TEDx Greensboro is a local event of TED Talks (TED = Technology, Entertainment, Design). I was fortunate to attend this inagural event in Greensboro (attendence was choosen through lottery). The theme was 'Dreamsboro' -- with the idea that presentations would focus on actions that make Greensboro, NC a place where folks want to live, work, play, retire, and do other good things.
In the first few minutes of the TEDx event, the moderator, Justin Catanoso asked that cell phones, etc. be turned off. So, in the first few minutes, there wasn't much activity. Then, slowly, folks couldn't contain their enthusiasm any longer and the tweeting began in earnest.
The viz disects the words used by the Tweeters as well is data about who tweeted, which speakers were mentioned, what tags were used. A user has the ability to see the Tweeters' profile, the tweet itself or open links mentioned within a tweet.
Now, let's talk about how this viz was built --
First, it was inspired by Andy Cotgreave's efforts that culled Tweets and used at the Tableau Customer Conference 2012, as well as other conferences (directions found here). To obtain the raw data, I installed Python and ran the Python script to create a CSV file of the tweets. That was the easy part.
I wanted to extend Andy's approach so that I could use Tableau 8's new word cloud viz. To do that, I needed to split the tweets into separate words. Likely this could have been done in Python, as likely could have the rest of the tasks, but I used what I knew well -- Excel. I created a macro that splits the tweets into words - each word placed into a separate column on the same line as the tweet. What did I have now? The basic structure needed to use the Tableau Reshaper Tool Add-In (found here). This then gave me a row for each word, along with the full tweet info.
Next, I found that many words were noise. Adjectives, pronouns, mis-spellings, etc. existed. A little manual effort identified these noise words and set them aside in a separate worksheet tab. I then connected Tableau public to the Excel, created a join condition that excluded rows with the noise words. From there it was just a matter of creating the word clouds, tweet timeline, and some parameter driven choices for how to present the data.
Did I mention that all the tools used for data acquisition, reshaping and vizualization are free? Python, Tableau Data Reshaper, and Tableau Public will handle up to ONE MILLION rows of data! You can use the above guidance to recreate a viz of your own against your own media event. I entered this into a 2013 Tableau User Conference contest, please tweet about the viz using the hashtag #TableauTEDxGSO.
Hope you enjoy!