Making NYT-style building maps with data from Microsoft

nyt_style_buildings.R

library(tidyverse)
library(sf)
library(tigris)

# start by picking a state from https://github.com/Microsoft/USBuildingFootprints
# WARNING: these files can be pretty big. using arizona for its copious subdivisions and reasoanable 83MB.
url_footprint <- "https://usbuildingdata.blob.core.windows.net/usbuildings-v1-1/Arizona.zip"
download.file(url_footprint, "Arizona.zip")
unzip("Arizona.zip")

# read in building shapes 
buildings <- st_read("Arizona.geojson")

# use the tigris package to get zctas boundaries (basically zip codes)
options(tigris_class = "sf", tigris_use_cache = TRUE) # defaults to sp; manually set to sf
zctas_national <- zctas(cb = TRUE)

# get the geometry for the zip codes we want to map
selected_zip <- zctas_national %>%
  # input the zipcode(s) you want here. using 85308 for glendale, az
  filter(GEOID10 %in% c(85308)) %>%
  st_transform(st_crs(buildings))

# clip buildings file to just the selected zipcode
building_select <- st_intersection(selected_zip, buildings)

# plot using using geom_sf
glendale <- ggplot() +
  geom_sf(data = building_select, fill = "black", color = NA) +
  theme_void() +
  # manually dropping the grid to solve a bug in geom_sf 
  theme(panel.grid.major = element_line(colour = 'transparent'))

ggsave("glendale_az.png", glendale, dpi = 400, width = 8, height = 5)

Do you have a census API key? tigris may require that...