Browsing through other visualizations, I found remarkable diversity in the types of data folks visualized using these plots. One particularly striking example was [Adam McCann](https://twitter.com/adamemccann) [visualization](https://www.flerlagetwins.com/2019/08/ternary.html) based on text from the Game of Thrones series. Given the upcoming U.S. elections, I thought it would be fun to compare the speech used by the leading Democrat, Republican, and Independent nominees.
I computed the top 250 words (excluding stop words) for each candidate by frequency and merged them all together. I normalized the frequency by the amount of text available. I interpret the resulting number as a measure of how *important* that word is to the candidate’s speech. Comparing the importance of words across the 3 candidates yields the Ternary Plot above. The bubbles are scaled based on the average importance of the word. All analysis was done in Python using standard tools (i.e., Pandas, sklearn) and the graphic was edited in Canva for visual appeal.
I don’t see “brain worm” in RFK Jr’s corner…
I’m surprised the word “unburdened” wasn’t there
Of course Trump’s word is “China” – couldn’t help but read it in his voice lol. I’m curious what the unlabeled circled are.
I’m shocked “work” falls near kamala.
Maybe flip the triangle up side down? It doesn’t make visual sense for RFK to be on top.
Ah the political virtues of “said”, “going” and “like”
Now for non-Americans, who tfk is RFK?
this is an interesting graph but there are fewer words appearing than large bubbles. it’s like some words were picked to be labeled and some not. what is your cutoff? can you lower it?
or better could make an interactive version to observe all word occurences? it would make much more sense.
“I know words. I have the best words.”
Well this tells me nothing useful.
You graphed word cloud data without most of the words. Being about politics, cherry picking data points seems disingenuous.
![Ternary Plot of Key Words from Political Speech Transcripts [OC]](https://www.europesays.com/wp-content/uploads/2024/08/du748gf9ksfd1-1920x1024.jpeg)
12 comments
While scrolling through YouTube videos, I came across the Ternary Plot. Wikipedia defines it as follows:
>A **ternary plot**, **ternary graph**, **triangle plot**, **simplex plot**, or **Gibbs triangle** is a [barycentric](https://en.wikipedia.org/wiki/Barycentric_coordinates_(mathematics)) [plot](https://en.wikipedia.org/wiki/Plot_(graphics)) on three variables which [sum](https://en.wikipedia.org/wiki/Summation) to a constant. It graphically depicts the ratios of the three variables as positions in an [equilateral triangle](https://en.wikipedia.org/wiki/Equilateral_triangle). It is used in [physical chemistry](https://en.wikipedia.org/wiki/Physical_chemistry), [petrology](https://en.wikipedia.org/wiki/Petrology), [mineralogy](https://en.wikipedia.org/wiki/Mineralogy), [metallurgy](https://en.wikipedia.org/wiki/Metallurgy), and other physical sciences to show the compositions of systems composed of three species. Ternary plots are tools for analyzing [compositional data](https://en.wikipedia.org/wiki/Compositional_data) in the three-dimensional case.
Browsing through other visualizations, I found remarkable diversity in the types of data folks visualized using these plots. One particularly striking example was [Adam McCann](https://twitter.com/adamemccann) [visualization](https://www.flerlagetwins.com/2019/08/ternary.html) based on text from the Game of Thrones series. Given the upcoming U.S. elections, I thought it would be fun to compare the speech used by the leading Democrat, Republican, and Independent nominees.
Text was taken from the following sources:
* [https://www.rev.com/blog/transcript-tag/donald-trump](https://www.rev.com/blog/transcript-tag/donald-trump)
* [https://www.rev.com/blog/transcript-tag/kamala-harris](https://www.rev.com/blog/transcript-tag/kamala-harris)
* [https://www.rev.com/blog/transcript-tag/rfk-jr](https://www.rev.com/blog/transcript-tag/rfk-jr)
I computed the top 250 words (excluding stop words) for each candidate by frequency and merged them all together. I normalized the frequency by the amount of text available. I interpret the resulting number as a measure of how *important* that word is to the candidate’s speech. Comparing the importance of words across the 3 candidates yields the Ternary Plot above. The bubbles are scaled based on the average importance of the word. All analysis was done in Python using standard tools (i.e., Pandas, sklearn) and the graphic was edited in Canva for visual appeal.
I don’t see “brain worm” in RFK Jr’s corner…
I’m surprised the word “unburdened” wasn’t there
Of course Trump’s word is “China” – couldn’t help but read it in his voice lol. I’m curious what the unlabeled circled are.
I’m shocked “work” falls near kamala.
Maybe flip the triangle up side down? It doesn’t make visual sense for RFK to be on top.
Ah the political virtues of “said”, “going” and “like”
Now for non-Americans, who tfk is RFK?
this is an interesting graph but there are fewer words appearing than large bubbles. it’s like some words were picked to be labeled and some not. what is your cutoff? can you lower it?
or better could make an interactive version to observe all word occurences? it would make much more sense.
“I know words. I have the best words.”
Well this tells me nothing useful.
You graphed word cloud data without most of the words. Being about politics, cherry picking data points seems disingenuous.