Using a Hexagonal Grid for Canada's Election Map

Using a Hexagonal Grid for Canada's Election Map

Us Canadians are going to be looking at a lot of election maps in the next few hours, so I thought I’d try to make a few! First, we need to download the Riding geography from open Canada (there is more extensive information on Elections and Ridings at the Library of Parliament website). Next, we need the tidyverse and the sf packages.

library(tidyverse)
library(sf)
## Linking to GEOS 3.6.1, GDAL 2.1.3, PROJ 4.9.3

The riding boundaries come in shapefile format, which is handled by a one-liner to read_sf(). I’m going to simplify the boundaries a bit to speed up the plotting time. The dTolderance argument is in map units, and it took some experimenting to settle on the number 100.

ridings <- read_sf("boundaries_2015_shp_en/FED_CA_2_2_ENG.shp") %>%
  st_simplify(dTolerance = 100)
ridings
## Simple feature collection with 347 features and 13 fields
## geometry type:  GEOMETRY
## dimension:      XY
## bbox:           xmin: 3658201 ymin: 658873 xmax: 9019157 ymax: 6083005
## epsg (SRID):    NA
## proj4string:    +proj=lcc +lat_1=49 +lat_2=77 +lat_0=63.390675 +lon_0=-91.86666666666666 +x_0=6200000 +y_0=3000000 +datum=NAD83 +units=m +no_defs
## # A tibble: 347 x 14
##    FED_NUM NID   FEDNUM ENNAME FRNAME PROVCODE CREADT REVDT REPORDER
##      <dbl> <chr>  <dbl> <chr>  <chr>  <chr>    <chr>  <chr> <chr>   
##  1   35029 {30F…  35029 Etobi… Etobi… ON       20131… <NA>  2013    
##  2   35032 {39A…  35032 Guelph Guelph ON       20131… <NA>  2013    
##  3   48017 {C3B…  48017 Edmon… Edmon… AB       20131… <NA>  2013    
##  4   48018 {C5B…  48018 Edmon… Edmon… AB       20131… <NA>  2013    
##  5   48021 {BBF…  48021 Edmon… Edmon… AB       20131… <NA>  2013    
##  6   48022 {F45…  48022 Footh… Footh… AB       20131… <NA>  2013    
##  7   48023 {B4C…  48023 Fort … Fort … AB       20131… <NA>  2013    
##  8   48024 {4CA…  48024 Grand… Grand… AB       20131… 2014… 2013    
##  9   48025 {722…  48025 Lakel… Lakel… AB       20131… <NA>  2013    
## 10   24065 {858…  24065 Marc-… Marc-… QC       20131… 2014… 2013    
## # … with 337 more rows, and 5 more variables: DECPOPCNT <dbl>,
## #   QUIPOPCNT <dbl>, ENLEGALDSC <chr>, FRLEGALDSC <chr>, geometry <POLYGON
## #   [m]>

Next, we can plot! I opted to use ggplot2’s geom_sf(), using theme_void() to strip away the axes/labels.

ggplot(ridings) + 
  geom_sf(aes(fill = PROVCODE)) + 
  theme_void()

This is the type of map that all the major outlets are using to communicate election results (I’m looking at you, CBC and Global), but it’s misleading because each riding has an equal say in forming government.

A better way would be to give each riding an equal area. Luckily many others have thought of this, and in R we can use the geogrid package to calculate a hexagonal (or rectangular) grid, to which we can then assign ridings. This takes about 20 minutes…it may sound slow, but it’s way faster than trying to do this by hand!

library(geogrid)
ridings_grid <- ridings %>%
  calculate_grid(grid_type = "hexagonal", seed = 1938)

ridings_hex <- assign_polygons(ridings, ridings_grid)

Now we can use similar code to make a “map” of ridings, this time with each riding having equal area:

ggplot(ridings_hex) +
  geom_sf(aes(fill = PROVCODE)) +
  theme_void()

The default grid/riding assignment definitely has some problems: PEI’s ridings are mashed up with New Brunswick, Newfoundland, and Labrador. Setting the seed argument in calculate_grid() may help calculate alternate grids, although it’s probably a better bet to calculate by province and arrange things manually. Either way, it’s far superior to the geographical map!

A cool way to see where the seats are coming from is animation! Luckily, the gganimate package supports spatial objects out of the box. In order to combine the objects, we need to make sure they have the same CRS (I set it to NA here because we’re no longer in geographic coordinates with the hex grid) and the same columns. Then, we can use transition_states() to animate!

library(gganimate)
ridings_combined <- rbind(
  ridings %>% 
    mutate(type = "geographic") %>% 
    st_set_crs(NA) %>% 
    select(PROVCODE, type, geometry),
  ridings_hex %>%
    mutate(type = "hex") %>% 
    st_set_crs(NA) %>% 
    select(PROVCODE, type, geometry)
)

plot_obj <- ggplot(ridings_combined) +
  geom_sf(aes(fill = PROVCODE)) +
  theme_void() +
  transition_states(type, transition_length = 1, state_length = 5)

animate(plot_obj, width = 700, height = 700, res = 96)

Avatar
Dewey Dunnington
Geoscientist, Programmer, Educator