Many presidential hopefuls see the best way to win over voters as promising to tackle a prominent issue of their day. With President Obama, it was healthcare; with President George W. Bush, it was tax-cuts. It is often wondered whether these issues that they speak about during pre-election debates will actually be what they focus on most while in office.
The goal of this project is to draw parallels through textual analysis of what a president committed to solving in pre-election debates, and what they actually took the steps to enact during their presidency. The first dataset I chose to approach these questions is the Presidential Candidate Debate transcripts from 1960 - 2016 from The American Presidency Project at UC Santa Barbara. The other was presidential speech transcripts from The Miller Center at University of Virginia. Once I scraped this data and processed it using Python's NLTK, I learned a method to classify each speech by topics contained within it from my course, Data 88: Language Modeling and Text Analysis. This was done in a topic modeling technique called Latent Dirichlet Allocation, and visualized with network analysis. I also measured speech similarity, both across presidents and for a single president before and after his election. Check out my poster for more information on my workflow, results, and visualizations!