I was interested if there was a way to discern how Twitter felt about the football league I watch, the prem. As a short exercise I decided to scrape 1000 most recent tweets that included a hashtag used by the club and used Natural Language Toolkit’s sentiment analyser to find a mean score per club for 3 metrics; positivity, negativity, and a compound score of the two also including neutrality. Positivity and Negativity ranged from 0 to 1 (1 being most extreme), and compound ranged from 1 (most positive) to -1 (most negative), Here is what I found:

Interestingly the standouts are Brighton, Arsenal, and Liverpool being the most hated clubs in England, with Brighton (0.064), close to double the league’s average (0.035). Perhaps due to their apparent over performance leading to a lot of upsets? The other comment is on Manchester United being the stand out most positive team, with their positive score (0.11) greatly exceeding the league average (0.08). Lastly a nice takeaway is that the perception of teams in the league, despite rivalries, is overall more positive than it is negative (at least at time of writing!).

I haven’t finished exploring this, in the future I plan on refining the twitter scraping to include more relevant search terms, as well as including a time series. On top of this I plan to improve the sentimental analysis to increase its accuracy, and also potentially include weight of emotion.