The Tweets

Sentiment Analysis is a powerful tool that allows us to extract and explore public sentiment as distributed on social media. As part of the Pride Project, we scraped over 10,000 tweets regarding the various teams' Pride observations. We used machine learning to evaluate the sentiments in the tweets and categorize them on a 1-5 scale, with 1 being the most negative and 5 being the most positive. We believe this can help us uncover trends among different fan bases. Identifying the most inclusive fan bases may help raise examples for all.

Visualizing the Tweets, Grouped by Team and Sentiment Score



Initial Observations:

  • The team's win/loss record does not seem to have an influence on the fan's participation in the Twitter conversations on this topic.
  • Win/loss record also does not seem to influence the fan base's perception of Pride Night; in fact, one of the most negative fan bases follows one of the winningest teams, and one of the most positive fan bases follows a team with one of the worst wining records.
  • Most fan bases with high survey response counts also had high Twitter activity.

Category Breakdown



By visualizing the sentiment categories as percentage of overall responses, we can identify trends.

  • There is some consistency between the survey and the Twitter scrape, in regards to the teams with the strongest positive and negative feedback. Teams with more neutral survey responses also generated relatively neutral Twitter sentiment.
  • These highly polarized teams also tend to be the most active across social media.