README

arcgisgeocode provides access to ArcGIS geocoding services from R. It supports address candidate identification, batch geocoding, reverse geocoding, and autocomplete suggestions.

Installation

# install from CRAN 
install.packages("arcgisgeocode")

You can also install the development version from r-universe as a binary for Mac, Windows, or Ubuntu from r-universe like so:

# install from R-universe
install.packages("arcgisgeocode", repos = "https://r-arcgis.r-universe.dev")

Or you can install the package from source which requires Rust to be available. Follow the rustup instructions to install Rust and verify your installation is compatible using rextendr::rust_sitrep(). Then install the development version from GitHub:

# install pak if not available
if (!requireNamespace("pak")) install.packages("pak")

# install development version of {arcgisgeocode}
pak::pak("r-arcgis/arcgisgeocode")

Usage

By default, the ArcGIS World Geocoder will be used. This geocoding server provides public access to the /findAddressCandidates, /reverseGeocode, and /suggest endpoints made available via the find_address_candidates(), reverse_geocode(), and suggest_places() functions respectively.

The batch geocoding endpoint /geocodeAddresses is available via geocode_addresses(). However, this requires the use of an authorization token and may consume credits.

Refer to the ArcGIS World Geocoder official documentation for additional information on use restrictions and licensing. For example, a valid token is required to store the results of geocoding transactions.

Reverse geocoding

library(arcgisgeocode)

# Find addresses from locations
rev_res <- reverse_geocode(c(-117.172, 34.052))

# preview results
dplyr::glimpse(rev_res)
#> Rows: 1
#> Columns: 23
#> $ match_addr   <chr> "600-620 Home Pl, Redlands, California, 92374"
#> $ long_label   <chr> "600-620 Home Pl, Redlands, CA, 92374, USA"
#> $ short_label  <chr> "600-620 Home Pl"
#> $ addr_type    <chr> "StreetAddress"
#> $ type_field   <chr> ""
#> $ place_name   <chr> ""
#> $ add_num      <chr> "608"
#> $ address      <chr> "608 Home Pl"
#> $ block        <chr> ""
#> $ sector       <chr> ""
#> $ neighborhood <chr> "South Redlands"
#> $ district     <chr> ""
#> $ city         <chr> "Redlands"
#> $ metro_area   <chr> ""
#> $ subregion    <chr> "San Bernardino County"
#> $ region       <chr> "California"
#> $ region_abbr  <chr> "CA"
#> $ territory    <chr> ""
#> $ postal       <chr> "92374"
#> $ postal_ext   <chr> ""
#> $ country_name <chr> "United States"
#> $ country_code <chr> "USA"
#> $ geometry     <POINT [°]> POINT (-117.172 34.05204)

Address search

The find_address_candidates() function returns geocoding candidate results. The function is vectorized over the input and will perform multiple requests in parallel. Each request geocodes one location at a time.

One or more candidates are returned from the endpoint. You can limit the number of candidates using the max_locations argument (with a maximum of 50).

# Find addresses from address search
candidates <- find_address_candidates(
  address = "esri",
  address2 = "380 new york street",
  city = "redlands",
  country_code = "usa",
  max_locations = 2
)

dplyr::glimpse(candidates[, 1:10])
#> Rows: 2
#> Columns: 11
#> $ input_id    <int> 1, 1
#> $ result_id   <int> NA, NA
#> $ loc_name    <chr> "World", "World"
#> $ status      <chr> "M", "M"
#> $ score       <dbl> 100.00, 98.57
#> $ match_addr  <chr> "Esri", "380 New York St, Redlands, California, 92373"
#> $ long_label  <chr> "Esri, 380 New York St, Redlands, CA, 92373, USA", "380 Ne…
#> $ short_label <chr> "Esri", "380 New York St"
#> $ addr_type   <chr> "POI", "PointAddress"
#> $ type_field  <chr> "Business Facility", NA
#> $ geometry    <POINT [°]> POINT (-117.1957 34.05609), POINT (-117.1948 34.05726)…

Suggest locations

Geocoding services can also provide a location suggestion based on a search term and, optionally, a location or extent. The suggest_places() function (/suggest endpoint) is intended to be used as part of a client-facing application that provides autocomplete suggestions.

In this example we create a search extent around a single point and find suggestions based on the search term "bellwood".

# identify a search point as a simple feature column
location <- sf::st_sfc(
  sf::st_point(c(-84.34, 33.74)),
  crs = 4326
)

# buffer and create a bbox object to search within the extent
search_extent <- sf::st_bbox(
  sf::st_buffer(location, 10)
)

# find suggestions within the bounding box
suggestions <- suggest_places(
  "bellwood",
  location,
  search_extent = search_extent
)

suggestions
#> # A data frame: 5 × 3
#>   text                                                   magic_key is_collection
#> * <chr>                                                  <chr>     <lgl>        
#> 1 Bellwood Coffee, 1366 Glenwood Ave SE, Atlanta, GA, 3… dHA9MCN0… FALSE        
#> 2 Bellwood, Atlanta, GA, USA                             dHA9MCN0… FALSE        
#> 3 Bellwood Church, Atlanta, GA, USA                      dHA9MCN0… FALSE        
#> 4 Bellwood Yard, Atlanta, GA, USA                        dHA9MCN0… FALSE        
#> 5 Bellwood, IL, USA                                      dHA9NCN0… FALSE

The result is intended to be provided to find_address_candidates() to complete the geocoding process. The column text contains the address to geocode. The column magic_key is a special identifier that makes it much faster to fetch results. Pass this into the argument magic_key.

# get address candidate information
# using the text and the magic key
res <- find_address_candidates(
  suggestions$text,
  magic_key = suggestions$magic_key
)

dplyr::glimpse(res[, 1:10])
#> Rows: 7
#> Columns: 11
#> $ input_id    <int> 1, 2, 3, 4, 5, 5, 5
#> $ result_id   <int> NA, NA, NA, NA, NA, NA, NA
#> $ loc_name    <chr> NA, NA, NA, NA, NA, NA, NA
#> $ status      <chr> "M", "M", "M", "M", "T", "T", "T"
#> $ score       <dbl> 100, 100, 100, 100, 100, 100, 100
#> $ match_addr  <chr> "Bellwood Coffee", "Bellwood, Atlanta, Georgia", "Bellwood…
#> $ long_label  <chr> "Bellwood Coffee, 1366 Glenwood Ave SE, Atlanta, GA, 30316…
#> $ short_label <chr> "Bellwood Coffee", "Bellwood", "Bellwood Church", "Bellwoo…
#> $ addr_type   <chr> "POI", "Locality", "POI", "POI", "Locality", "Locality", "…
#> $ type_field  <chr> "Snacks", "City", "Church", "Building", "City", "City", "C…
#> $ geometry    <POINT [°]> POINT (-84.34273 33.74034), POINT (-84.41243 33.77455), PO…

Important: Storing results

By default, the argument for_storage = FALSE meaning that the results of the geocoding operation cannot be persisted. If you intend to persist the results of the geocoding operation, you must set for_storage = TRUE.

Batch geocoding

Many addresses can be geocoded very quickly using the geocode_addresses() function which calls the /geocodeAddresses endpoint. Note that this function requires an authorization token. geocode_addresses() sends the input addresses in chunks as parallel requests.

library(arcgisutils)
library(arcgisgeocode)

set_arc_token(auth_user())

# Example dataset from the Urban Institute
fp <- "https://urban-data-catalog.s3.amazonaws.com/drupal-root-live/2020/02/25/geocoding_test_data.csv"

to_geocode <- readr::read_csv(fp, readr::locale(encoding = "latin1"))

geocoded <- to_geocode |>
  dplyr::reframe(
    geocode_addresses(
      address = address,
      city = city,
      region = state,
      postal = zip
    )
  )

dplyr::glimpse(res[, 1:10])

Using other locators

{arcgisgeocode} can be used with other geocoding services, including custom locators hosted on ArcGIS Online or Enterprise. For example, we can use the AddressNC geocoding service available on ArcGIS Online.

Create a new GeocodeServer object using geocode_server(). This geocoder can be passed into the geocoder argument to any of the geocoding functions.

arcgisgeocode