = POP70) |>
mutate(GEOID = as.character(GEOID))
S_2020 <- F_2020 |>
select(GEOID = TRTID2010, POP2020 = pop20) |>
mutate(GEOID = as.character(GEOID))
# combine
pop_data <-
reduce(list(S_1970, S_2020),
right_join,
by = "GEOID"
)
# categorize tracts by population change
pop_data <- pop_data |>
mutate(change = (POP2020 - POP1970) / POP1970) |>
mutate(status = case_when(
is.na(change) ~ "Uncategorized",
change > 0.10 ~ "1 - Growing",
change < -0.10 ~ "3 - Declining",
TRUE ~ "2 - Low Change"
))Combine LTDB data with current geography
# get current geography
F_2022 <- get_acs(
geography = "tract",
state = "IL",
county = "Cook",
variables = "B01001_001",
year = 2022,
geometry = TRUE,
) |>
select(GEOID)
# merge on GEOID, ensure its an sf
pop_data_geo <- pop_data |>
inner_join(F_2022, by = "GEOID") |>
st_as_sf()Create map
# create map
pop_data_geo |>
ggplot() +
geom_sf(aes(fill = status)) +
scale_fill_manual(
values = c(
"1 - Growing" = "#2ca25f",
"2 - Low Change" = "#99d8c9",
"3 - Declining" = "#e5f5f9",
"Uncategorized" = "black"
)
) +
labs(fill = "Population Change") +
geom_sf(
data = chi_bnd, color = "yellow",
linewidth = 0.5, fill = NA
) +
theme_void()
Conclusion
You can see clear population grown in several areas in Cook Country but outside of the city, suggesting some suburbanization. To make this map more clear, I think it would be interesting to combine tracts with their neighbors somehow, reducing the number of polygons and making larger trends (potentially) easier to see.