Package 'geomander'

Description

A compilation of tools to complete common tasks for studying gerrymandering. This focuses on the geographic tool side of common problems, such as linking different levels of spatial units or estimating how to break up units. Functions exist for creating redistricting-focused data for the US.

Package Content

Index of help topics:

add_edge                Add Edges to an Adjacency List
adjacency               Build an Adjacency List
alarm_states            List Available States from ALARM Data
baf_to_vtd              Estimate Plans from a Block Assignment File to
                        Voting Districts
block2prec              Aggregate a Block Table by Matches
block2prec_by_county    Aggregate a Block Table by Matches Within
                        County
check_contiguity        Check Contiguity by Group
check_polygon_contiguity
                        Check Polygon Contiguity
checkerboard            Checkerboard
checkerboard_adj        Checkerboard Adjacency
clean_vest              Clean VEST Column Names
compare_adjacencies     Compare Adjacency Lists
count_connections       Count How Often Pairs of Units Share a District
create_block_table      Create a Census Block-Level Table
create_tract_table      Create an ACS Tract-Level Table
dra2r                   Convert DRA Export Data to an R Spatial Object
estimate_down           Estimate Values Down Using Precomputed Matches
estimate_up             Aggregate Values Up Using Precomputed Matches
geo_estimate_down       Estimate Values Down to a Finer Geography
geo_estimate_up         Aggregate Values Up to a Larger Geography
geo_filter              Filter to Intersecting Pieces
geo_match               Match Features Across Geographic Layers
geo_plot                Plot an 'sf' Object with Row Numbers
geo_plot_group          Plot Groups with Connected Components Colored
geo_sort                Sort Features by Northwest-to-Southeast Order
geo_trim                Trim Away Small Pieces
geomander-package       Geographic Tools for Studying Gerrymandering
geos_centerish          Get the kind of center of each shape
geos_circle_center      Get the centroid of the maximum inscribed
                        circle
get_alarm               Get an ALARM Dataset
get_dra                 Get a Dave's Redistricting App Dataset
get_heda                Get a Harvard Election Data Archive ("HEDA")
                        Dataset
get_lewis               Get Historical United States Congressional
                        District Shapefiles
get_rpvnearme           Get the RPV Near Me Dataset
get_vest                Get a Voting and Election Science Team ("VEST")
                        Dataset
global_gearys           Compute Global Geary's C
global_morans           Compute Global Moran's I
gstar_i                 Compute Standardized Getis-Ord G*i
heda_states             List Available HEDA States
local_gearys            Compute Local Geary's C
local_morans            Compute Local Moran's I
nrcsd                   nrcsd
orange                  orange
precincts               precincts
r2dra                   Convert an R Plan to DRA Block Assignment
                        Format
regionalize             Estimate Regions Separated by Geographic
                        Features
rockland                rockland
seam_adj                Filter Adjacency to the Edges Along a Seam
seam_geom               Filter Shapes to Units Along a Seam
seam_rip                Remove Edges Across an Administrative Boundary
seam_sew                Suggest Edges to Sew a Seam
split_precinct          Split a Precinct
st_centerish            Get the kind of center of each shape
st_circle_center        Get the Center of the Maximum Inscribed Circle
subtract_edge           Remove Edges from an Adjacency List
suggest_component_connection
                        Suggest Connections for Disconnected Groups
suggest_neighbors       Suggest Neighbors for Lonely Precincts
towns                   towns
va_blocks               va_blocks
va_vtd                  va_vtd
va18sub                 va18sub
vest_states             List Available VEST States

Maintainer

Christopher T. Kenny <[email protected]>

Author(s)

Christopher T. Kenny [aut, cre] (ORCID: <https://orcid.org/0000-0002-9386-6860>), Cory McCartan [ctb] (ORCID: <https://orcid.org/0000-0002-6251-669X>)

Description

Add one or more undirected edges to an adjacency list in place and return the modified list.

Usage

add_edge(adj, v1, v2, ids = NULL, zero = TRUE)

Arguments

Value

Adjacency list with the requested edges added symmetrically.

Examples

data(towns)
adj <- adjacency(towns)

add_edge(adj, 2, 3)
add_edge(adj, "West Haverstraw", "Stony Point", towns$MUNI)

Description

This mimics redist's redist.adjacency using geos to create the patterns, rather than sf. This is faster than that version, but forces projections.

Usage

adjacency(shp, epsg = 3857)

Arguments

Value

Zero-indexed adjacency list with nrow(shp) entries.

Examples

data(precincts)
adj <- adjacency(precincts)

Description

List Available States from ALARM Data

Usage

alarm_states()

Value

character vector of state abbreviations inferred from the ALARM repository contents.

Examples

## Not run: 
# relies on internet availability and interactivity on some systems
alarm_states()

## End(Not run)

Description

District lines are often provided at the census block level, but analyses often occur at the voting district level. This provides a simple way to estimate the block level to the voting district level.

Usage

baf_to_vtd(baf, plan_name, GEOID = "GEOID", year = 2020)

Arguments

Details

If a voting district is split between blocks, this currently uses the most common district.

When a voting district contains blocks assigned to more than one district, the output uses the modal district assignment.

Value

tibble with one row per voting district and the requested plan column

Examples

# Not guaranteed to reach download from redistrict2020.org
## Not run: 
# download and read baf ----
url <- paste0('https://github.com/PlanScore/Redistrict2020/', 
              'raw/main/files/DE-2021-01/DE_SLDU_bef.zip')
tf <- tempfile('.zip')
utils::download.file(url, tf)
utils::unzip(tf, exdir = dirname(tf))
baf <- readr::read_csv(
  file = paste0(dirname(tf), '/DE_SLDU_bef.csv'),
  col_types = 'ci'
)
names(baf) <- c('GEOID', 'ssd_20')

# convert to vtd level ----
baf_to_vtd(baf = baf, plan_name = 'ssd_20', 'GEOID')

## End(Not run)

Description

Aggregate block-level attributes up to a larger geography, usually precincts, using a vector of group assignments such as the output of geo_match().

Usage

block2prec(block_table, matches, geometry = FALSE)

Arguments

Details

If matches carries a "matching_max" attribute, the output preserves all target groups up to that size, even when some groups receive no matched rows. Missing nonnegative numeric summaries are filled with 0.

Value

dataframe with one row per matched group. When empty target groups are preserved, the number of rows may be larger than length(unique(matches)).

Examples

set.seed(1)
data(rockland)
rockland$id <- sample(c(1:2, 4), nrow(rockland), TRUE)
block2prec(rockland, rockland$id)

Description

Variant of block2prec() that matches blocks to precincts separately within each county. This helps avoid cross-county mismatches when county boundaries and census blocks do not line up cleanly.

Usage

block2prec_by_county(block_table, precinct, precinct_county_fips, epsg = 3857)

Arguments

Value

dataframe with one row per row of precinct, ordered to align with the input precinct object.

Examples

## Not run: 
# Need Census API
data(towns)
towns$fips <- '087'
block <- create_block_table('NY', 'Rockland')
block2prec_by_county(block, towns, 'fips')

## End(Not run)

Description

Identify contiguous sets of units and numbers each set. Can be extended to repeat the procedure within a subgeography.

Usage

check_contiguity(adj, group)

cct(adj, group)

ccm(adj, group)

Arguments

Details

Given a zero-indexed adjacency list and an array of group identifiers, this returns a tibble which identifies the connected components. The three columns are group for the inputted group, group_number which uniquely identifies each group as a positive integer, and component which identifies the connected component number for each corresponding entry of adjacency and group. If everything is connected within the group, then each element of component will be 1. Otherwise, the largest component is given the value 1, the next largest 2, and so on.

If nothing is provided to group, it will default to a vector of ones, checking if the adjacency graph is connected.

cct() is shorthand for creating a table of the component values. If everything is connected within each group, it returns a value of 1. In general, it returns a frequency table of components.

ccm() is shorthand for getting the maximum component value. It returns the maximum number of components that a group is broken into. This returns 1 if each group is connected. #'

Value

tibble with contiguity indicators. Each row is the units of adj. Columns include

• group Values of the inputted group argument. If group is not specified, then all values will be 1.

• group_number Integer encoding of group.

• component A number for each contiguous set of units within a group. If all units within a group are contiguous, all values are 1. If there are two sets, each discontiguous with the other, the larger one will be numbered 1 and the smaller one will be numbered 2.

Examples

data(checkerboard)
adj <- adjacency(checkerboard)
# These each indicate the graph is connected.
check_contiguity(adj) # all contiguous
# If there are two discontiguous groups, there will be 2 values of `component`
cct(adj)
ccm(adj)

Description

Cast shp to component polygons, build the adjacency, and check the contiguity. Avoids issues where a precinct is actually a multipolygon

Usage

check_polygon_contiguity(shp, group, epsg = 3857)

Arguments

Value

tibble in the same format as check_contiguity().

Examples

data(checkerboard)
check_polygon_contiguity(checkerboard, i)

Description

This data set contains 64 squares in an 8x8 grid, like a checkerboard.

Usage

data("checkerboard")

Format

An sf dataframe with 64 observations

Examples

data('checkerboard')

Description

This data contains a zero indexed adjacency list for the checkerboard dataset.

Usage

data("checkerboard_adj")

Format

A list with 64 entries

Examples

data('checkerboard_adj')

Description

Rename VEST election-result columns into a more uniform office_year_party_candidate style.

Usage

clean_vest(data)

Arguments

Value

input object with renamed columns

Examples

data(va18sub)
va <- clean_vest(va18sub)

Description

Compare two adjacency lists and report edges that differ between them.

Usage

compare_adjacencies(adj1, adj2, shp, zero = TRUE)

Arguments

Value

tibble of differing edge pairs. When shp is supplied, includes DE-9IM relation and simple intersection class diagnostics.

Examples

data(towns)
rook <- adjacency(towns)
sf_rook <- lapply(sf::st_relate(towns, pattern = 'F***1****'), function(x) {
  x - 1L
})
compare_adjacencies(rook, sf_rook, zero = FALSE)

Description

Summarize a district-membership matrix into pairwise co-assignment counts.

Usage

count_connections(dm, normalize = FALSE)

Arguments

Value

tibble in long form with columns x, y, and fill, where fill stores the count or proportion of shared assignments.

Examples

set.seed(1)
dm <- matrix(sample(1:2, size = 100, TRUE), 10)
count_connections(dm)

Description

Download block-level decennial census data with the standard redistricting variables used throughout the package.

Usage

create_block_table(
  state,
  county = NULL,
  geometry = TRUE,
  year = 2020,
  mem = FALSE,
  epsg = 3857
)

Arguments

Value

dataframe or sf object with one row per block. Includes population and voting-age-population summaries by race and ethnicity.

Examples

## Not run: 
# uses the Census API
create_block_table(state = 'NY', county = 'Rockland', geometry = FALSE)

## End(Not run)

Description

Download tract-level ACS data with the standard redistricting variables used throughout the package.

Usage

create_tract_table(
  state,
  county,
  geometry = TRUE,
  year = 2019,
  mem = FALSE,
  epsg = 3857
)

Arguments

Value

dataframe or sf object with one row per tract. Includes population, voting-age population, and citizen voting-age population summaries by race and ethnicity.

Examples

## Not run: 
# Relies on Census Bureau API
tract <- create_tract_table('NY', 'Rockland', year = 2018)

## End(Not run)

Description

Read a DRA block assignment export and attach it to 2020 census blocks, optionally collapsing the assignments to a precinct layer.

Usage

dra2r(dra, state, precincts, epsg = 3857)

Arguments

Value

sf object at the block level when precincts is omitted, or at the precinct level when precincts is supplied.

Examples

## Not run: 
# Needs Census Bureau API
# dra_utah_test is available at https://bit.ly/3c6UDKk
blocklevel <- dra2r('dra_utah_test.csv', state = 'UT')

## End(Not run)

Description

Non-geographic companion to geo_estimate_down(). Use this when you already have a vector of matches and want to avoid recomputing them.

Usage

estimate_down(wts, value, group)

Arguments

Value

numeric vector of length length(group) with values split by weight

Examples

library(dplyr)
set.seed(1)
data(checkerboard)
counties <- checkerboard |>
  group_by(id <= 32) |>
  summarize(geometry = sf::st_union(geometry)) |>
  mutate(pop = c(100, 200))
matches <- geo_match(checkerboard, counties)
estimate_down(wts = rep(1, nrow(checkerboard)), value = counties$pop, group = matches)

Description

Non-geographic companion to geo_estimate_up(). Use this when you already have a vector of group assignments.

Usage

estimate_up(value, group)

Arguments

Value

numeric vector of length max(group) with values aggregated by group

Examples

library(dplyr)
set.seed(1)
data(checkerboard)
counties <- checkerboard |>
  group_by(id <= 32) |>
  summarize(geometry = sf::st_union(geometry)) |>
  mutate(pop = c(100, 200))
matches <- geo_match(checkerboard, counties)
estimate_up(value = checkerboard$i, group = matches)

Description

Split values observed on a larger geography across a smaller geography after first matching each row of to to a row of from. This is commonly used to distribute precinct-level election totals to blocks.

Usage

geo_estimate_down(from, to, wts, value, method = "center", epsg = 3857)

Arguments

Details

If all weights for a matched group are zero, the function falls back to equal allocation within that group. Rows in to that do not match any row of from receive 0.

Value

numeric vector of length nrow(to) with estimated values

Examples

library(dplyr)
set.seed(1)
data(checkerboard)
counties <- checkerboard |>
  group_by(id <= 32) |>
  summarize(geometry = sf::st_union(geometry)) |>
  mutate(pop = c(100, 200))
geo_estimate_down(from = counties, to = checkerboard, value = counties$pop)

Description

Aggregate values from a smaller geography to a larger geography after matching each row of from to a row of to. This is commonly used to roll block-level counts up to precincts or districts.

Usage

geo_estimate_up(from, to, value, method = "center", epsg = 3857)

Arguments

Details

Groups in to with no matched rows are included in the output and receive 0.

Value

numeric vector of length nrow(to) with values aggregated by group

Examples

library(dplyr)
set.seed(1)
data(checkerboard)
counties <- checkerboard |>
  group_by(id <= 32) |>
  summarize(geometry = sf::st_union(geometry)) |>
  mutate(pop = c(100, 200))
geo_estimate_up(from = checkerboard, to = counties, value = checkerboard$i)

Description

Keep only the rows of from whose geometry intersects any geometry in to.

Usage

geo_filter(from, to, bool = FALSE, epsg = 3857)

Arguments

Value

subset of from, or a logical vector when bool = TRUE

Examples

## Not run: 
# Needs Census Bureau API
data(towns)
block <- create_block_table('NY', 'Rockland')
geo_filter(block, towns)

## End(Not run)

data(towns)
data(rockland)
sub <- geo_filter(rockland, towns)

Description

Match each row of from to one row of to, typically when from is a finer geography nested inside to. The result is an integer vector of row indices in to and can be reused by downstream helpers such as estimate_up(), estimate_down(), and block2prec().

Usage

geo_match(
  from,
  to,
  method = "center",
  by = NULL,
  tiebreaker = TRUE,
  epsg = 3857
)

Arguments

Details

Methods are as follows:

• "centroid": match using the geometric centroid of each row in from.

• "center": use the centroid when it lies inside the polygon and otherwise fall back to a point-on-surface.

• "point": use point on surface for each row in from.

• "circle": use the centroid of the maximum inscribed circle.

• "area": match to the row in to with the largest area overlap.

When the output does not reference every row of to, an attribute "matching_max" is attached and records nrow(to). Functions such as block2prec() use that attribute to preserve empty target groups.

Value

integer vector of length nrow(from). Values are row indices in to, or -1 / -2 when tiebreaker = FALSE. Empty geometries are returned as NA.

Examples

library(dplyr)
data(checkerboard)
counties <- sf::st_as_sf(as.data.frame(rbind(
  sf::st_union(checkerboard |> filter(i < 4)),
  sf::st_union(checkerboard |> filter(i >= 4))
)))

geo_match(from = checkerboard, to = counties)
geo_match(from = checkerboard, to = counties, method = 'area')

Description

Quick diagnostic plot that labels each row with its row number.

Usage

geo_plot(shp)

Arguments

Value

ggplot object

Examples

data(checkerboard)
geo_plot(checkerboard)

Description

Plot Groups with Connected Components Colored

Usage

geo_plot_group(shp, adj, group, save = FALSE, path = "")

Arguments

Value

list of ggplot objects, one per unique group

Examples

library(dplyr)
data('checkerboard')
data('checkerboard_adj')

checkerboard <- checkerboard |> mutate(discont = as.integer(j == 5 | j == 6))

p <- geo_plot_group(checkerboard, checkerboard_adj, checkerboard$discont)

p[[1]]
p[[2]]

Description

Reorder rows by the centroid distance from the northwest corner of the overall bounding box.

Usage

geo_sort(shp, epsg = 3857)

Arguments

Value

reordered sf dataframe

Examples

data(checkerboard)
geo_sort(checkerboard)

Description

Keep only rows of from whose intersection with to covers more than a given share of the row's area.

Usage

geo_trim(from, to, thresh = 0.01, bool = FALSE, epsg = 3857)

Arguments

Value

subset of from, or a logical vector when bool = TRUE

Examples

## Not run: 
# Needs Census Bureau API
data(towns)
block <- create_block_table('NY', 'Rockland')
geo_trim(block, towns, thresh = 0.05)

## End(Not run)

data(towns)
data(rockland)
sub <- geo_filter(rockland, towns)
rem <- geo_trim(sub, towns, thresh = 0.05)

Description

Returns points within the shape, near the center. Uses the centroid if that's in the shape, or point on surface if not.

Usage

geos_centerish(shp, epsg = 3857)

Arguments

Value

geos geometry vector of points.

Examples

data(towns)
geos_centerish(towns)

Description

Returns the centroid of the largest inscribed circle for each shape

Usage

geos_circle_center(shp, tolerance = 0.01, epsg = 3857)

Arguments

Value

geos geometry vector of points.

Examples

data(towns)
geos_circle_center(towns)

Description

Download a state dataset from the Algorithm-Assisted Redistricting Methodology Project. Depending on state and year, the data are returned at either the voting-district level or the census-block level. The current supported data is the 2020 and 2010 retabulations of the VEST data, which can be downloaded with get_vest().

Usage

get_alarm(state, year = 2020, geometry = TRUE, epsg = 3857)

Arguments

Details

See the full available data at https://github.com/alarm-redist/census-2020.

Value

tibble or sf object with ALARM election data and, optionally, geometry.

Examples

ak <- get_alarm('AK', geometry = FALSE)

Description

Download a state election dataset from the DRA public repository. Depending on state and year, the geometry may be joined at the VTD or block-group level.

Usage

get_dra(state, year = 2020, geometry = TRUE, clean_names = TRUE, epsg = 3857)

Arguments

Details

See the full available data at https://github.com/dra2020/vtd_data.

Value

tibble or sf object with DRA election data and, optionally, geometry.

Examples

ak <- get_dra('AK', geometry = FALSE)

Description

Download a state file from the Harvard Election Data Archive and return it as an sf object. Some states require special handling because the source files differ in format or level of geography.

Usage

get_heda(state, path = tempdir(), epsg = 3857, ...)

Arguments

Details

Most states are distributed as zipped shapefiles. California is stored differently and is aggregated here to 2010 tracts. Some states are shipped as loose shapefile components rather than zip archives. If the source file has no CRS metadata, the function assumes EPSG 4140 before optionally reprojecting.

The function currently returns the raw source column names because internal name cleaning is disabled in the implementation.

Value

sf tibble containing the requested HEDA dataset

Examples

shp <- get_heda('ND')

Description

Data sourced from the United States Congressional District Shapefiles, primarily hosted at https://cdmaps.polisci.ucla.edu/. Files are fetched through the GitHub repository at https://github.com/JeffreyBLewis/congressional-district-boundaries.

Usage

get_lewis(state, congress, path_only = FALSE)

Arguments

Value

sf tibble of congressional district boundaries, or a character URL when path_only = TRUE.

References

Jeffrey B. Lewis, Brandon DeVine, Lincoln Pitcher, and Kenneth C. Martis. (2013) Digital Boundary Definitions of United States Congressional Districts, 1789-2012. [Data file and code book]. Retrieved from https://cdmaps.polisci.ucla.edu on [date of download].

Examples

path <- get_lewis(state = 'NM', congress = 111, path_only = TRUE)
if (attr(curlGetHeaders(path), 'status') == 200) {
  get_lewis(state = 'NM', congress = 111)
}

Description

Download precinct-level racially polarized voting estimates from the RPV Near Me repository.

Usage

get_rpvnearme(state, version = c(1, 2))

Arguments

Value

tibble of precinct-level estimates of party vote by race

Examples

get_rpvnearme('DE')

Description

Download a state-year VEST precinct file from Harvard Dataverse.

Usage

get_vest(state, year, path = tempdir(), clean_names = TRUE, epsg = 3857, ...)

Arguments

Value

sf tibble containing the requested VEST dataset

Examples

## Not run: 
# Requires Dataverse API
shp <- get_vest('CO', 2020)

## End(Not run)

Description

Computes the Global Geary's Contiguity statistic. Can produce spatial weights from an adjacency or sf dataframe, in which case the spatial_mat is a contiguity matrix. Users can also provide a spatial_mat argument directly.

Usage

global_gearys(shp, adj, wts, spatial_mat, epsg = 3857)

Arguments

Value

numeric scalar containing the global Geary's C statistic

Examples

library(dplyr)
data('checkerboard')
checkerboard <- checkerboard |> mutate(m = as.numeric((id + i) %% 2 == 0))
global_gearys(shp = checkerboard, wts = checkerboard$m)

Description

Computes the Global Moran's I statistic and expectation. Can produce spatial weights from an adjacency or sf dataframe, in which case the spatial_mat is a contiguity matrix. Users can also provide a spatial_mat argument directly.

Usage

global_morans(shp, adj, wts, spatial_mat, epsg = 3857)

Arguments

Value

named list returned by the underlying C++ implementation, including the global statistic and its moments.

Examples

library(dplyr)
data('checkerboard')
checkerboard <- checkerboard |> mutate(m = as.numeric((id + i) %% 2 == 0))
global_morans(shp = checkerboard, wts = checkerboard$m)

Description

Returns standardized local Getis-Ord G*i scores using either an sf object, a zero-indexed adjacency list, or a spatial weights matrix.

Usage

gstar_i(shp, adj, wts, spatial_mat, epsg = 3857)

Arguments

Value

numeric vector of standardized G*i scores, one per observation

Examples

library(dplyr)
data('checkerboard')
checkerboard <- checkerboard |> mutate(m = as.numeric((id + i) %% 2 == 0))
gstar_i(shp = checkerboard, wts = checkerboard$m)

Description

List Available HEDA States

Usage

heda_states()

Value

character vector of lowercase state abbreviations with packaged HEDA support.

Examples

heda_states()

Description

Compute Local Geary's C for each observation using either an sf object, a zero-indexed adjacency list, or a spatial weights matrix.

Usage

local_gearys(shp, adj, wts, spatial_mat, epsg = 3857)

Arguments

Value

tibble with one column, geary, and one row per observation

Examples

library(dplyr)
data('checkerboard')
checkerboard <- checkerboard |> mutate(m = as.numeric((id + i) %% 2 == 0))
local_gearys(shp = checkerboard, wts = checkerboard$m)

Description

Compute Local Moran's I for each observation using either an sf object, a zero-indexed adjacency list, or a spatial weights matrix.

Usage

local_morans(shp, adj, wts, spatial_mat, epsg = 3857)

Arguments

Value

tibble with columns moran, expectation, and variance

Examples

library(dplyr)
data('checkerboard')
checkerboard <- checkerboard |> mutate(m = as.numeric((id + i) %% 2 == 0))
local_morans(shp = checkerboard, wts = checkerboard$m)

Description

The data contains the North Rockland Central School District.

Usage

data('nrcsd')

Format

An sf dataframe with 1 observation

Examples

data('nrcsd')

Description

This data contains the blocks for Orange County NY, with geographies simplified to allow for better examples.

Usage

data("orange")

Format

An sf dataframe with 10034 observations

Details

It can be recreated with: orange <- create_block_table('NY', 'Orange') orange <- rmapshaper::ms_simplify(orange, keep_shapes = TRUE)

Examples

data('orange')

Description

This data contains the election districts (or precincts) for Rockland County NY, with geographies simplified to allow for better examples.

Usage

data("precincts")

Format

An sf dataframe with 278 observations

References

https://www.rocklandgis.com/portal/apps/sites/#/data/datasets/2d91f9db816c48318848ad66eb1a18e9

Examples

data('precincts')

Description

Project a plan defined on precincts down to 2020 census blocks and return the two-column format expected by DRA exports.

Usage

r2dra(precincts, plan, state, path, epsg = 3857)

Arguments

Value

tibble with columns GEOID20 and District, suitable for writing to a DRA-style CSV.

Examples

## Not run: 
# Needs Census Bureau API
cd <- tinytiger::tt_congressional_districts() |> filter(STATEFP == '49')
cnty <- tinytiger::tt_counties(state = 49)
matchedcty <- geo_match(from = cnty, to = cd)
# use counties as precincts and let the plan be their center match:
r2dra(cnty, matchedcty, 'UT', 'r2dra_ex.csv')

## End(Not run)

Description

Divide an adjacency graph into regions by removing edges whose centerlines intersect a set of line features, such as roads or rivers.

Usage

regionalize(shp, lines, adj = adjacency(shp), epsg = 3857)

Arguments

Value

integer vector of region labels with nrow(shp) entries

Examples

data(towns)
# make some weird roadlike feature passing through the towns
lines <- sf::st_sfc(sf::st_linestring(sf::st_coordinates(sf::st_centroid(towns))),
  crs = sf::st_crs(towns)
)
regionalize(towns, lines)

Description

This data contains the blocks for Rockland County NY, with geographies simplified to allow for better examples.

Usage

data("rockland")

Format

An sf dataframe with 4764 observations

Details

It can be recreated with: rockland <- create_block_table('NY', 'Rockland') rockland <- rmapshaper::ms_simplify(rockland, keep_shapes = TRUE)

Examples

data('rockland')

Description

Keep only the adjacency edges that connect the two sides of a selected seam.

Usage

seam_adj(adj, shp, admin, seam, epsg = 3857)

Arguments

Value

adjacency list containing only cross-seam edges

Examples

data('rockland')
data('orange')
data('nrcsd')

o_and_r <- rbind(orange, rockland)
o_and_r <- o_and_r |>
  geo_filter(nrcsd) |>
  geo_trim(nrcsd)
adj <- adjacency(o_and_r)

seam_adj(adj, shp = o_and_r, admin = 'county', seam = c('071', '087'))

Description

Keep only rows in shp that lie on the border between two administrative units and have at least one adjacency connection across that border.

Usage

seam_geom(adj, shp, admin, seam, epsg = 3857)

Arguments

Value

subset of shp containing only rows on the selected seam

Examples

data('rockland')
data('orange')
data('nrcsd')

o_and_r <- rbind(orange, rockland)
o_and_r <- o_and_r |>
  geo_filter(nrcsd) |>
  geo_trim(nrcsd)
adj <- adjacency(o_and_r)

seam_geom(adj, shp = o_and_r, admin = 'county', seam = c('071', '087'))

Description

Remove adjacency connections between two or more administrative units, such as counties on opposite sides of a seam.

Usage

seam_rip(adj, shp, admin, seam, epsg = 3857)

Arguments

Value

adjacency list with all cross-seam edges removed

Examples

data('rockland')
data('orange')
data('nrcsd')

o_and_r <- rbind(orange, rockland)
o_and_r <- o_and_r |>
  geo_filter(nrcsd) |>
  geo_trim(nrcsd)
adj <- adjacency(o_and_r)

seam_rip(adj, o_and_r, 'county', c('071', '087'))

Description

Build a set of candidate cross-seam edges by constructing Voronoi cells from representative points and identifying neighboring cells across the seam.

Usage

seam_sew(shp, admin, seam, epsg = 3857)

Arguments

Value

tibble with columns v1 and v2, suitable for use with add_edge()

Examples

data('rockland')
data('orange')
data('nrcsd')

o_and_r <- rbind(orange, rockland)
o_and_r <- o_and_r |>
  geo_filter(nrcsd) |>
  geo_trim(nrcsd)
adj <- adjacency(o_and_r)

adds <- seam_sew(o_and_r, 'county', c('071', '087'))
adj <- adj |> add_edge(adds$v1, adds$v2)

Description

States often split a precinct when they create districts but rarely provide the geography for the split precinct. This allows you to split a precinct using a lower geography, typically blocks.

Usage

split_precinct(lower, precinct, split_by, lower_wt, split_by_id, epsg = 3857)

Arguments

Details

The function first filters lower and split_by to the selected precinct, matches lower-level units to split_by, and unions matched pieces. When lower_wt is supplied, a wt column is added with summed weights.

Value

sf dataframe with one row per observed split piece

Examples

library(sf)
data(checkerboard)
low <- checkerboard |> dplyr::slice(1:3, 9:11)
prec <- checkerboard |>
  dplyr::slice(1:3) |>
  dplyr::summarize(geometry = sf::st_union(geometry))
dists <- checkerboard |>
  dplyr::slice(1:3, 9:11) |>
  dplyr::mutate(dist = c(1, 2, 2, 1, 3, 3)) |>
  dplyr::group_by(dist) |>
  dplyr::summarize(geometry = sf::st_union(geometry))

split_precinct(low, prec, dists, split_by_id = 'dist')

Description

Returns points within the shape, near the center. Uses the centroid if that's in the shape, or point on surface if not.

Usage

st_centerish(shp, epsg = 3857)

Arguments

Value

sf dataframe with the same rows and attributes as shp, but point geometries.

Examples

data(towns)
st_centerish(towns)

Description

Compute the centroid of the largest inscribed circle for each feature.

Usage

st_circle_center(shp, tolerance = 0.01, epsg = 3857)

Arguments

Value

sf dataframe with the same rows and attributes as shp, but point geometries.

Examples

data(towns)
st_circle_center(towns)

Description

Remove one or more undirected edges from an adjacency list and return the modified list.

Usage

subtract_edge(adj, v1, v2, ids = NULL, zero = TRUE)

Arguments

Value

Adjacency list with the requested edges removed symmetrically.

Examples

data(towns)
adj <- adjacency(towns)

subtract_edge(adj, 2, 3)
subtract_edge(adj, "West Haverstraw", "Stony Point", towns$MUNI)

Description

Suggest nearest cross-component links that could reconnect disconnected groups.

Usage

suggest_component_connection(shp, adj, group, epsg = 3857)

Arguments

Value

tibble with columns x and y, giving candidate row pairs to connect with add_edge().

Examples

library(dplyr)
data(checkerboard)
checkerboard <- checkerboard |> filter(i != 1, j != 1)
adj <- adjacency(checkerboard)
suggest_component_connection(checkerboard, adj)

Description

For precincts which have no adjacent precincts, this suggests the nearest precinct as a friend to add. This is useful for when a small number of precincts are disconnected from the remainder of the geography, such as an island.

Usage

suggest_neighbors(shp, adj, idx, neighbors = 1)

Arguments

Value

tibble with columns x and y, giving pairs of row indices that could be connected with add_edge().

Examples

library(dplyr)
data(va18sub)
va18sub <- va18sub |> filter(!VTDST %in% c('000516', '000510', '000505', '000518'))
adj <- adjacency(va18sub)
suggests <- suggest_neighbors(va18sub, adj)
adj <- adj |> add_edge(v1 = suggests$x, v2 = suggests$y)

Description

This data contains 7 town boundaries for the towns which overlap North Rockland School District in NY.

Usage

data("towns")

Format

An sf dataframe with 7 observations

References

https://www.rocklandgis.com/portal/apps/sites/#/data/items/746ec7870a0b4f46b168e07369e79a27

Examples

data('towns')

Description

This data contains the blocks Henrico County, VA with geographies simplified to allow for better examples.

Usage

data("va_blocks")

Format

An sf dataframe with 6354 observations

Details

blocks87 <- create_block_table(state = 'VA', county = '087') va_blocks <- rmapshaper::ms_simplify(va_blocks, keep_shapes = TRUE)

Examples

data('va_blocks')

Description

This data contains the blocks for Henrico County, VA with geographies simplified to allow for better examples.

Usage

data("va_blocks")

Format

An sf dataframe with 93 observations

Details

va_vtd <- tinytiger::tt_voting_districts(state = 'VA', county = '087', year = 2010) va_vtd <- rmapshaper::ms_simplify(va_vtd, keep_shapes = TRUE)

Examples

data('va_blocks')

Description

This data contains a 90 precinct subset of Virginia from the 2018 Senate race. Contains results for Henrico County

Usage

data("va18sub")

Format

An sf dataframe with 90 observations

References

Voting and Election Science Team, 2019, "va_2018.zip", 2 018 Precinct-Level Election Results, https://doi.org/10.7910/DVN/UBKYRU/FQDLOO, Harvard Dataverse, V4

Examples

data('va18sub')

Description

List Available VEST States

Usage

vest_states(year)

Arguments

Value

character vector of state abbreviations available for that year

Examples

## Not run: 
# Requires Dataverse API
vest_states(2020)

## End(Not run)