I've created this GitHub respository that analyzes nearly 7,000 Facebook posts by NC State Senators between Jan 2014 to Dec 2016. Specifically, I use Structural Topic Modeling to identify the topics and estimate the effect party has on each topic's prevalence.
I haven't uploaded the data used onto GitHub but email me if you're interested in the raw data.
See either Part 1 or Part 2 below for the code and commentary.
Part | Topics |
---|---|
Part 1 | Pre-processing & Exploratory |
Part 2 | Structural Topic Modeling |
For the code on how the data was pulled, see the file pullFacebook.R