Yuxiao Luo 2023-02-15
What is the meaning of geocoding? Geocoding is the process of transforming a description of a location—such as a pair of coordinates, an address, or a name of a place—to a location on the earth’s surface. You can geocode by entering one location description at a time or by providing many of them at once in a table.
Reverse geocoding is a way to convert latitude and longitude coordinates to addresses. It’s the opposite of geocoding which lets you find latitude and longitude by address. All kinds of companies use reverse geocoding to find accurate address data for things like delivery updates, location-based advertising, or as part of an address verification solution.
We can use thetidygeocoder package in R to do both work.
First, let’s install the package and load it into R environment.
# install the package if you don't have it
# install.packages('tidygeocoder')
library(tidygeocoder)
library(tidyverse)Create a dataframe with addresses
addresses <- tibble::tribble(
~name, ~addr,
"White House", "1600 Pennsylvania Ave NW, Washington, DC",
"Transamerica Pyramid", "600 Montgomery St, San Francisco, CA 94111",
"Willis Tower", "233 S Wacker Dr, Chicago, IL 60606"
)This is the look of the addresses data
frame/tibble.
addresses## # A tibble: 3 × 2
## name addr
## <chr> <chr>
## 1 White House 1600 Pennsylvania Ave NW, Washington, DC
## 2 Transamerica Pyramid 600 Montgomery St, San Francisco, CA 94111
## 3 Willis Tower 233 S Wacker Dr, Chicago, IL 60606
Geocode the addresses using geocode function. The returned message
from the API query contains many columns, including class of the
address, type of the address, license information, and coordinates.
Let’s extract the latitudes and longitude in the return message with
lat = latitude and long = longtitude.
# geocode the addresses
lat_longs <- addresses %>%
geocode(addr, method = 'osm', lat = latitude , long = longitude)
# this result only returns latitudes and longitudes
lat_longs## # A tibble: 3 × 4
## name addr latit…¹ longi…²
## <chr> <chr> <dbl> <dbl>
## 1 White House 1600 Pennsylvania Ave NW, Washington, DC 38.9 -77.0
## 2 Transamerica Pyramid 600 Montgomery St, San Francisco, CA 941… 37.8 -122.
## 3 Willis Tower 233 S Wacker Dr, Chicago, IL 60606 41.9 -87.6
## # … with abbreviated variable names ¹latitude, ²longitude
If you need more information from the address that is provided from the
API serivce, you can use the argument full_results = TRUE. Notice that
the address details may vary from different API, which is the
method = 'osm' argument you specify in the function.
full_result <- addresses %>% geocode(addr, method = 'osm', full_results = TRUE)
# overview of the returned full results
glimpse(full_result)## Rows: 3
## Columns: 14
## $ name <chr> "White House", "Transamerica Pyramid", "Willis Tower"
## $ addr <chr> "1600 Pennsylvania Ave NW, Washington, DC", "600 Montgome…
## $ lat <dbl> 38.89770, 37.79520, 41.87887
## $ long <dbl> -77.03655, -122.40279, -87.63591
## $ place_id <int> 164605957, 110070947, 119999324
## $ licence <chr> "Data © OpenStreetMap contributors, ODbL 1.0. https://osm…
## $ osm_type <chr> "way", "way", "way"
## $ osm_id <int> 238241022, 24222973, 58528804
## $ boundingbox <list> <"38.8974908", "38.897911", "-77.0368537", "-77.0362519">…
## $ display_name <chr> "White House, 1600, Pennsylvania Avenue Northwest, Washin…
## $ class <chr> "office", "tourism", "building"
## $ type <chr> "government", "attraction", "commercial"
## $ importance <dbl> 1.054721, 1.139861, 1.224511
## $ icon <chr> NA, "https://nominatim.openstreetmap.org/ui/mapicons/poi_…
In the returned full results (14 columns), you can see extra address information other than the coordinates.
There are many geocoding services you can choose. Here is a full
list
of the services offered in tidygeocoder. I tested with osm
(OpenStreetMap provided by Nominatim API), ArcGIS (ArcGIS) and Bing
(Microsoft Bing Map), and here is my feedback:
-
osmcan return complete and reliable result, the speed limit is 1 query per second. It doesn’t need API key. -
ArcGIScan return the primary results in a fast and more efficient manner, but some information might be missing, ex., postal code. However, it doesn’t have any usage limitation as far as I know. It doesn’t require an API key either. -
Bingcan return complete and reliable result with a nice speed. It needs registration and requires API key. It offers a free quota for trial usage (30000 queries). However, if you are working on a research project in a university or use the API queries for educational purpose, you can apply for education license, which will give you 30000 queries per day. You can find the instruction and email address in the Bing Developer Portal for applying to eudcational license after you register with the link above.
To do reverse geocoding, we need geolocation coordinates (latitude and longitude). And then we can find the address and assoicated details of that address.
I downloaded the crime reports in NYC from the NYC Open Data webiste, and subset 5 observations from the large dataset. I anonymized the data by replacing the original complaint ID with a randomly generated ID for each report and removing the dates.
First, let’s load the data into R environment. We can load it from the RUG GitHub folder.
address <- read_csv('https://raw.githubusercontent.com/YuxiaoLuo/RUG-RUserGroup/main/data/address.csv')Let’s glimpse the data. OFNS_DESC is the offense description. PD_DESC is the police department standard description of the crime. We will use the Latitude and Longitude to obtain the detailed address.
glimpse(address)## Rows: 5
## Columns: 5
## $ id <dbl> 478325, 478326, 478327, 478328, 478329
## $ OFNS_DESC <chr> "GRAND LARCENY", "HARRASSMENT 2", "PETIT LARCENY", "CRIMINAL…
## $ PD_DESC <chr> "LARCENY,GRAND BY IDENTITY THEFT-UNCLASSIFIED", "HARASSMENT,…
## $ Latitude <dbl> 40.73683, 40.66779, 40.74126, 40.71443, 40.72780
## $ Longitude <dbl> -73.98875, -73.90587, -73.81912, -73.91424, -73.97923
Reverse geocoding uses reverse_geocode function. You need to specify
the latitude and longitude columns for the arguments lat and long.
And specify the which API service you will use in method argument,
here I’m using osm.
address_osm <- reverse_geocode(address,
lat = Latitude,
long = Longitude,
method = 'osm')
glimpse(address_osm)## Rows: 5
## Columns: 6
## $ id <dbl> 478325, 478326, 478327, 478328, 478329
## $ OFNS_DESC <chr> "GRAND LARCENY", "HARRASSMENT 2", "PETIT LARCENY", "CRIMINAL…
## $ PD_DESC <chr> "LARCENY,GRAND BY IDENTITY THEFT-UNCLASSIFIED", "HARASSMENT,…
## $ Latitude <dbl> 40.73683, 40.66779, 40.74126, 40.71443, 40.72780
## $ Longitude <dbl> -73.98875, -73.90587, -73.81912, -73.91424, -73.97923
## $ address <chr> "213, Park Avenue South, Manhattan Community Board 5, Manhat…
The result will return the address associated with the coordinates. If
you want more details with the address. You can specify
full_result = TRUE.
address_osm_full <- reverse_geocode(address, Latitude, Longitude, method = 'osm', full_results = TRUE)
glimpse(address_osm_full)## Rows: 5
## Columns: 24
## $ id <dbl> 478325, 478326, 478327, 478328, 478329
## $ OFNS_DESC <chr> "GRAND LARCENY", "HARRASSMENT 2", "PETIT LARCENY", "C…
## $ PD_DESC <chr> "LARCENY,GRAND BY IDENTITY THEFT-UNCLASSIFIED", "HARA…
## $ Latitude <dbl> 40.73683, 40.66779, 40.74126, 40.71443, 40.72780
## $ Longitude <dbl> -73.98875, -73.90587, -73.81912, -73.91424, -73.97923
## $ address <chr> "213, Park Avenue South, Manhattan Community Board 5,…
## $ place_id <int> 168178290, 167543808, 174392047, 167576877, 168170242
## $ licence <chr> "Data © OpenStreetMap contributors, ODbL 1.0. https:/…
## $ osm_type <chr> "way", "way", "way", "way", "way"
## $ osm_id <int> 249657335, 249544986, 284492649, 250451993, 249657796
## $ osm_lat <chr> "40.7367253", "40.667945200000005", "40.7412718", "40…
## $ osm_lon <chr> "-73.98848375618527", "-73.9059354", "-73.81881304999…
## $ house_number <chr> "213", "363", "59-21", "52-20", "186"
## $ road <chr> "Park Avenue South", "Sutter Avenue", "150th Street",…
## $ neighbourhood <chr> "Manhattan Community Board 5", "Brownsville", NA, "Ri…
## $ suburb <chr> "Manhattan", "Brooklyn", "Queens", "Queens", "Manhatt…
## $ county <chr> "New York County", NA, NA, NA, "New York County"
## $ city <chr> "City of New York", "City of New York", "City of New …
## $ state <chr> "New York", "New York", "New York", "New York", "New …
## $ `ISO3166-2-lvl4` <chr> "US-NY", "US-NY", "US-NY", "US-NY", "US-NY"
## $ postcode <chr> "10003", "11212", "11355", "11378", "10009"
## $ country <chr> "United States", "United States", "United States", "U…
## $ country_code <chr> "us", "us", "us", "us", "us"
## $ boundingbox <list> <"40.7366331", "40.7368174", "-73.9886512", "-73.9883…
I can extract whatever information I need in the full results. For example, I need the county, city, and postcode of the coordinates.
address_osm_full_subset <- address_osm_full %>% select(id, county, city, postcode)
address_osm_full_subset ## # A tibble: 5 × 4
## id county city postcode
## <dbl> <chr> <chr> <chr>
## 1 478325 New York County City of New York 10003
## 2 478326 <NA> City of New York 11212
## 3 478327 <NA> City of New York 11355
## 4 478328 <NA> City of New York 11378
## 5 478329 New York County City of New York 10009
If you have a large amount of geocoding queries, I would recommend to
use any of service that requires API key since it’s providing a stable
and fast service. Assume you’ve registered any of the service and
obtained an API key. You can set up the API key in Sys.setenv() by
replacing ***Your API key*** with your own unique key (a series of
digits and letters). Please keep your API key in a secret and a safe
place.
Sys.setenv(HERE_API_KEY = '***Your API key***')And then when you run geocode or reverse_geocode functions, choose
the corresponding API service method. For example, I’m assuming you are
using Google API service, then you should use method = 'google'.
geocode(some_addresses, address = addr, method = 'google', lat = latitude, long = longitude)That’s it. I hope you enjoyed reading this tutorial for Geocoding in R.