47 Use API-wrapping packages

This section explores how to use R packages that wrap APIs, allowing easy access to web data. We’ll cover various approaches from simple downloads to more complex API interactions. This content builds on Jenny Bryan’s stat545 materials, with significant updates and additions to reflect current best practices. For a more in-depth look at APIs, check out the API chapter in the R for Data Science book. For a huge list of free apis to play with, check out this list

47.1 The Data Acquisition Spectrum

When it comes to obtaining data from the internet, we can categorize methods into four main type * Direct Download: Grabbing readily available flat files (CSV, XLS, etc.) * Wrapped API Access: Using R packages designed for specific APIs * Raw API Interaction: Crafting custom queries for APIs * Web Scraping: Extracting data embedded in HTML structures

We’ll focus primarily on the second method, but now you know about the full spectrum of options at your disposal. For a comprehensive list of R tools for interacting with the internet, check out the rOpenSci repository. These tools include packages for APIs, scraping, and more.

47.2 Direct Download

In the simplest case, the data you need is already on the internet in a tabular format. Effectively, you just need to click and download whatever data you need. There are a couple of strategies here:

  • Use read.csv or readr::read_csv to read the data straight into R.
  • Use the command line program curl to do that work, and place it in a Makefile or shell script (see the section on make for more on this).

The second case is most useful when the data you want has been provided in a format that needs cleanup. For example, the World Value Survey makes several datasets available as Excel sheets. The safest option here is to download the .xls file, then read it into R with readxl::read_excel() or something similar. An exception to this is data provided as Google Spreadsheets, which can be read straight into R using the googlesheets package.

47.2.1 From rOpenSci web services page

From rOpenSci’s CRAN Task View: Web Technologies and Services:

  • downloader::download() for SSL.
  • curl::curl() for SSL.
  • httr::GET data read this way needs to be parsed later with read.table().
  • rio::import() can “read a number of common data formats directly from an https:// URL”. Isn’t that very similar to the previous?

What about packages that install data?

47.3 Data supplied on the web

Many times, the data that you want is not already organized into one or a few tables that you can read directly into R. More frequently, you find this data is given in the form of an API. Application Programming Interfaces (APIs) are descriptions of the kind of requests that can be made of a certain piece of software, and descriptions of the kind of answers that are returned.

Many sources of data – databases, websites, services – have made all (or part) of their data available via APIs over the internet. Computer programs (“clients”) can make requests of the server, and the server will respond by sending data (or an error message). This client can be many kinds of other programs or websites, including R running from your laptop.

47.4 Streamlined Data Retrieval with API Wrappers

Many common web services and APIs have been “wrapped”, i.e. R functions have been written around them which send your query to the server and format the response. This is a great way to get started with APIs, as you don’t need to worry about the details of the API itself. You can just focus on the data you want to get.

API-wrapping packages act as intermediaries between your R environment and web services. They handle the nitty-gritty of API calls, authentication, and data parsing, allowing you to focus on analysis rather than data acquisition logistics. These packages are especially useful for beginners, as they abstract away the complexities of web service interaction. For added bonuses, they often ensure that the data you receive is actually from the API you intended to query, and they provide a structured reproducible way to access the data.

47.4.1 Case Study: Ornithological Data with rebird

Let’s dive into a practical example using the rebird package, which interfaces with the eBird database. eBird lets birders upload sightings of birds, and allows everyone access to those data.

library(tidyverse)
library(rebird)
#> Error in library(rebird): there is no package called 'rebird'

First, let’s fetch recent bird sightings from a specific location:

47.4.1.1 Search birds by geography

The eBird website categorizes some popular locations as “Hotspots”. These are areas where there are both lots of birds and lots of birders. One such location is at Iona Island, near Vancouver. You can see data for this Hotspot at http://ebird.org/ebird/hotspot/L261851.

At that link, you will see a page like this:

Iona Island

Figure 47.1: Iona Island

The data already looks to be organized in a data frame! rebird allows us to read these data directly into R (the ID code for Iona Island is “L261851”).

ebirdhotspot(locID = "L261851") %>%
  head() %>%
  kable()

We can use the function ebirdgeo() to get a list for an area (note that South and West are negative):

vanbirds <- ebirdgeo(lat = 49.2500, lng = -123.1000)
vanbirds %>%
  head() %>%
  kable()

Note: Check the defaults on this function (e.g. radius of circle, time of year).

We can also search by “region”, which refers to short codes which serve as common shorthands for different political units. For example, France is represented by the letters FR.

frenchbirds <- ebirdregion("FR")
frenchbirds %>%
  head() %>%
  kable()

Find out when a bird has been seen in a certain place! Choosing a name from vanbirds above (the Bald Eagle):

eagle <- ebirdgeo(species = "Haliaeetus leucocephalus", lat = 42, lng = -76)
eagle %>%
  head() %>%
  kable()

rebird knows where you are:

ebirdgeo(species = "Buteo lagopus")

47.4.2 Searching geographic info: geonames

rOpenSci has a package called geonames for accessing the GeoNames API. First, install the geonames package from CRAN and load it.

# install.packages("geonames")
library(geonames)
#> Error in library(geonames): there is no package called 'geonames'

The geonames package website tells us that there are a few things we need to do before we can use geonames to access the GeoNames API:

  1. Go to the GeoNames site and create a new user account.
  2. Check your email and follow the instructions to activate your account.
  3. Click [here] to enable the free web services for your account (Note! You must be logged into your GeoNames account already for the link to work).
  4. Tell R your GeoNames username.

To do the last step, we could run this line in R…

options(geonamesUsername="my_user_name")

…but this is insecure. We don’t want to risk committing this line and pushing it to our public GitHub page!

Instead, we can add this line to our .Rprofile so it will be hidden. One way to edit your .Rprofile is with the helper function edit_r_profile() from the usethis package. Install/load the usethis package and run edit_r_profile() in the R Console:

# install.packages("usethis")
library(usethis)
edit_r_profile()

This will open up your .Rprofile file. Add options(geonamesUsername="my_user_name") on a new line (replace “my_user_name” with your GeoNames username).

Important: Make sure your .Rprofile ends with a blank line!

Save the file, close it, and restart R. Now we’re ready to start using geonames to search the GeoNames API.

(Also see the Cache credentials for HTTPS chapter of Happy Git and GitHub for the useR.)

47.4.2.1 Using GeoNames

What can we do? We can get access to lots of geographical information via the various GeoNames WebServices.

countryInfo <- GNcountryInfo()
glimpse(countryInfo)

This countryInfo dataset is very helpful for accessing the rest of the data because it gives us the standardized codes for country and language.

47.4.2.2 Remixing geonames and rebird

What are the cities of France?

francedata <- countryInfo %>%
  filter(countryName == "France")
frenchcities <- with(francedata, GNcities(
  north = north, east = east,
  south = south, west = west,
  maxRows = 500
))
glimpse(frenchcities)

47.4.3 Wikipedia searching

We can use geonames to search for georeferenced Wikipedia articles. Here are those within 20 km of Rio de Janerio, comparing results for English-language Wikipedia (lang = "en") and Portuguese-language Wikipedia (lang = "pt"):

rio_english <- GNfindNearbyWikipedia(
  lat = -22.9083, lng = -43.1964,
  radius = 20, lang = "en", maxRows = 500
)
rio_portuguese <- GNfindNearbyWikipedia(
  lat = -22.9083, lng = -43.1964,
  radius = 20, lang = "pt", maxRows = 500
)
nrow(rio_english)
nrow(rio_portuguese)

47.4.4 Searching the Public Library of Science: rplos

PLOS ONE is an open-access journal. They allow access to an impressive range of search tools, and allow you to obtain the full text of their articles. rOpenSci has a package called rplos that we can use to interact with the PLOS API. They have a nice tutorial on the rOpenSci website that you can see here. First, install/load the rplos package from CRAN.

# install.packages("rplos")
library(rplos)
#> Error in library(rplos): there is no package called 'rplos'

47.4.4.1 Searching PLOS ONE

Let’s follow along with the rOpenSci tutorial and do some searches:

searchplos(q = "Helianthus", fl = "id", limit = 5)
#> Error in searchplos(q = "Helianthus", fl = "id", limit = 5): could not find function "searchplos"
searchplos("materials_and_methods:France", fl = "title, materials_and_methods")
#> Error in searchplos("materials_and_methods:France", fl = "title, materials_and_methods"): could not find function "searchplos"
searchplos("materials_and_methods:study site", fl = "title, materials_and_methods")
#> Error in searchplos("materials_and_methods:study site", fl = "title, materials_and_methods"): could not find function "searchplos"
searchplos("*:*", fl = "id")
#> Error in searchplos("*:*", fl = "id"): could not find function "searchplos"

Here is a list of options for the search or you can run data(plosfields) followed by plosfields in the R Console.

47.4.4.2 Take a highbrow look

The highplos() function does “highlighted searches on PLOS Journals full-text content”.

highlighted <- highplos(q = "alcohol", hl.fl = "abstract", rows = 10)
#> Error in highplos(q = "alcohol", hl.fl = "abstract", rows = 10): could not find function "highplos"

We can then pass this output to highbrow(), which will open up our default browser where we can browse the highlighted fragments. When we run highbrow(highlighted) in our R Console this is what we see in our browser:

Example rplos highlights

Figure 47.2: Example rplos highlights

47.4.4.3 Plots over time

We can use the plot_throughtime() function to visualize the results of a search over time.

plot_throughtime(terms = "phylogeny", limit = 200)
#> Error in plot_throughtime(terms = "phylogeny", limit = 200): could not find function "plot_throughtime"

47.4.5 Is it a boy or a girl? gender-associated names throughout US history

The gender package allows you access to data on the gender of names in the US. Because names change gender over the years, the probability of a name belonging to a man or a woman also depends on the year.

First, install/load the gender package from CRAN. You may be prompted to also install the companion package, genderdata. Go ahead and say yes. If you don’t see this message no need to worry, it is a one-time install.

# install.packages("gender")
library(gender)
#> Error in library(gender): there is no package called 'gender'

Let’s do some searches for the name Kelsey.

gender("Kelsey")
#> Error in gender("Kelsey"): could not find function "gender"
gender("Kelsey", years = 1940)
#> Error in gender("Kelsey", years = 1940): could not find function "gender"

47.5 Conclusion

API-wrapping packages in R provide powerful tools for accessing diverse data sources. As you progress in your data science journey, mastering these tools will greatly expand the range of data available for your analyses. Remember to always check the terms of service for any API you use, and be mindful of rate limits and data usage policies.