redistio

redistio provides a point-and-click districting interface powered by Shiny and mapgl. For regular sf objects, it can be used to draw districts and export assignment files. For redist_map objects, algorithmic assistance is enabled for map drawing. It also includes an interactive adjacency graph editor via adj_editor().

Installation

You can install the development version of redistio from GitHub with:

pak::pak('christopherkenny/redistio')

Or from R-universe with:

install.packages('redistio', repos = c('https://christopherkenny.r-universe.dev', 'https://cloud.r-project.org'))

Example

The most basic application of redistio starts with an sf tibble and a column of district assignments.

library(redistio)

draw(dc, dc$ward)

Plans Browser and Comparisons

When working with redist_map objects and redist_plans, you can pass pre-computed plans to draw() for browsing and comparison. This enables two additional tabs in the app:

• Plans: Browse and preview simulated plans on the map, and adopt any plan as your current working plan.
• Comparisons: Compare your current plan against the simulated plans using summary statistics and plots.

library(redistio)
library(redist)

map <- redist_map(dc, existing_plan = ward)
plans <- redist_smc(map, nsims = 100)

draw(map, init_plan = map$ward, plans = plans, plans_fn = add_plan_stats)

The plans_fn argument accepts a function that adds the current plan to the redist_plans object and computes summary statistics for comparison. The built-in add_plan_stats() function handles this automatically, computing statistics like population parity, compactness, partisan metrics, and administrative splits to match whatever columns are present in plans.

Adjacency Editor

redistio includes an interactive adjacency graph editor for fixing adjacency issues in your data (e.g., removing edges across rivers or adding edges for missing connections):

library(redistio)

adj_editor(dc, init_plan = dc$ward)

The editor lets you add or remove edges by clicking pairs of precincts, and provides an export tab with an edit log and reproducible code using geomander functions.

Configuration

Editor options

The editor can be configured primarily through redistio_options(). Some configurable options:

• theme: the bslib theme to use
• panels: which panels to include. Allows for removing panels like elections, if drawing districts in a party-blind manner
• palette_party: the colors to use for partisan maps
• palette_pop: the colors to use for population maps
• palette_pct: the colors to use for percentage maps. Diverging palettes are recommended.
• map_tiles: the base map style function to use from mapgl
• use_algorithms: whether to enable algorithmic assistance for map drawing. Requires a redist_map object.
• alg_max_districts: the maximum number of districts allowed in algorithmic simulations
• alg_max_sims: the maximum number of simulations allowed in algorithmic simulations
• use_plans: whether to show the plans browser tab when plans is provided to draw()
• use_comparisons: whether to show the comparisons tab when plans and plans_fn are provided to draw()
• use_planscore: whether to use the PlanScore API to evaluate plans. Requires internet access and a PlanScore key.
• locked_districts: districts to lock at app start, preventing edits to those districts
• projection: the Maplibre map projection to use (default is 'mercator')

Data Dependent Options

Several data-based options may be configured inside draw():

• layers: Columns to use as toggle layers, where you can show things like county lines above the shapes
• elect_cols: Specify election columns directly. This use a guessing approach which follows the ALARM Project column naming schema by default.
• demog_cols: Specify demographic columns directly. This use a guessing approach which follows the ALARM Project column naming schema by default.
• split_cols: Specify columns which contain administrative units to check splits for.
• hover_fn: A function to display precinct-data based on the row of shp that the mouse is over.
• plans: A redist_plans object for browsing and comparison.
• plans_fn: A function to add a reference plan and compute summary statistics (e.g., add_plan_stats()).

More complex apps

These features allow for analyzing more complex maps, with live comparisons with simulations. For example, with the following code, you can have many more options:

library(redistio)
nj <- alarmdata::alarm_50state_map('NJ')
pl <- alarmdata::alarm_50state_plans('NJ')
draw(
  nj,
  init_plan = nj$cd_2020,
  palette = ggredist::ggredist$alaska,
  layers = list(County = 'county'),
  plans = pl,
  plans_fn = add_plan_stats
)