This is a guest post by Andy Cotgreave, the Social Content Manager at Tableau. He’s been with the company for three years, sharing his passion for data visualization and helping people see and understand their data. Prior to joining Tableau he was a data analyst at the University of Oxford. You can find him on twitter as @acotgreave.
What a four weeks it’s been. The FIFA World Cup has been the biggest social sport event ever and has provided endless opportunities for analysis of a truly global conversation. I’m Andy Cotgreave (@acotgreave), the social content manager at Tableau, and I teamed up with Datasift for the duration to delve into the data and see what insights we could find.
We found real business insight. We found fun facts. We discovered new ways of looking at events. Here are some of my personal highlights:
Which Hashtags Won the World Cup?
Traditionally, the only way to reach a global audience was via TV and official merchandise. Brands had to pay huge amounts to become official FIFA sponsors or partners, at huge cost. In a real-time social media world, however, that’s no longer the case.
Using just a basic stream (click here to see the one we used) to capture all brands’ World Cup hashtags, we discovered that Nike, not a sponsor, outperformed all but Adidas (when measured by number of Tweets with a hashtag). Obviously there are more ways to measure success, but on this measure, maybe brands no longer need to be a FIFA partner?
Which Players Got the Most Mentions?
Still thinking about brands, it’s vital that they choose the best players to sponsor. By looking at mentions of players, we discovered plenty of surprising facts.
England were hugely disappointing during the World Cup and suffered an embarrassing defeat by Uruguay. Despite that, England’s Wayne Rooney was mentioned much more on Twitter than Suarez, who played exceptionally during the match. This is good news for Nike, who sponsor Rooney.
Looking at players from the USA, we saw that goalkeepers can be more interesting than strikers. Howard made an amazing save during the USA vs Portugal. In fact, it was talked about more than Jones’ spectacular goal when the USA were 1-0 down.
For the player mention charts, we used the salience entities to identify topics.
What Does Sentiment Look Like During a Match?
In any sporting event, one set of fans are going to be happy, and another sad. Breaking down Tweets using Datasift’s sentiment analysis is an excellent way of doing this. During the opening match between Brazil and Croatia, we saw a clear display of unhappiness when a penalty was awarded.
Goal? Or Goooaaaalll?
Finally, my favorite exploration was inspired by excitable South American commentators. It turns out that thousands of people love to use lots and lots of letter for the word “goal” in their Tweets when a team scores.
For this view, we used this stream containing a regular expression to filter tweets containing any length of the word “goal”. After importing the data into MySQL, we calculated the length of the word using a SOUNDEX() function (I am grateful to Imranul Hoque for his example).
The streams we used are linked to in the text above. Depending on the volume of tweets shown in Historic Previews, we varied between a 1%, 10% and 100% sample rate. Data was stored in MySQL and explored in Tableau.