Shift or rotate states when mapping (good for simpler US maps)

move_state.R

# This process is adapted from https://sesync-ci.github.io/blog/transform-Alaska-Hawaii.html
# Here, it's offered as a function to simplify trial-and-error.

move_state <- function(
    df, # spatial dataframe
    choice, # value in "state" column
    rotation, # eg -39 * pi/180
    right, # amount to move eastward; use negative for left/west
    up# amount to move northward; use negative for down/south
  ){
    rot <- function(a){
      matrix(c(cos(a), sin(a), -sin(a), cos(a)), 2, 2)
    }
    
    df_state <- df |> 
      filter(state == choice)
    
    centroid <- df_state |> 
      st_geometry() |> 
      st_union() |> 
      st_centroid()
    
    transformation <- (st_geometry(df_state) - centroid) * rot(rotation) + centroid + c(right, up)
    
    df_state <- df_state |> 
      st_set_geometry(transformation) |> 
      st_set_crs(st_crs(df))
    
    df |> 
      filter(state != choice) |> 
      rbind(df_state)
  }

Mapping just the United States can result in an awkward map:

my_map <- 
  tigris::states(cb = TRUE) |> 
  filter(NAME %in% state.name) |> 
  st_transform("ESRI:102008")

my_map |> 
  ggplot() + 
  geom_sf(fill = "salmon")

Quentin Read shows a good process to move the spatial coordinates of states, including moving, rotating, and shrinking them here: https://sesync-ci.github.io/blog/transform-Alaska-Hawaii.html

I wasn't satisfied with Alaska's shrinking because it's more inaccurate than is needed. (See Claus Wilke's Fundamentals of Data Visualization chapter 15.) Keeping Alaska its original size necessitated playing around with the numbers to get it someplace not bumping into another state. This function simplifies that trial and error. Note that it expects the data frame to have a filtering column called state.

my_map2 <- my_map |> 
  rename(state = NAME) |> 
  move_state("Alaska", 
             rotation = -40 * pi/180,
             right = 1350000,
             up = -5500000) |> 
  move_state("Hawaii", 
             rotation = -35 * pi/180,
             right = 6800000,
             up = -2000000) 

my_map2 |>
  ggplot() +
  geom_sf(fill = "salmon") + 
  theme_void()