Jets Fan Sentiment Analysis

Ryan Young
4 min readDec 26, 2020

Welcome to my data blog. I’m a student at Duke Fuqua’s MS in Business Analytics program and my hope is that you will get to know me better over the next few weeks through this blog.

I know that 2020 has been one of the most tumultuous years in U.S. history so I will try to find the solutions to our most pressing issues…

JK…I’m mostly just here to have some fun (as you can probably tell by the title of this post). But I plan on using critical thinking, data analysis and visualizations to enlighten and entertain…myself (at least).

Monday’s NY Daily News Back Cover

I’m a quasi-Jets fan, but I would rather root for chaos and the Jets win on Sunday was well worth watching Twitter melt down. New York entered Sunday 0–13, in the driver’s seat to land the number one spot in next year’s NFL draft (to be in position to draft the coveted QB Trevor Lawrence). However, the win placed them in a tie with the 1–13 Jacksonville Jaguars and because of tiebreakers that favor the Jaguars, there is now little hope of salvaging a franchise quarterback out of this pitiful season.

It’s hard to earn one lonely win in a season and still be denied the top pick, but this certainly would be the one franchise to do it. In fact, here is a quick sentiment analysis of the last 50 years of Jets fandom: PAIN.

But I thought a sentiment analysis of Sunday’s Jets win could be unique…because how much negativity could actually come out of a win? Surely, there are some [misguided] Jets fans that do not want the ignominy of a winless season.

I used the Twitter API to gather replies to the Jets’ twitter account from the start of the game until two hours after its conclusion (from 4–9pm EST). I felt that using replies to @nyjets tweets was the best way to get a representation of actual Jets fans. This limits national commentary and neutral observers, but of course there will still be a handful of trolls.

I broke down these 5,033 tweets into individual words, removed “stopwords” and bleeped a common four-letter word that fans tended to use.

It is not very hard to gauge the general sentiment my taking a peek at the most-used words. (It’s also pretty amusing that no Jets player was among the top-26 words.) However, I wanted to test out three of the primary sentiment analysis dictionaries in R (bing, nrc and afinn) and see if they would agree that Jets fans were miserable and angry while their team looked great on the field against the 9–4 Rams.

The table on the left uses the “bing” sentiment dictionary, while the right uses “nrc” that categorizes the words into emotions (including positive and negative).

These results seem to incorrectly indicate that the sentiment was nearly evenly split between positive and negative emotions. There are several reasons for this. Many words are misinterpreted in this context, including the three words (in these dictionaries) that were used most frequently. “Win”, “lose” and “fans” are not the signals they traditionally may be. Jets fans are telling their team (or at least their social media person) what to do. In fact, the probability that any given “win” was preceded by a “don’t” is pretty high. This leads to another limitation…not using bigrams, as individual words often lose their context, especially when it comes to negation. It is also difficult to account for sarcasm, trolls and Jaguars fans, all of which likely contributed to the positive sentiments. Although, fans of the opposing team, who usually would also greatly distort the results, were not a factor in this case because Rams fans were very much not pleased with the game either.

To try and account for some of these limitations, I eliminated some of the most popular words that I believed were misinterpreted (such as “win”, “lose” and “fans”). I then used the “afinn” dictionary, which assigns a value for the sentiment of each word (ranging from -5 for the most negative to 5 for the most positive). I then summed the sentiment values of the tweeted words for each minute and created a plot.

This chart seems to accurately convey Jets fans’ sentiment, which negatively spiked following the touchdown that put New York ahead by three scores and then again when the Jets tweeted the final score at 7:01pm. This has to be one of the most unique fan sentiment charts you will ever see for a win.

If it’s any consolation to Jets fans, they will find a way to mess up their draft selection regardless of their position.

Merry Christmas, Jets fans. (And for those not celebrating, Happy March 300th.)

Data obtained through Twitter API

Check out my github for the code behind each post.