How to conduct a tombola with R

Two weeks ago, we announced to raffle off three hardcover versions of our ADCR book among all followers of our Twitter account @RDataCollection. Tomorrow is closing day, so it is high time to present the drawing procedure, which, as a matter of course, is conducted with R.

Connecting with Twitter

We start with installing the latest version of Jeff Gentry's twitteR package from GitHub, which makes the OAuth authentication handshake procedure very comfortable:

devtools::install_github("geoffjentry/twitteR")
library(twitteR)

Next, we have to authenticate our Twitter app. Note that this requires that we already registered our app (for free) at https://dev.twitter.com/ and stored the credentials (API key and secret as well as access token and secret) locally. An excellent way of storing your credentials locally as R environment variables has been described by @JennyBryan at http://stat545-ubc.github.io/bit003_api-key-env-var.html, which we follow here. To do so, we first write the tokens in an R object as name-value pairs (it goes without saying that the keys presented here are fictional):

credentials <- c(
  "twitter_api_key=rN3Td2zZADLWZBN9Pj7X2eBN",
  "twitter_api_secret=abcqBpUzE7BQ65QJ6BRzpUzjyaRCfwn3ndrUUcqDWfhCN7Fj",
  "twitter_access_token=9287465372-6ckQsXGP83eaXCsQHFQFx5pUNhmYYqknnCwWScVk8n7L",
  "twitter_access_token_secret=ZHUxEW5fefntdyWBBB95fuXY5umZzWXdtPKtjUEP9GDcJs6w"
  )

In order to write the keys to a local .Renviron file in the default working directory, we write:

fname <- paste0(normalizePath("~/"),".Renviron")
writeLines(credentials, fname)

We have to do this only once to retrieve the keys in later R sessions. The benefit of this approach is that we do not have to store the keys in actual R code which we plan to publish. To see if this worked, we can retrieve the file again and inspect its content:

browseURL(fname)

After reloading R, we can use Sys.getenv() to retrieve the keys again and feed them to twitteR's setup_twitter_oauth() function, which takes care of the entire handshake procedure from start to finish:

api_key <- Sys.getenv("twitter_api_key")
api_secret <- Sys.getenv("twitter_api_secret")
access_token <- Sys.getenv("twitter_access_token")
access_token_secret <- Sys.getenv("twitter_access_token_secret")
setup_twitter_oauth(api_key,api_secret,access_token,access_token_secret)

Retrieving the followers

Now that we have access to @RDataCollection's Twitter account, we collect information about the account and the number of followers:

user <- getUser("RDataCollection")
user$getFollowersCount()

## 122

At the moment, the account has 122 followers, which gives each of them a chance of approximately 2.5 percent to win one of the books. In order to draw from the list of followers, we extract their screennames using twitteR's internal methods:

user_followers <- user$getFollowers()
followers_n <- length(user_followers)
followers_screennames <- vector()
for (i in 1:followers_n) {
  followers_screennames[i] <- user_followers[[i]]$screenName
}

To be fair, we exclude the author's accounts from the tombola:

authors <- c("simonsaysnothin", "christianrubba", "marvin_dpr", "jonas_nijhuis", "phdwhipbot")
followers <- setdiff(followers_screennames, authors)

Drawing the winners

Finally we are ready to draw the three winners!

sample(followers, 3)

Well, don't worry – we won't perform the actual draw before tomorrow evening. Best of luck in winning one of the books, and a peaceful 4th Advent Sunday everyone!