Introduction
Since Englands World Cup opener against Iran, I’ve been attempting to capture the Tweets of the, sometimes fickle, England fanbase on which to conduct some analysis.
My goal was to capture as many relevant tweets as possible, then apply a model to conduct sentiment analysis on the contents of the Tweet. At a high level, the model analyses the Tweet text and applies an emotion to it, either Fear, Joy, Surprise, Neutral, Sadness, Anger or Disgust with a score between 0 and 100, where scores closer to 100 indicate a higher probability that the assessment of emotion is accurate (find more technical details at the bottom the page).
As a result, we can see general posting trends during matches, which begs the question, are people angry or joyful while England are playing? Read to the end to find out.
We can also see, for any England player, what the general sentiment is towards them. Most people would agree that Jude Bellingham has had an incredible World Cup so far, but does Twitter reflect this?
Well, yes, unsurprisingly. He has the most Tweets registered with the emotion; ‘Joy’.
My Approach
Across 4 England matches so far (3 group matches & 1 last 16 match & at least one more to come), I have collected 124,386 usable tweets. To collect the data, I’ve used the Twitter API where, in each case, Tweets were collected if they included the official England hashtag ‘#ThreeLions’ or, if they included the match hashtag, for example ‘#ENGUSA’ in the case of England vs United States.
Tweets were streamed and captured in real-time to my laptop then uploaded to the Domo platform after the game. Collection of data during the games went fairly well, with the exception of the Wales game which unfortunately was frustratingly disconnected by Twitter due to the sheer volume of tweets being posted by users. I was watching the match at the pub at the time and couldn’t restart it, so I have data for only around the first 15 minutes of the game. Apologies for this, but only slight apologies because I’d honestly rather be at the pub than watching my Twitter script.
I’ve used Domo’s superb integration of Jupyter Notebooks to process the data and to apply the model (Jochen Hartmann, “Emotion English DistilRoBERTa-base”).
And for the follow up post around the technical side of things, follow me on LinkedIn.
And, finally, if you’re still here and desperate to see inside the mind of the average England fan on Twitter during a match then allow me to oblige. If the England v USA game was anything to go by, the leading emotion of an England fan is Fear, closely followed by Anger. Lovely.
The Dashboard
Warning: This is unfiltered Twitter data posted by real people. There is, at best, foul language in the analysis and, given that this is from the English fanbase, who knows what else, so please bear this in mind before proceeding.