Time to continue where I left off. With each US senator’s twitter handle handled, it’s time to get their tweets following a similar create/use-function-and-iterate pattern.

Prep work

Getting started with rtweet is relatively simple, and you can follow the directions on the package’s GitHub page. I have an app already in place. In R, this is all we need to do to get started:

t_token <- create_token(app = "text_mining_for_r",
                        consumer_key = "Your_Consumer_Key",
                        consumer_secret = "Your_Consumer_Secret")


Now we’ll iterate with get_timeline():

safe_timeline <- possibly(get_timeline, NULL)
tweets <- twitter_accounts$identifier %>% 
  map(safe_timeline, n = 10, token = t_token, lang = "en")

This was only partially successful, as I ran up against Twitter’s API Rate limits. That meant that my tweets object had about 40 NULL values, and I had to wait an hour before continuing the iteration and binding everything together:

null_handles <- which(unlist(lapply(tweets, is.null)))

tweets2 <- twitter_accounts$identifier[null_handles] %>% 
  map(safe_timeline, n = 10, token = t_token, lang = "en")
tweets3 <- Filter(Negate(is.null), c(tweets[-null_handles], tweets2))

tweets4 <- tweets3 %>% 
  map(select, created_at, screen_name, text, is_retweet) %>% 
  bind_rows() %>% 
  left_join(twitter_accounts, by = c("screen_name" = "identifier"))

Text Analysis

…Not today. Looks like this just became a three part series.