Introduction to Tidygeocoder

The tidygeocoder package provides an intuitive tidyverse-style interface for geocoder services. Currently the US Census and Nominatim (OSM) services are supported. The US Census service requires a street level address that is located in the United States. The OSM service does not have these restrictions, but has usage limits that will cause you to be locked out of the service temporarily. Because of these usage limits the default service for the geocode() function is the US Census which we will use to geocode a few street addresses below.

library(dplyr)
library(tidygeocoder)
library(knitr)

Geocode the addresses in our ‘sample_addresses’ dataset:

lat_longs <- sample_addresses %>% 
  geocode(addr,lat=latitude,long=longitude)

Latitude and longitude columns are attached to our input dataset:

kable(lat_longs)

Note that non-US addresses and non-street addresses were not found since we are using the US Census geocoder service.

if ((require("ggplot2") & require("maps") & require("ggrepel"))) {

ggplot(lat_longs %>% filter(!is.na(longitude)),aes(longitude,latitude),color="grey98") +
  borders("state") +
  theme_classic() +
  geom_point() +
  theme(line = element_blank(),
        text = element_blank(),
        title = element_blank()) +
  geom_label_repel(aes(label =name),show.legend=F) +
  scale_x_continuous(breaks = NULL) + 
  scale_y_continuous(breaks = NULL)
}

#ggsave("us_map.png",width=8,height=5)

To find non-US addresses and non-street addresses we can use the OSM service. The ‘cascade’ method will first attempt to use the US Census method for each address and only use the OSM service if the Census method fails (the Census service is tried first since the OSM service has usage limits).

cascade_points <- sample_addresses %>% 
  geocode(addr,method='cascade')

kable(cascade_points)