Mapping All Fifty U.S. States

William Murphy

2016-11-13

This data package provides an easy way to plot 50-state choropleth thematic maps with ggplot2 in R.

Motivation

It is far too difficult to create a U.S. state data visualization in R that includes Alaska and Hawaii. The U.S. state map included in the maps package includes only the 48 contiguous states. A search for existing solutions will lead to Stack Overflow threads (1, 2), but the solutions available require either limiting plotting options by rendering with a specific package, installing a large list of dependencies and tracking down shape files to perform arcane transformation calculations, or using annotations or visibility-based maps that compromise the aesthetic of the visualization.

This package is a simple solution: a data file, with no dependencies, that can be dropped into ggplot2 with no effort. To create it, I followed the steps in a Stack Overflow answer from Barry Rowlingson to transform Alaska and Hawaii into insets next to the contiguous states and packaged the resulting data frame for simple drop-in use.

Usage Examples

To use the fifty state map, simply point the map argument in geom_map to fifty_states. As usual, you’ll also create an aesthetic mapping for map_id to the column in your data frame that contains the lower case state names.

If you need to be highly technically accurate, you can add bounding boxes to highlight the insets so that no viewer confuses Alaska and Hawaii for regions in a deformed Mexico that you randomly decided to include. The function called fifty_states_inset_boxes will return a layer created via annotate, so you can just add it to your plot.

library(ggplot2)
library(fiftystater)

data("fifty_states") # this line is optional due to lazy data loading

crimes <- data.frame(state = tolower(rownames(USArrests)), USArrests)

# map_id creates the aesthetic mapping to the state name column in your data
p <- ggplot(crimes, aes(map_id = state)) + 
  # map points to the fifty_states shape data
  geom_map(aes(fill = Assault), map = fifty_states) + 
  expand_limits(x = fifty_states$long, y = fifty_states$lat) +
  coord_map() +
  scale_x_continuous(breaks = NULL) + 
  scale_y_continuous(breaks = NULL) +
  labs(x = "", y = "") +
  theme(legend.position = "bottom", 
        panel.background = element_blank())

p
# add border boxes to AK/HI
p + fifty_states_inset_boxes()

This package also plays nicely with my colorplaner package for bivariate color scales.

# Map a second variable to each state's fill color with colorplaner
library(colorplaner)
p + aes(fill2 = UrbanPop) + scale_fill_colorplane() +
  theme(legend.position = "right")

The map data included in this package can be generated with the code below. If you want to run this code, you will need to ensure the packages listed below are installed; they are not automatically installed with fiftystater.

# Create map data with AK, HI inset.
library(maptools)
library(rgeos)
library(rgdal)
library(dplyr)

transform_state <- function(object, rot, scale, shift){
  object %>% elide(rotate = rot) %>%
    elide(scale = max(apply(bbox(object), 1, diff)) / scale) %>%
    elide(shift = shift)
}

#state shape file from
# http://www.arcgis.com/home/item.html?id=f7f805eb65eb4ab787a0a3e1116ca7e5
loc <- file.path(tempdir(), "stats_dat")
unzip(system.file("extdata", "states_21basic.zip", package = "fiftystater"),
      exdir = loc)
fifty_states_sp <- readOGR(dsn = loc, layer = "states", verbose = FALSE) %>%
  spTransform(CRS("+init=epsg:2163"))

alaska <- fifty_states_sp[fifty_states_sp$STATE_NAME == "Alaska", ] %>%
  transform_state(-35, 2.5, c(-2400000, -2100000))
proj4string(alaska) <- proj4string(fifty_states_sp)

hawaii <- fifty_states_sp[fifty_states_sp$STATE_NAME == "Hawaii", ] %>%
  transform_state(-35, .75, c(-1170000,-2363000))
proj4string(hawaii) <- proj4string(fifty_states_sp)

fifty_states <-
  fifty_states_sp[!fifty_states_sp$STATE_NAME %in% c("Alaska","Hawaii"), ] %>%
  rbind(alaska) %>%
  rbind(hawaii) %>%
  spTransform(CRS("+init=epsg:4326")) %>%
  fortify(region = "STATE_NAME") %>%
  mutate(id = tolower(id))