About

For a data science project, my team of 4 decided to explore themes in a dataset of trending YouTube videos in the United States.

I performed a sentiment analysis on the video titles using the syuzhet package in R. With 4 different lexicons, I explored the valence (positivity or negativity) and sentiments (anger, anticipation, disgust, fear, joy, sadness, surprise, and/or trust) conveyed in titles.

I found that those that conveyed a high level of trust were common among trending titles, especially among the very highest level of popularity. It was also interesting that while we saw a higher frequency of positive words by count, there were a wider variety of (more unique) negative words in titles.