dfw_beeswarm.R

library(tidycensus)
library(tidyverse)
library(tigris)
library(sf)
library(ggbeeswarm)
library(viridis)
library(extrafont)
options(tigris_use_cache = TRUE)

df <- get_acs(geography = "tract", state = "TX",  
              variables = c("B03002_012", "B03002_003", "B03002_004", "B03002_006"), 
              summary_var = "B19013_001", geometry = TRUE) %>%
  mutate(variable = recode(variable, B03002_003 = "White", 
                           B03002_004 = "Black", 
                           B03002_006 = "Asian", 
                           B03002_012 = "Hispanic")) %>%
  group_by(GEOID) %>%
  filter(estimate == max(estimate, na.rm = TRUE)) %>%
  ungroup() %>%
  filter(estimate != 0)

metro <- core_based_statistical_areas(cb = TRUE, class = "sf") %>%
  filter(str_detect(NAME, "Dallas"))

dfw <- df[metro, op = st_within]

ggplot(dfw, aes(x = variable, y = summary_est, color = summary_est)) +
  geom_quasirandom(alpha = 0.5) + 
  coord_flip() + 
  theme_minimal(base_family = "Tahoma") + 
  scale_color_viridis(guide = FALSE) + 
  scale_y_continuous(labels = scales::dollar) + 
  labs(x = "Largest group in Census tract", 
       y = "Median household income", 
       title = "Household income distribution by largest racial/ethnic group", 
       subtitle = "Census tracts, Dallas-Fort Worth metropolitan area", 
       caption = "Data source: 2011-2015 ACS")

Here is my attempt at getting all 3 metro areas into a single plot. (Hours later than I said, because I forgot to hit send)

library(tidycensus)
library(tidyverse)
library(tigris)
library(sf)
library(ggbeeswarm)
library(viridis)
library(extrafont)
options(tigris_use_cache = TRUE)

df2 <- get_acs(geography = "tract", state = c("TX", "MO", "KS", "CA"),  
              variables = c("B03002_012", "B03002_003", "B03002_004", "B03002_006"), 
              summary_var = "B19013_001", geometry = TRUE) %>%
  mutate(variable = recode(variable, B03002_003 = "White", 
                           B03002_004 = "Black", 
                           B03002_006 = "Asian", 
                           B03002_012 = "Hispanic")) %>%
  group_by(GEOID) %>%
  filter(estimate == max(estimate, na.rm = TRUE)) %>%
  ungroup() %>%
  filter(estimate != 0)


dfw <- core_based_statistical_areas(cb = TRUE, class = "sf") %>%
  filter(str_detect(NAME, "Dallas"))
kc <- core_based_statistical_areas(cb = TRUE, class = "sf") %>%
  filter(str_detect(NAME, "Kansas City"))
sac <- core_based_statistical_areas(cb = TRUE, class = "sf") %>%
  filter(str_detect(NAME, "Sacramento"))

metroD <- df2[dfw, op = st_within] %>% mutate(metro = "DFW")
metroKC <- df2[kc, op = st_within] %>% mutate(metro = "KC")
metroSAC <- df2[sac, op = st_within] %>% mutate(metro = "SAC")

allmetro <- rbind(metroD, metroKC, metroSAC)

ggplot(allmetro, aes(x = variable, y = summary_est, color = summary_est)) +
  geom_quasirandom(alpha = 0.5) + 
  coord_flip() + 
  theme_minimal(base_family = "Tahoma") + 
  scale_color_viridis(guide = FALSE) + 
  scale_y_continuous(labels = scales::dollar) +
  facet_grid(metro ~ .) +
  labs(x = "Largest group in Census tract", 
       y = "Median household income", 
       title = "Household income distribution by largest racial/ethnic group", 
       subtitle = "Census tracts", 
       caption = "Data source: 2011-2015 ACS")