Who will take the 2017 NBA MVP crown? It certainly will not be a unanimous decision like last year’s.
I thought it’d be interesting to see what great fans think about this year’s packed NBA MVP race. The weight of public opinion on sports now, relative to what it was a decade ago, has increased thanks in large part to social media platforms, like Twitter. Many of us follow the NBA regularly, tweet about a good nightly game between intriguing teams, and get into an occasional spout with someone whose opinion on an NBA player/team differs.
Twitter provides a way that we can analyze tweets in order to understand trends about groupthink stances on certain NBA topics, as well as smaller communities that might be in the minority.
I used the Twitter API for the authorization, searched through 10000s of tweets with R program to find commonalities, and attempted to ascertain how highly NBA Twitter thinks of five NBA MVP candidates through some preliminary sentiment analysis.
Creative Process
2017 NBA MVP Twitter Word Clouds
After receiving authorization through a few commands, I began some data mining with a “searchTwitter()” tool from the twitteR package in RStudio.
I chose the following five candidates:
- Russell Westbrook
- James Harden
- Kawhi Leonard
- LeBron James
- Stephen Curry
Then, I searched over 5000 tweets (as of 4/8/17) involving their names and the word MVP (“Westbrook+MVP”, like this) to gather relevant recent and popular tweets. After collecting the character strings and placing them in corpuses, I used a function that would remove stop words, whitespace, and punctuation, as those wouldn’t be of much interest when we tried to find the counts of words that were prevalent among NBA fans.
Lastly, with regards to text cleansing, I changed the text to lowercase. That way, “Westbrook”, “westbrook” and “WESTBROOK” can all be considered as the same word when we build a word cloud.
What is a wordcloud?
A word cloud is an image composed of words concerning a subject of interest.
In our case, we’ll be producing word clouds that contain words that are often found when searching for “PlayerX+MVP”.
Furthermore, when viewing (and attempting to decipher the word cloud visualization), remember:
- Words that are most often found in review will be larger, brighter, and closer to the center of the diagram.
Take a look at some of the clouds for our five designated superstar candidates. Keeping an eye out for some of the smaller words will give you insight about what NBA twitter fans think about when discussing MVP candidacy.
If you also see symbols, it is likely because emojis and other unorthodox characters weren’t compatible with this method of text mining. There could also be a few links in the diagrams if the links were often referred to when the engine searched for a “candidate+MVP”.
How about James Harden?
The most often word used when mulling Harden’s MVP candidacy (or berating him, which is also common for the internet) is “Westbrook”! This is truly representative of how competitive our race is.
Look at some others; you’ll see more of the following:
- Team, teammates – which could represent the extent to which Harden has a better assortment of talent has been proclaimed throughout Twitter
- Rigged
- Close
- Endheroball – Very intriguing
Speaking of Westbrook, what has Twitter said about Russell Westbrook?
Look no further than these 3 words: triple, double and average.
But how about some of these others? Westbrook-centric tweets generated the most intriguing word cloud.
- Congratulations – it seems as if a large sum of individuals believe that Westbrook has already taken the crown.
- Durant – Because Russell’s current consideration could be fueled by narratives involving Kevin Durant’s departure
- “Y’all” – By the way, did y’all see that shot Westbrook had against the Denver Nuggets?
- Officially – This could refer to how Westbrook officially qualified for a triple double season average.
- “Statpadding” – Hey….
- Matter – Hmm….
How about some other candidates who haven’t been giving the same campaign pedestal?
Kawhi Leonard’s NBA Twitter search was often met with terms, such as:
- Untold
- Unheralded
- Defensive, power, strong, stopper
- Contested
LeBron James’s wordcloud:
- Passionately
- Unbelievable
- Anyone – Potentially written for a tweet, such as “Why would anyone else win the award? LeBron is the league’s best!”
- Winner
- Every – Should he win the MVP every year, as some contend?
Stephen Curry’s cloud:
- Reigning
- Doesn’t – Geez, that seems harsh
- Straight
- Unanimous
- Outplayed – which could refer to his most recent stretch against several other MVP candidates
Sentiment Analysis
Beyond the word clouds, I sought a way to ascertain if NBA fans referred to these candidates fondly, sardonically, slanderously, etc. I used the RSentiment package for this pursuit.
This method isn’t golden, but it’s going to produce some funny/insightful results.
Six potential sentiment categories exist:
- Sarcasm
- Very Negative
- Negative
- Neutral
- Positive
- Very Positive
There are preset words in the package such that if a tweet has words or phrases that represent the sentiment that the preset words are generally associated with, the sentiment calculation will lean toward one of the six categories. For example, “like” may be a preset word in the RSentiment package; because it is most often associated with positivity, a tweet that uses this in abundance will likely be “positive”. The package is a little quirky, so there will be times when we would like to impose a different classification on the program’s interpretation.
For example: “Allen Iverson was an inefficient player, but he’s still worthy of our respect.” – Sentiment: Neutral
I think that’s fair here.
I chose a random sample of 20 tweets for each player (which isn’t quite enough for a comprehensive look, nevertheless, we can get a little insight about how NBA fans feel about an MVP candidate). Let’s get a feel for the sentiment surrounding each player and then compare their samples to each other in a graph.
James Harden, Sentiment = Neutral
(This was likely the result because he received a small complement, although the tweet was more of a Westbrook praise than anything else.)
@russwest44 better get mvp. Harden good but he ain’t westbrook. ???? pic.twitter.com/CRvmIWC1yv
— Black&Woke? (@RealTru225) April 8, 2017
Russell Westbrook, Sentiment = Positive:
(yes, this was one of the 20 random selections)
I said soon after Durant left that Russell Westbrook would average a triple double and win MVP. Congratulations, Russ.
— Skip Bayless (@RealSkipBayless) April 8, 2017
Kawhi Leonard, Sentiment = Sarcasm:
@vedangraul all the mvp candidates have been invisible against warriors. kawhi harden westbrook. Only my guy isaiah thomas showed up :P
— Sachin Panda (@PkumarSachin) April 1, 2017
LeBron James, Sentiment = Very Positive
Lebron is the best player in the world, having his best season ever, MVP. Don’t @ me
— PG 14 ?????? (@PG13PG14) April 6, 2017
Stephen Curry, Sentiment = Sarcasm
(wonder if the program believes that Stephen Curry is overrated):
@DamonBruce Is it to late to include #Curry in the MVP race since he’s been the best player in the league the last 2 months? #2thetweets
— Peter James (@petoejames) April 7, 2017
Here’s a chart that tracked the calculated sentiment over the sets of the tweets. Who came out most favorably among them? Least favorably?
You can enlarge this graph, or view it in Tableau, if you’d like.
Most/Least Favorable According to NBA Twitter
The least favorable reviews likely belonged to Harden. It just appears as if NBA Twitter is conditioned to refer to him with a little austerity at times. But of course, the sample size is quite small; this concern might not be universally true.
In contrast, it appears that LeBron James-related comments registered most reverently among the five. Rightfully so, he’s consistently shown us that he has an extra gear. There were only two tweets among the 20 that were considered negative by RSentiment.
If I had to choose a 2017 NBA MVP winner based on NBA fan sentiment (not necessarily my personal choice), then it would likely become Russell Westbrook. The fascination with a 30-point triple-double season average is proven to be rampant among fans on Twitter. Additionally, his ability to squirm out of precarious situations with clutch fourth-quarter comebacks adds to the allure. Some enthusiasts go as far as to say that we should already congratulate Russell Westbrook for his award. We’ll see in June, I suppose.
Keep it up NBA Twitter. Maybe we can create a database that will give the community a bit (very minuscule) of agency in the actual race at some point in the future.
