Viz : Text Mining Stranger Things (Season 1 vs. Season 2)
Tamanna Hossain-Kay / 2017-12-03
Right after my husband and I finished binge watching Season 2 , a friend sent me a really interesting sentiment analysis of Stranger Things by Jordan Dworkin. I loved it so much I did a remix !
The viz is set up as a comparison of episodes across seasons . You can pick a chapter and compare most frequently used words , top trigrams (or three word phrases) , and sentiment trajectories in Season 1 versus Season 2.
A sliding window technique of taking a 40-word moving average is used to create the sentiment trajectories . Interpolation is used to smoothen and normalize all episode trajectories to 500 time points. The AFINN lexicon is used for emotional valence association of words.
R and Tableau were my chosen tools. I mostly followed Dworkin’s code for script scraping and creating sentiment trajectories. I didn’t include the community detection (Louvain Method) he does for grouping episodes. Instead, I added trigrams and took a long detour into SVG parsing to re-create R-style wordclouds in Tableau. All of my code is available on GitHub.