I spent the majority of the most recent presidential debate thinking of a fun #rstats project. I had three objectives for the post:
(1) that it be fun to do; (2) that it showcase some functionality of rcicero
or rcanvas
; and (3) that it be interesting
to the public. The third objective was admittedly negotiable.
In the next two posts, I’m going to analyze the tweets from our US senators in response to #debate. Part I will show how to obtain each
senator’s twitter account with the rcicero
package, and Part II will perform some basic text analysis on the corpus of tweets with the awesome rtweet
package.
Prep Work
rcicero
requires some prep work before use which I’ve spelled out on the package’s GitHub README. Let’s
jump right in:
To get the twitter accounts, I needed a pair of longitude and latitude coordinates from each state, so I just grabbed
some from the zipcodes
package:
Get the Twitter Handles
Now I need to pass each pair of coordinates through the Cicero API with a custom function that gets each senator’s twitter account, and, just for fun, their party:
Iterate
With our custom function in hand, it’s time to iterate with the purrr
package which is so beautiful and elegant, I can’t even bear to look at some of my old code. Warning: the code below is fairly expensive–it will run you about 59 Cicero API credits.
Voila! Our twitter_accounts
data.frame
now has each senator’s last_name, first_name, twitter handle, and party:
Stay tuned for Part II.