ProfOak

Update: It appears that we misinterpreted this data in our original story, specifically the ‘sentiment’ analysis. After reviewing the findings again, the zero value tweets that display “no emotion at all” appear to include the #BringBackNationalPokedex example itself, which we presumed had been excluded. With the implicit sentiment contained in the hashtag, the picture painted by these findings is more starkly one of disappointment among fans.

We apologise for the error in our reading of these results. We’ve left the original article intact below so as not to compound the confusion.


Original Story: The recent news that upcoming mainline Pocket Monsters game Pokémon Sword & Shield won't feature the full National Pokédex has come as an obvious disappointment to many fans. At the weekend Nintendo Life contributor James Bralant discussed why he believes that everyone will ultimately benefit from the purge, but passions are running hot and you'd be forgiven for thinking 90% of the internet are painting #BringBackNationalPokedex banners, grabbing pitchforks and preparing to storm Game Freak to demand the inclusion of all 1000+ Pokémon (in all their forms).

Well, data scientist Juan De Dios Santos has applied some hard science to the situation in an attempt to get a better read on the feeling of the community. Using natural language processing and other data-based analysis techniques, he looked at a sample of over 2000 tweets and analysed the top nouns, adjectives, verbs and adverbs being used to discuss the cull, as well as looking at the 'sentiment' of each tweet and which individual Pokémon was mentioned the most.

Sitting in a labcoat surrounded by bubbling test tubes of distilled internet bile (we like to imagine), the analyst studied 2724 tweets using the hashtag #BringBackNationalPokedex. The most frequently used words are hardly surprising: 'game' was the most popular noun, 'know' the most popular verb, with 'old' and 'not' winning the respective adjective and adverb categories.

More interestingly, the 'sentiment' analysis (using Google Cloud’s Natural Language API) revealed that feelings on the subject were almost perfectly balanced according to the samples analysed. Here's a histogram:

See? Science!
See? Science!

Mr Santos explained his findings:

Surprisingly enough, the sentiments are almost perfectly balanced; the mean value is -0.045, with a standard deviation of 0.27. The peak in the center of the histogram indicates that most of the tweets had no emotion at all, and upon manually inspecting the tweets and the values, I found out that these zero-valued tweets are those made of only hashtags, so no emotions at all.

Examples from the extremes ranged from "THEY RUINED THE SAGA!” and “THAT GAME IS SO BRITISH THEY ARENT EVEN LETTING POKEMON FROM OTHER REGIONS IN” to more positive examples such as: "It’s amazing to see how passionate the pokemon community is about the games, #BringBackNationalDex is a true example.”

Finally, the most frequently mentioned Pokémon turned out to be something of a surprise - it wasn't Pikachu. Fifth gen legendary Zekrom beat out Cyndaquil, Blaziken and Swampert, with the yellow mouse coming in a lowly fifth place. In reality, Wingull topped the mentions thanks to a viral retweet at the time, but this was discounted as an anomaly.

Zekrom: not as cute or easy to carry on your shoulder as Pikachu, but he's got his fans.
Zekrom: not as cute or easy to carry on your shoulder as Pikachu, but he's got his fans.

So there you have it - internet ire might be quick and loud, but science has shown that the reality of the situation is perhaps more reasoned than you might think. Definitely check out the published article for a more exhaustive analysis.

Where do you fall on the spectrum of opinion? Do you agree with these findings? Share your thoughts with a little comment, if you fancy.

[via towardsdatascience.com]