Exploring the #RStats Community on Twitter

Twitter is a venue for professional skills development …

… not just a place to post cat pics and complain to airlines

library(twitteR)

setup_twitter_oauth(consumer_key,
                    consumer_secret,
                    access_token,
                    access_secret)

[1] "Using direct authentication"

#Who uses the #RStats hashtag the most?
r.stats.search <- searchTwitter("#Rstats", n=1000)

# For the top ten tweeters of the #RStats hashtag...

followers <- user$getFollowers(n=200)
favorites <- user$getFavorites(n=200)

(1) Play with API to know it's capabilities (2) Create a dummy dataset to plan out DB build (3) Go back to the API to get data in a reasonable format

sean <- c("nate", "matt", "sara", "pete", "laura")
laura <- c("maura", "sean", "greg", "matt", "sara")
albert <- c("nate", "sara", "roger", "maura")

add.rel <- function(flwr, user){
  user = getOrCreateNode(graph, "Name", name=user)
  flwr = getOrCreateNode(graph, "Name", name=flwr)
  createRel(flwr, "FOLLOWS", user)}

user <- "sean"
a <- lapply(sean, FUN = add.rel, user = user)

add.rel.followers <- function(flwr, user){
  user = getOrCreateNode(graph, "Name", name=user)
  flwr = getOrCreateNode(graph, "Name", name=flwr)
  createRel(flwr, "FOLLOWS", user)}

for (i in 1:length(users)){
  a <- lapply(unlist(unname(followers[i])), FUN = add.rel.followers, user = users[i])}

add.rel.favorites <- function(flwr, user){
  user = getOrCreateNode(graph, "Name", name=user)
  flwr = getOrCreateNode(graph, "Name", name=flwr)
  createRel(flwr, "FAVORITED", user)}

for (i in 1:length(users)){
  a <- lapply(unlist(unname(favorites[i])), FUN = add.rel.favorites, user = users.fav[i])}

• API limits depth of analysis
  • Favorites limited to 200 by Twitter API
• Processing/memory limits bredth of analysis
  • For how many users are we getting their favorites/followers? Every time you get the followers for another user your node count increases by their total follower count
  • If you are focusing on key influencers (e.g. Hadley Wickham with 15K followers), your relationship count increases by n x (average follower count) and your node count by slightly less than that since some users will already be in the graph