London derby on Twitter

Arsenal has been through a tumultuous time recently. It seems the the fans are slowly coming to a consensus of change after a few seasons of division. This division has played out very publicly on social media so I sought to look at the fans thoughts on twitter.

On the 30th of April, Arsenal faced their biggest rivals in Tottenham Hotspur in a match that seems to highlight a shift of authority, in this season at least, between two North London teams. Arsenal have just come of three ‘confidence boosting wins’ against Middlesborough and Leicester City in the EPL, but most notably against Manchester City in the FA cup. But before that had been showing some unexpectedly poor form.

I am an Arsenal fan and as such I don’t know too much what Tottenham has been up to but I googled it and they seem to have been winning a lot.

Below is a review of the tweets that occurred from 0109 to 2002 on the 30th of May 2017 (which is about when I downloaded the tweets) from Arsenal fans. This gave me around 4500 tweets, I don’t know the ins and outs of the twitter API - so I’m not sure if this is a sample of the tweets at the time or all of them.

The first thing I looked at is when are people actually tweeting? Below is a graph showing when people are tweeting about the game, where the red lines indicate when the game is actually played.

plot of chunk tweet_time

So most people tweet right before, during and right after the game. Not that surprising but a good start to the analysis none the less!

Next I looked at what are people tweeting about. I do some text standardizing stuff to analyze the data. If you’re not interested skip the next bit - suffice to say things are all lower case and I cut the ends off some of the words.

I have cleaned these tweets by removing non-ASCII characters, RT-text, mentions, punctuation, numbers, urls/websites. Then lower cased and stemmed everything using a snowball stemmer.

All analysis is performed in R, primarily with the tidytext library. I did this following the examples set out in tidy text mining, by Julia Silge & David Robinson.

Stemmed word Count
arsen 823
coyg 595
nld 352
spur 287
gunner 239
northlondonderbi 233
game 221
giroud 209
afc 198
win 198

The results aren’t particularly exciting, so I only included the top 10. The jist of it is that most words describe the match being played. COYG, we need to win, this is a north London derby etc. etc.

Players are also mentioned a lot.

Giroud is somewhat surprisingly Arsenal’s most mentioned played, mentioned more often than Tottenham with 209 occurrences. Followed by Ramsey (153), Cech (84) & Ozil (83). Sanchez was only mentioned 64 times in this sample! Perhaps most surprising is that Gabriel was only mentioned 55 times, since he gave away a significant penalty in the second half.

funny side point: Swearing wasn’t uncommon, fuck mentioned 131 times and shit 71 times.

So what are people saying about Giroud?

plot of chunk giroud

Not very nice things. Twitter is not happy with Giroud, whether that being the way he played, or being involved at all (Danny Welbeck anyone?). Personally, I like Giroud, but his performance during this match doesn’t make him easy to defend.

If we look at when the negative feeling toward Giroud over time though - it looks like his performance didn’t matter too much.

plot of chunk giroud_time

Most people seem to be annoyed most when the teams are announced and the early parts of the game. There also looks like a bit of a spike around half-time when Wenger hasn’t made a change. The pro-Giroud team seem show a bit of a rally at the start of the match, but Giroud doesn’t really impose himself and they (and I) soon go quiet.

I also look at the emotion associated across all the tweets, which gives a good view of the game from an Arsenal point of view.

plot of chunk sentiment

The red lines show (from left to right) start of the game, the Alli chance, Eriksen chance, Alli goal followed by Kane’s penalty and finally the final whistle that couldn’t have come sooner.

All in all not a great day for Arsenal, and the twitter-verse certainly reflects that. Hopefully there is a turn around and we can have a more positive view of twitters reaction to Arsenal. Maybe we will see the team kick up a gear or three against United next week.

I’m not holding my breath though.