austin_distance.R

library(tidycensus)
library(tigris)
library(tidyverse)
library(sf)
library(extrafont)
options(tigris_use_cache = TRUE, tigris_class = "sf")

########################################################################
########################################################################
# Chart 1
########################################################################
########################################################################

# Get the data
austin_metro <- map(c(2010, 2016), function(x) {
  get_acs(geography = "tract", 
          state = "TX", 
          county = c("Bastrop", "Caldwell", "Hays", "Travis", "Williamson"),
          variables = c(totalpop = "B15002_001"), 
          geometry = TRUE, 
          year = x) %>%
    mutate(endyear = x)
}) %>%
  reduce(rbind)

austin_hall <- c(-97.747300, 30.265682) %>%
  st_point() %>%
  st_sfc(crs = st_crs(austin_metro)) %>%
  st_transform(26914)

austin_dist <- austin_metro %>%
  st_transform(26914) %>%
  mutate(dist = as.numeric(
    st_distance(
      st_centroid(.), austin_hall
    )
  )) %>%
  mutate(distcat = case_when(
    dist < 10000 ~ "0-10km", 
    dist < 20000 ~ "10-20km", 
    dist < 30000 ~ "20-30km", 
    dist < 40000 ~ "30-40km", 
    TRUE ~ "40km+"
  )) %>%
  as.data.frame() %>%
  group_by(distcat, endyear) %>%
  summarize(pop = sum(estimate, na.rm = TRUE)) %>%
  ungroup() %>%
  spread(key = endyear, value = pop) %>%
  rename(pop2010 = `2010`, pop2016 = `2016`) %>%
  mutate(pctchange = 100 * ((pop2016 - pop2010) / pop2010))

ggplot(austin_dist) + 
  theme_minimal(base_family = "Verdana", base_size = 12) + 
  geom_linerange(aes(ymin = -2, x = reorder(distcat, desc(distcat)),  
                   ymax = pctchange), 
               color = "blue", size = 1) + 
  geom_point(aes(y = pctchange, x = reorder(distcat, desc(distcat))), 
             color = "blue", size = 4) + 
  coord_flip() + 
  labs(y = "Percent growth between 2006-10 and 2012-16", 
       x = "", 
       title = "Population growth age 25+ by distance band", 
       subtitle = "Austin, TX metropolitan area", 
       caption = "Data acquired with the R tidycensus package, ACS table B15002.\nDistance measured from Census tract centroids to Austin City Hall.\nChart by @kyle_e_walker.") + 
  scale_y_continuous(labels = function(g) {paste0(g, "%")}) + 
  theme(panel.grid.major.x = element_blank(), 
        panel.grid.minor.x = element_blank(),
        panel.grid.minor.y = element_blank(), 
        plot.title = element_text(face = "bold"), 
        plot.caption = element_text(size = 7), 
        plot.margin = unit(c(0.2, 0.5, 0.2, 0.2), "in"))

ggsave("general.png", dpi = 300, height = 5, width = 8)



########################################################################
########################################################################
# Chart 2
########################################################################
########################################################################

# By whether you do - or don't - have a college degree
vars <- paste0("B15002_0", c(15:18, 32:35))

austin_metro <- map(c(2010, 2016), function(x) {
  get_acs(geography = "tract", 
          state = "TX", 
          county = c("Bastrop", "Caldwell", "Hays", "Travis", "Williamson"),
          variables = vars, 
          summary_var = "B15002_001", 
          year = x, 
          geometry = TRUE) %>%
    mutate(endyear = x) %>%
    group_by(GEOID) %>%
    mutate(ba_or_higher = sum(estimate, na.rm = TRUE)) %>%
    ungroup() %>%
    distinct(GEOID, .keep_all = TRUE) %>%
    mutate(less_than_ba = summary_est - ba_or_higher)
}) %>%
  reduce(rbind) %>%
  select(GEOID, NAME, endyear, ba_or_higher, less_than_ba, over25 = summary_est)



austin_hall <- c(-97.747300, 30.265682) %>%
  st_point() %>%
  st_sfc(crs = st_crs(austin_metro)) %>%
  st_transform(26914)

austin_dist <- austin_metro %>%
  st_transform(26914) %>%
  mutate(dist = as.numeric(
    st_distance(
      st_centroid(.), austin_hall
    )
  )) %>%
  mutate(distcat = case_when(
    dist < 10000 ~ "0-10km", 
    dist < 20000 ~ "10-20km", 
    dist < 30000 ~ "20-30km", 
    dist < 40000 ~ "30-40km", 
    TRUE ~ "40km+"
  )) %>%
  as.data.frame() %>%
  group_by(distcat, endyear) %>%
  summarize(ba_or_higher_grouped = sum(ba_or_higher, na.rm = TRUE), 
            less_than_ba_grouped = sum(less_than_ba, na.rm = TRUE)) %>%
  ungroup() %>%
  gather(key = level, value = pop, -distcat, -endyear) %>%
  spread(key = endyear, value = pop) %>%
  rename(pop2010 = `2010`, pop2016 = `2016`) %>%
  mutate(pctchange = 100 * ((pop2016 - pop2010) / pop2010), 
         level = case_when(
           level == "ba_or_higher_grouped" ~ "Bachelor's degree or higher", 
           level == "less_than_ba_grouped" ~ "Less than bachelor's degree"
         ))

ggplot(austin_dist) + 
  theme_minimal(base_family = "Verdana") + 
  geom_linerange(aes(color = level, ymin = -0.2, x = reorder(distcat, desc(distcat)),  
                   ymax = pctchange), 
               size = 1, position = position_dodge(width = 0.5)) + 
  geom_point(aes(color = level, y = pctchange, x = reorder(distcat, desc(distcat))), 
             size = 4, position = position_dodge(width = 0.5)) + 
  coord_flip() + 
  scale_color_manual(values = c("navy", "lightblue")) + 
  labs(y = "Percent change between 2006-10 and 2012-16", 
       x = "", 
       title = "Population change age 25+ by distance band", 
       subtitle = "Austin, TX metropolitan area", 
       caption = "Data acquired with the R tidycensus package, ACS table B15002.\nDistance measured from Census tract centroids to Austin City Hall.\nChart by @kyle_e_walker.", 
       color = "") + 
  scale_y_continuous(labels = function(g) {paste0(g, "%")}) + 
  theme(panel.grid.major.x = element_blank(), 
        panel.grid.minor.x = element_blank(),
        panel.grid.minor.y = element_blank(), 
        plot.title = element_text(face = "bold"), 
        plot.caption = element_text(size = 7)) + 
  guides(color = guide_legend(reverse = TRUE))


ggsave("bycollege.png", dpi = 300)