Warning: Contains major spoilers for this episode.
This week’s episode of Game of Thrones, “Home”, brought back to the show some of its most beloved characters while others bit the dust. As part of our continuing series of how the show’s fans are reacting to every episode, we once again carried out a sentiment analysis of Twitter users commenting on the latest episode.
Sentiment analysis of Twitter users – 11 Do’s and Don’ts
Sentiment analysis is the analysis of emotions, attitudes and opinions that are useful for making better business decisions.
Check out the fan reaction to episode one here
Chart 1: #branstark emotional states word cloud and bar chart
Bran’s return to the show, after his absence in season five, along with his grown-up looks, put most of the show’s fans in the joy category as can be seen in the bar chart. In the word cloud, you can see that there are hardly any tweets in the fear and disgust category compared to the surprise and joy category.
Tweet examples –
Alert
See
Chart 2: #branstark polarity analysis
A polarity analysis of the tweets for Bran show that the tweets were almost equally divided between neutral and positive.
Chart 3: #ramsaybolton emotional states word cloud and bar chart
Ramsay Bolton is the character that viewers love to hate and this fact is obvious in the tweets as well. Ramsay killed his father, Roose Bolton, in almost the same way that Roose had killed Robb Stark, delivering a much deserved comeuppance to the senior Bolton. But Ramsay also killed his stepmother and newly born half-brother by unleashing his dogs on them. Both these acts seem to have divided the Twitter fans into evenly split joy and anger camps. There is also a fair amount of disgust, fear and sadness at his acts as can be seen in the bar chart. The word cloud shows people expressing their disgust and fear with words like vile, sick, cruel, omg and afraid.
Tweet examples –
Cruel
OMG
Chart 4: #ramsaybolton polarity analysis
A polarity analysis of tweets about Ramsay Bolton shows that most of them fall in the negative category.
Chart 5: #jonsnow emotional states word cloud and bar chart
The biggest surprise of the show, and the most anticipated moment of season six, Jon Snow’s resurrection, was unveiled in the final moments of the show and Twitter users are overwhelmingly in joy over it. The word cloud shows them using terms such as biggest, amaz(e), love and vibe to describe the scene. The character’s return is even being compared to Gandalf’s return in the Lord of the Rings!
Tweet examples –
Vibe
Time
Chart 6: #jonsnow polarity analysis
A polarity analysis of the tweets shows that most of them were in Jon’s favor, but there seem to be a number of tweets that were classified as negative. This means that they were either from fans not pleased with enough screen time given to their beloved character or supporters of Daenerys.
Chart 7: #gameofthrones emotional states word cloud and bar chart
The overall response to the episode was that most of the show’s fans found the episode good since their tweets fall into the joy category, far outweighing all the other five emotional states. With words such as nice, like, bigger and amazing being mentioned repeatedly, the episode seems to have given fans much needed closure on last season’s cliffhanger – what will happen to Jon Snow?
Tweet examples –
Death
God
Chart 8: #gameofthrones polarity analysis
A polarity analysis of the show’s main hashtag, #gameofthrones, shows that while the positive tweets outweigh the number of negative ones, the difference is not very huge. This is because the polarity analysis includes people commenting on everything happening in the show, including other characters’ hashtags that we have not analyzed.
How we do it
Our data visualization team analyzed the latest two thousand tweets for four of the most popular hashtags for this episode – #gameofthrones, #jonsnow, #ramsaybolton and #branstark.
The first step was doing some data cleaning, which involved removing incorrect punctuation, whitespaces, tabs, HTML links and so on.
Once we got the “clean” tweets, we then classified them based on emotions using the naive Bayes algorithm after which we classified the text based on polarity.
The script was written in R which classifies the tweets into one of the six emotional states – sadness, joy, disgust, surprise, fear and anger. The next stage was classifying these tweets into one of the polarities – positive, negative and neutral.
Come back next week to find out how Twitter users are reacting to the third episode.
Image Source: HBO.COM
