Please note that if you agree with the content of this article, please click here to send a pre-written email to the Padres front office expressing your support.

Helpful shortcuts
Jump to Introduction
Jump to Basic Descriptors
Jump to Anecdotal Evidence
Jump to Topic Modeling
Jump to Sentiment Analysis

Introduction

Twice this offseason, Ron Fowler has erroneously stated that San Diego Padres fans do not want brown uniforms. First, during an interview with the UT, Fowler repeated his previous claims about fans not wanting brown uniforms. More recently, he somehow claimed that more Padres fans want the PCL uniforms than brown ones.

In both cases, Ron Fowler did not provide the data the Padres supposedly used to draw these conclusions – namely the “focus group” data that we’ve heard about for years. Not that the Padres could produce fake data that quickly, nor that they would give it to their fans anyway. If you recall, when journalist Bernie Wilson asked for the data used by Padres ownership in similar past statements, Ron Fowler informed him that “no, you may not” have the data:

But here’s the thing: we actually do have a lot of the data. No, not their mythical focus group data, but a whole lot of data that the public provides the Padres. For example, the Padres’ Facebook page is publicly viewable and full of hundreds or thousands of comments for almost every post they’ve ever made. Same thing, to a lesser degree, with their padres.com message board and other third-party “surveys” like comments at Gaslampball and the Padres subreddit, among others.

These data sources are arguably more valuable and more representative of Padres fans than the focus groups they allegedly proctored. On Facebook alone, there are hundreds of thousands of users who like their page. Even if their focus group included a response from every single season ticket holder, it would still be far less representative than the free online data just sitting there for the team to use. One could also argue that the types of season ticket holders who continue to buy this shitty product will always buy this shitty product and could be effectively ignored in the decision making process. Let alone the fact that proctoring a focus group only using your current best customers is heavily biased and not broadly representative of Padres fans.

With the proliferation of open-source web scrapers and text analytics packages, it’s now surprisingly easy to acquire and analyze this information ourselves, provided it’s attempted by someone with some sort of coding background. Turns out I do have that background.

So in mid-December, I wrote a computer script to scrape every Facebook status the Padres have posted since 2009, along with every comment written on those statuses and every reply to every comment. That might seem like a lot of data – over 7,000 status updates, some with thousands of comments per update – but it doesn’t take more than a few minutes of time to acquire once you’ve written a successful computer script. It actually took my code under one minute to scrape all this data on a Microsoft Surface, which isn’t exactly flush with computational power or RAM. Anyone worth a grain of salt in the Padres front office, marketing team, or business unit should be able to pull off this same query with little effort and without requiring sophisticated hardware. Assuming they have a laptop.

With this data in hand, it was time to do some analysis. And unlike the Padres, I’m actually sharing the data – all Padres status updates and all comments on uniform status updates, with VADER sentiment scores (more on this later).

First things first: the Padres have used the term “uniform” in 36 status updates since 2009. Some of these are orthogonal to the topic of uniform colors themselves, like when the Padres had Will Ferrell “wearing a Padres uniform” or when they announced that Mark Kotsay would be “back in a Padres uniform”. We don’t care about these status updates all that much, so they’ll mostly be ignored or used as a rough reference throughout this analysis.

What we do care about, however, are the twenty-nine status updates that actually are about the uniforms. These vary in degree from a somewhat passive way of soliciting uniform feedback, like this status update from April 2011:

Mat Latos in the new Marine digital camouflage uniform on Padres Military Opening Day. Who’s coming to watch him pitch tomorrow?

to a much more direct way of receiving feedback, like their recent uniform unfurling:

Introducing the 2017 Padres uniform lineup!

Check out the uniforms returning for next season and the two new ones making their debut!

In both cases, fans comment, like, share, and use the new expressive emoticons to provide the Padres with feedback. All this feedback is free to the team and provides them with a larger volume of empirical information than any focus group they may have proctored. Instead of a survey asking one generally older, generally wealthier season ticket holder how they would feel if the Padres did X, we have a dataset where nearly 3/4 of a million people who like the Padres on Facebook are asked how they do feel in response to a real event – this makes our focus group truly pragmatic.

Our goal is to transform this raw data from the Padres’ largest “focus group” into usable analytics that give us true insight into what Padres fans really say about our uniforms. Do fans actually like the PCL uniforms more than brown? Do fans not actually want the brown uniforms? Did fans like the new uniforms? We can answer all these problems in a number of ways.

First, we can examine the basic descriptor data points. How many people liked the most recent uniform post compared to similar ones in the past? What types of expressive emoticons were used and how does it compare to other posts?

Second, we can examine the most liked comments to each post to create anecdotal evidence most representative of a plurality of fans.

Third, we can examine the distribution of words and topics that fans talk about in their comments. Is there more to fan complaints and wishes than just “go back to brown”? Are there any topics in support of the PCL uniforms or that express satisfaction with the current uniforms? We can actually derive this using topic modeling.

And finally, we can score the positivity and negativity – the sentiment – of every uniform comment. What is the overall fan sentiment of each uniform status?

The basic descriptors

Jump to Introduction
Jump to Basic Descriptors
Jump to Anecdotal Evidence
Jump to Topic Modeling
Jump to Sentiment Analysis

The Padres have twenty-nine true uniform-related status updates since 2009. In terms of pure likes, their top three are unsurprising:

“Excited for retro weekend? Get your throwback brown and yellow Padres gear out because you’ll need it at Petco Park starting Thursday!

Here’s the uniform lineup for this week’s series against the Cubs:

A first look at the Padres Way Back Wednesday uniform and batting helmet

“The first Way Back Wednesday is one week from today! Padres will wear throwback uniforms when they take on the Astros next Wednesday at 12:40.

Get your tickets now!”

In other words, the Padres most popular Facebook posts related to uniforms are those which explicitly involve the Padres wearing brown uniforms.

Now consider this: the second post above occurred during the emoticon era on Facebook. It received only one ‘sad’ emoticon, and zero ‘angry’ or ‘laughing’ emoticons. The post which unveiled the 2017 uniforms received 77 sad, 52 angry, and 15 laughing emoticons. That’s fairly cut-and-dry: the brown post was more liked and didn’t experience the negative emoticon feedback that the blue uniform post experienced.

Perhaps the most laughable assertion by Ron Fowler, that we can pretty quickly disprove using basic data points, was his statement that more people tell him they want the old PCL uniforms than brown uniforms. Scanning every single Facebook comment ever written as a response to a uniform-related status update by the Padres produces only 10 comments that use the term “PCL”:

Comment	Author	Timestamp
It’s a homage to the PCL days.	Christopher Norris	11/22/2016 21:03:00
Bring back 1990’s Jerseys! There almost like PCL Padres Jerseys.	Jay de Leon	7/10/2013 13:01:00
1948 PCL Padres jerseys are the best ever!	Pete Gabaldon	3/3/2012 20:39:00
Really cool! I guess there’s one benefit from having TOO MANY uniform styles throughout the years! I can even see the PCL Padre unis!	Mmandem Norris	3/3/2012 16:42:00
PCL unis !!!!!	Miguel Klique Alatorre	2/25/2012 12:41:00
Padres = Zero Traditions except losing T_T wished they had brought back the PCL 1936 jersey.	Jay de Leon	11/12/2011 0:28:00
Like them but still hate the logo on the home whites, it looks like it says “Radres” plus it looks like it is outlined in a dark color? why didnt they use the Padres logo from the PCL uni, it was very cool.	Jorge Verdin	11/10/2011 0:37:00
Support the return of PCL baseball to San Diego – the Ballpark for Escondido on Facebook 🙂	Randolph Ortlieb	6/11/2011 20:02:00
I hate the old PCL uniforms. Old does not always equal good.	Leo Palmer	6/11/2011 19:49:00
“I’ve been a Padres fan since they were the parent club of the Hawaii Islanders (AAA-PCL) in the 70s … In my opinion, the Padres should revert to brown, since that what a padre wears … whether it’s the classic 70s & 80s brown & gold, or the early 90s brown & orange! Brown must be the base color of the San Diego Padres!!! 🙂 The move to blue was stupid!!! Bring back the the gold/brown caps too! :-)”	Ivan Nishimura	1/7/2010 14:22:00

Of the 10 PCL-related comments the Padres have ever received on a uniform status update, only four of them actually advocate or explicitly state their approval of the PCL uniforms. And two of those four comments are by the same person!

That means that in seven years of posting about the Padres uniforms on Facebook, the Padres have been told to revert to their PCL uniforms by just 0.0004% of that part of their fanbase: just three people out of an audience of over 747,000 fans that “like” the Padres on Facebook.

Empirically, Ron Fowler’s statement is bullshit. In 2017, we call this an alternative fact.

Anecdotal evidence

Jump to Introduction
Jump to Basic Descriptors
Jump to Anecdotal Evidence
Jump to Topic Modeling
Jump to Sentiment Analysis

Besides the direct feedback the Padres received through likes and emoticon responses to their posts, the Padres also receive an incredibly robust, free dataset of anecdotal evidence from their fans in the form of comments (brushed upon briefly in the PCL example above).

Other fans, reading these comments, also have the opportunity to like and respond through emoticon to these comments themselves, creating an additional layer of feedback for the Padres. Although a post itself may not solicit a lot of likes, perhaps one or two comments submitted by a fan was liked by many other fans.

For the most part, these comments don’t receive many likes. In the 29 uniform-related statuses the Padres have posted, they’ve received 2,324 comments back, so it only makes sense that most of these are lost in the internet ether. However, some of these do actually receive a lot of support (and all of them will matter in our sentiment analysis later).

When the Padres released their new uniforms on November 22nd, they received three comments that actually garnered over 50 likes from other Padres fans. All three are regarding the Padres failure to return to brown:

“Another example of the team being disconnected from the fan base. We want the authentic brown and yellow. Period.” – James Hall, receiving 160 likes

“So Friday is the only day you will be able to tell the difference between the Padres and the Brewers, Rays, Dodgers, Marlins, etc., etc. Maybe since half the fans in the stands are rooting for the other team they are trying to make the Padres look like the visiting team?” – Michael Shanahan, receiving 79 likes

“Unfortunate.

I’m going to try to provide some feedback here in as objective a way I can (despite being an obviously biased season-ticket holder who rocks a brown jersey to every game I go to).

These unis send one of two messages; that Padres ownership has no idea what their fans want, or that they know and simply don’t care.

In a year where the HoF #’s were moved to make way for that gigantic Syquan eyesore, a trade was partially reversed amidst accusations of withheld information, the GM was suspended, the president was fired, and that whole choir national anthem fiasco happened (not to mention the team underperformed and nearly every starter was traded), it begs the question – WHY NOT?

Why not take advantage of a golden (and brown) opportunity to do right by the fans? Why not change course from the same old boring jerseys that get unveiled year after year? Why not embrace a color that the team and its fans could feel like they truly OWN in the same way Dodger Blue and Cardinal Red are owned? Why not strive to be unique and relevant in a way that the team on the field cannot yet provide? (We’ll get there.)

Don’t get me wrong, there was some good (and some great) last year. The all-star game and home run derby at Petco Park were awesome. Wil Myers’ breakout was a sight to see and really made us feel like we had a true potential star on our hands for the first time since Peavy and Gonzo were traded. Catching a glimpse of the future in Renfroe, Margot, and the gang in September was fun and provided reason for optimism. The banter between Don Orsillo and Mark Grant was hilarious and I’m stoked that they’ll be in the booth together for years to come. With all that positive momentum going into what inevitably will be two or three more rough years of non-contention, WHY NOT be bold and embrace the brown that the fans have so publicly craved?

Unfortunate.” – Paul Matthys, receiving 52 likes

This is only a small fraction of the overall feedback the Padres receive and apparently ignore. They’ve had 298 comments on these 29 posts alone that use the word “brown”. Many others don’t explicitly use the word brown, but carry the same meaning.

What makes the Padres recent unveiling even more damning to Fowler’s assertion that brown uniforms are not what Padres fans want is the overall displeasure expressed by Padres fans through Facebook emoticons.

The Padres have had over 700 posts in the Facebook Emoticon Era. The 2017 unveiling of uniforms ranks in the 94th percentile of angriness and in the 97th percentile of sadness, in terms of the volume of fans who used the angry and sad emoticons.

The next ten angry posts are literally all game recaps which explicitly state that the Padres lost:

Post Text	Timestamp
RECAP: Padres come up short against the Dodgers.	7/9/2016 21:47:19
RECAP: Alex Dickerson’s 8th-inning grand slam not enough as the Padres fall to the Cubs in the series opener.	5/10/2016 23:14:20
RECAP: Padres four home runs not enough in loss to Marlins.	6/14/2016 1:06:58
RECAP: Yangervis Solarte goes 4-for-4 as Padres drop series finale to Yankees.	7/3/2016 19:31:26
RECAP: Padres drop series opener to Reds.	7/30/2016 1:02:44
RECAP: Padres fall to Reds in series finale.	6/26/2016 15:34:29
RECAP: Christian Friedrich strikes out 7 as Padres drop series finale to Dodgers.	7/10/2016 18:36:21
RECAP: Brett Wallace goes deep as Padres fall to Brewers.	5/15/2016 16:54:23
RECAP: Cesar Vargas impresses in Major League debut in #Padres loss vs. Cardinals.	4/23/2016 23:25:14
RECAP: The Padres fall to the Rockies in the series finale at Coors Field.	4/10/2016 18:41:46

And the post directly following the 2017 uniform unveiling in terms of sadness is the Padres tribute to the victims of the Orlando terror attack:

This Flag Day, our flags are flying at half-staff in memorial of the victims in Orlando.

In other words, Padres fans have the same negative social media reaction to the current blue uniforms as they do to actually losing real games to the Dodgers and mass murder. By refusing to go back to brown uniforms, the Padres are basically inflicting a series loss to the Dodgers to play out in the psyche of Padres fans.

And for as disappointing and demonstrably false as Fowler’s statements are, it’s just as frustrating that the well-reasoned, calculated responses are denied existence, while the owner drones on about a (fictional) fan affinity for PCL uniforms.

Ultimately, these Facebook comments are valuable feedback that many companies pay third parties to pry out of their customers; an entire industry exists solely for acquiring this information from customers. Ron Fowler lies about and ignores it.

Topic Modeling

Jump to Introduction
Jump to Basic Descriptors
Jump to Anecdotal Evidence
Jump to Topic Modeling
Jump to Sentiment Analysis

As convincing as the general statistics and anecdotes above are, they’re still an esoteric way of figuring out exactly what fans are saying. Perhaps there are more arguments and comments being made than “bring back the brown” or “I hate the brown”. There probably are other uniform topics within the 2,000+ comments that deserve consideration in our overall judgement. But short of manually reviewing every comment, how can we figure out what else is being said?

One method is called “topic modeling”. (The exact algorithm I used is called Latent Dirichlet Allocation or LDA.)

In brief technical detail, a probabilistic algorithm is applied to the words within the comments. Comments are broken down into the individual words being used and the model derives the distribution of a) most important words for each topic and b) determines the topic(s) each comment is referring to. The underlying assumptions are that each comment refers to a small number of topics and that some words are more indicative of a specific topic than other words.

LDA was originally proven as a method for automatically classifying scientific papers into the correct scientific topic. By simply having a computer automatically extract probabilistic similarities between all scientific papers based on word co-occurence, the researchers were able to determine, with no manual effort, that a paper belonged to neuroscience or biochemistry or electronics strictly due to the distribution of words in the paper.

As a quick example, pretend you made 1,000 different people write you an essay on either dogs or cats. If you forced an LDA model to create two topics based these essays, you’d likely end up with:

• “cat” topic where the words ‘kitten’, ‘meow’, and ‘independent’ are weighted very strongly
• “dog” topic where the words ‘puppy’, ‘bark’, and ‘friendly’ are weighted very strongly

New essays could then be fed into the model and automatically classified into a dog or cat (or both) essay based simply on the distribution of words – the occurrence of ‘kitten’ or ‘bark’ – in the text.

I’m not going to get too bogged down in the technical details behind how I constructed the model, but the best setup determined that within the 2,000+ comments on the uniform posts there are roughly 25 different topics being discussed by fans.

Some of these 25 topics are completely benign to the uniforms themselves. Consider the topic with these word stems as its nine most important:

1. venabl
2. upton
3. shield
4. set
5. outfield
6. power
7. alonso
8. 1st
9. decent

Clearly, fans talk about the team itself within these comments, even when the status isn’t about the players. Here is an example comment by Padres fan Gallardo Angel Miguel on March 4th, 2015 that the model gives a high score in this topic:

I would have asked Venable to make the change to first or at least have tried it out in the spring. Alonso has ZERO power as a firstbaseman. Venable is a lefty and has decent power.

Like earlier, I’m going to ignore these benign topics and focus on the ones that appear to be related to the uniforms. These uniform topics and their important word stems:

1. uniform, year, chang, keep, last, money, buy, differ, make, dont
2. san, diego, font, san diego padr, logo, militari, repres, club
3. home, jersey, better, realli, away, altern, home jersey, made, road, sd
4. look, like, love, team, fan, yellow, best, look like, uniform
5. back, uni, bring, brown, bring back, orang, back brown
6. blue, wear, white, brown, would, bore, one, uniform, pinstrip, jersey
7. friar, miss, sign, except, point, origin, swing, teach, club

Before I bounce through each topic, note than no topic related to the PCL uniforms exists. I know we went over the rarity of the actual term ‘PCL’ earlier, but there isn’t even the term ‘red’ in any topic. It’s safe to conclude that it isn’t something Padres fans are talking about or want.

Now, to each topic. Topic 5 is as clear as it gets: these are the pure “back to brown” advocates. There are many, many comments like these. Since we already pretty thoroughly acknowledged this topic, I’m going to leave it as is.

Take a look at Topic 6. It includes the tern ‘blue’, but also includes the words ‘brown’ and the word stem ‘bore’ – the stem for boring and bored. From the latest uniform unfurling, here’s the comment (from Padres fan Eric Smith) that the model scores the highest for this topic:

So boring. Is that font just something you downloaded off the internet? How many teams out there use blue / white? FFS, this is so boring.

And the next highest scoring for this topic:

Why not the brown and yellow uniforms? How about the blue and orange pinstripes? Stop being so plain and go back to our roots. We look like every other mediocre team in the majors.

I could read you more, but you probably get the point. This is not just a “go back to brown” category, which is already its own thing. This is a “your uniforms are boring” topic.

Ron Fowler won’t find any solace in Topic 1 either. Here are the top scoring comments for that Topic from the new uniform unveiling:

“Why do they keep changing their uniforms? Stupid and a waste of money if you ask me. How many times have the Yankee’s changed their uniforms in the last 100 years? Find an identity and stick with it!” – David Larson

“Stop already with all the jerseys..lol… All the great franchises stick with 1 jersey and build tradition, not a group of different jerseys..c’mon Padres, get it right already….it makes it look like you don’t even know what your identity is…” – Roman Ortiz

“Oh look, new Padres uniforms. No tradition whatsoever, and they keep doing it again and again. Keep the Friday, trash the rest.” – Randy Dumont

These comments are clearly about the Padres consistently changing uniforms and lacking tradition.

Topic 2 isn’t friendly, either. These comments tend to be about the lettering on the uniforms, focusing on both the logo and font, although occasionally delving into the camouflage kits. The comments aren’t generally saying nice things about the font or logo, though, like the highest scoring comment (by Padres fan Christopher John) on the recent uniform unveiling:

And they brought back the weird looking San Diego font logo from 2004 and the uniforms from 2011 wow I will miss the San Diego font logo that they used from 2012 until this year.

Topic 3 isn’t necessarily about liking or disliking something, but indicates that the commenter is talking about one of the specific jerseys. Sticking to the recent uniform unveiling theme, you have these nuggets among the top scoring comments:

“Y’all can do better. The road jersey is recycled from 5 seasons back. And home white might as well just say ACME across the chest.” – Richie Harman

“New home jerseys are awful. So boring. Really bummed we got rid of the yellow from last year. Hope they go back to the home uni’s from last year at some point.” – Phil Cobb

“Home and road are SO boring! The all blue alternate jersey is ugly and looks like a knock off jersey made by a lesser sportswear maker. I’d love to buy a New Jersey but this stuff is so uninspiring.” – Jason Lewis Danio

Topic 7 is similar to the last two: mostly about some niche aspects of the brand, like the swinging friar or ballpark signage. This topic was more popular in earlier years and there was actually only one comment about it on the recent uniform unveiling:

Yeah either the 2004 friar or the original friar would had been better thank goodness I bought a hat that has that original San Diego Padres logo with the Swinging Friar on it!

Put it all together and you understand that even though the Padres are consistently receiving feedback from their fans on specific elements of the uniforms – different kits, the font, the logo, the Friar – they are still unable to obtain a coherent grasp of how the fans view their product. You have to be either willfully ignorant or just plain stupid to ignore both the volume of information they receive back and the clear detail in these topics.

Finally, we reach Topic 4. The words ‘like’ and ‘love’ are featured prominently among its top words, perhaps hinting that some fans actually talk about what they like about the uniforms, and not that they’re just flaming piles of boring shit.

Unfortunately (for Fowler), these adjectives are largely used to describe the brown uniforms or express love for the team despite other complaints. Here are the top comments for Topic 4 from the November uniform unveiling:

“I also Wanted to see if they brought back the brown uniforms. Loved them in the 80s. Now they just look like every other team.” – Jaime Campos

“Yawn….and I love my Padres. Thankfully they will still have the brown uniforms on Friday. May the play on the field be better than the look of the uniform. :)” – Melody Crone

“Wow. Loving the response from everyone on here and I echo the sentiment. These look like bland, generic jerseys and it’s disappointing as it appeared that last year you guys were making concessions to what your fans wanted to see, but once again, the owners decided to ignore and move on. Come on ownership — you had a chance to set the Padres apart from the rest of the MLB. It’s irritating to your fanbase. The Friday Jerseys and the All-Star Game jerseys were amazing. The home and away jerseys are exactly the opposite. At least you kept the Brown and Yellow on Fridays, so at least you have one jersey that I like.” – Douglas Coleman

A reasonable fan would reach the following logical conclusions after reading through the topics created by an impartial mathematical algorithm, based solely on what the math says the fans are talking about:

1. The only color-related sentiment publicly advocated by a reasonable volume of Padres fans is a desire for the team to permanently move to a retro jersey which features brown and yellow or orange.
2. An absolutely insignificant number of fans want the PCL uniforms.
3. Padres fans believe there is a clear lack of tradition, confounded by the team consistently changing uniforms.
4. Padres fans believe the new uniforms are boring.
5. Padres fans have numerous complaints about individual details on the uniforms, be it the home/road/alternate jersey, the font, the logo, and the Friar.
6. If people who prefer the blue exist, they do not exist in a large enough volume to crack the top 25 topics of what Padres fans talk about on Padres uniform-related Facebook statuses.

Sentiment analysis

Jump to Introduction
Jump to Basic Descriptors
Jump to Anecdotal Evidence
Jump to Topic Modeling
Jump to Sentiment Analysis

The most in-depth way we’re going to analyze fan opinion on uniforms is by examining the sentiment of fan comments on uniform-related posts. That’s where we take the comment text and determine how positive or negative the comment is, and express that in a score from -1 to 1.

So how does this work? How can you score a sentence and determine it’s positive or negative?

Awhile back, some researchers gave a large group of people a series of words and asked these people to rank them in order, from most negative word to most positive word. After enough people had ranked enough words, the researchers had an overall sentiment score for each word. From that score they could then assess the sentiment of a statement purely through analyzing the overall score of the sentence, based on the words used. (Other words databases use a different approach, but these are all fairly robust word banks.)

That direct method isn’t entirely perfect, though, nor all that straightforward. For example, using the word ‘not’ should flip the sentiment of the sentence: “this is not nice” has the exact opposite sentiment of “this is nice”. Sarcasm isn’t easy to detect, either: “that’s just fantastic” scores highly from a word-based perspective, but is clearly not a positive sentiment.

There are ways around these imperfections, like examining pairs of words or using algorithms that ingest the entire sentence, break down the grammatical structure, and account for negation words or sarcasm. Google and other text analytic wunderkinds like Amazon spend a lot of capital and employ a lot of brain power to create incredibly accurate systems so that when you go to buy bluetooth headphones, they can instantly tell you which descriptive words from reviews are most pertinent.

But we don’t need to worry about these imperfections. (If I was actually in-charge of marketing, though, I would account for them. But until the Padres pay me…) While it’d be great to have a perfectly accurate picture of Padres fan sentiment, we really only care about the directionality of sentiment. Do we need to know if brown is 83% positive versus 77%? Not really. Just getting to “it’s more positive than negative” or “it’s definitely liked more than blue” is a huge improvement over Ron Fowler’s words.

Given that understanding, we’re going to use the VADER sentiment algorithm. Besides being a well-known algorithm with an awesome acronym, it also comes out-of-the-box with python’s NLTK package, continuing our open-source theme. The algorithm itself does a little of everything we mentioned above, bouncing the words in the comments against a sentiment database, while also applying some heuristics to negate sentiment when it encounters a negation word like ‘not’. We’ll still occasionally be burned by sarcasm with this approach, but we’ll be able to arrive at a directional answer.

For each comment, the sentiment analyzer will return four sentiment values: its overall rating, the positive rating, the neutral rating, and the negative rating. The overall rating is bound between -1 and 1, while each individual rating is between 0 and 1. So a positive rating of 0.99 is extremely positive while a negative rating of 0.20 is only moderately negative. Since a comment can have both positive and negative sentiment within the same comment – for example, one may like the logo, but hate the color scheme – VADER will give us those details in addition to the overall net sentiment. This is great because there is a difference between indifference and a polar response; “I don’t care about uniforms” and “these uniforms suck but I love the way you released the uniforms to the public” are very different responses that may score similarly overall.

To give you an example of how this works and get you comfortable with the scores, let’s take a look at how a few random comments score in each category:

they brought back the bow-time road jerseys. I hate them before, I still hate them now. Last year’s road unis are way better.

Category	Rating
Overall Sentiment	-0.67
Positive Sentiment	0.10
Negative Sentiment	0.26
Neutral Sentiment	0.64

This should make sense. The overall rating is negative, given the strong negative sentiment in the second sentence, but there is still some positive sentiment in that final sentence, keeping this comment’s overall score from being the lowest possible: -1.0.

Now for a more complicated example, to show you the true power here:

Bummer. The navy and yellow was a good look. But the brown and yellow is the classic and unique scheme of the Padres. Why not pick a good identity and stick with it? These are all terrible except for the brown jersey.

Category	Rating
Overall Sentiment	0.03
Positive Sentiment	0.12
Negative Sentiment	0.12
Neutral Sentiment	0.76

This makes some sense, right? The comment expresses “bummer”, follows it with a positive sentence, follows that with a neutral sentence, asks a question with a positive adjective, and then ends with a mixed sentence stating one negative with an exception. The ratings reflect all this well.

More than anything, this is a good reason why we should focus on the individual parts of the ratings, not just the overall rating; although it scores evenly in positive and negative sentiment, if you were the person in-charge of judging how fans view the uniforms, you’d want to know there was negative sentiment, despite the neutral score overall.

Onto the overall results…

Of the thirty plus Padres status updates containing the word “uniform”, nine of them received over 100 comments. That’s enough comments to make a meaningful directional ranking. If we do that by creating a ratio of negative to positive VADER attributes, and then rank these nine status updates – truncated for the readers’ sake – in order from most negative to most positive, it goes:

1. Introducing the 2017 Padres uniform lineup!
2. Today, we unveiled our updated 2012 uniforms and logos.
3. We present to you the first lineup of the 2015 season!
4. Voting for the 2012 Retro Night uniforms has ended.
5. Mark Kotsay will be back in a Padres uniform next season.
6. Excited for retro weekend? Get your throwback brown and yellow
7. Recently acquired: Will Ferrell. Ferrell will appear in tomorrow
8. A first look at the Padres Way Back Wednesday uniform
9. Amazing behind-the-scenes video of Will Ferrell and Wil Myers

Very clearly, the new blue uniform statuses are the most negative even if we make an attempt to consider offsetting positive sentiments, by utilizing a ratio. Conversely, Way Back Wednesday was as universally loved as Will Ferrell’s publicity stunt in a Padres uniform, ranking between both of those posts.

If instead of including the positive responses, we focus solely on the negative, the results are just as conclusive. Take a look at the plot below. It displays the cumulative percent of comments with a negative rating at least as large as the x-axis value, with several topical uniform-related statuses shown as separate series.

As you can plainly see, the percent of comments with a portion of their sentiment scoring negative is significantly higher for the 2012 and 2017 blue uniform updates than several of the brown statuses: roughly 28% of all comments on the 2017 and 2012 uniform updates have some negative sentiment, while just 8-14% of brown-related comments have some negative sentiment. And, again, this doesn’t even include sarcasm.

You may notice that I also threw in the Padres’ PCL throwback status against the Nationals, which ranked between the brown statuses and the blue statuses: it wasn’t hated as much as the new uniforms, but they still weren’t very well-received.

Conclusion

Ron Fowler’s statements (and all previous Padres statements) about fans preferring blue uniforms to brown uniforms are demonstrably false. Using data available to the public, representing a larger population of Padres fans than any focus group they claim to have proctored, we have been able to deduce this conclusion regardless of the analysis method: the basic descriptive statistics, anecdotes, topic models, and sentiment analysis all agree. And I’ve released this data to prove my point.

If you agree, please click here to send a pre-written email to the Padres front office expressing your support.

Based on this clear information and the Padres’ unwillingness to present data that supports Fowler’s position, along with rumors that Mrs. Fowler hates the brown uniforms, one can only conclude that the Padres refusing to embrace brown uniforms means that Ron Fowler is simply unable to fully satisfy Mrs. Fowler in other ways.