March Madness on Twitter

Wisconsin Badgers vs Duke Blue Devils April 6, 2015

Yedurag Babu, William Murphy, Jordan Huff, Huw Smith, Fadel M. Megahed

Data Analysts at the Samuel Ginn College of Engineering, Auburn University

Email: datasci@auburn.edu

Real Time Tweet Visualizations

The realtime visualizations below are made possible by dc.js , which is a cool javascript library with native cross filter support for exploration of multidimensional datasets. The map is supported by Mapbox . Other javascript libraries used here are d3.js , a highly popular javascript library which has revolutionized data visualization and crossfilter.js , which enabled fast multidimensional filtering for coordinated views. Server support is provided by Heroku , which is a cloud application platform that can host applications in Node.js, Python, Ruby and many other languages. The servers are coded entirely using Flask, which is a python microframework based on Werkzeug and Jinja 2. The database support is provided by ClearDB, a reliable, fault tolerant database-as-a-service for MySQL powered applications. In a future release of this tool, the database support will be provided by a PostgreSQL database hosted on Amazon (AWS). The sentiment of the tweets are approximate values calculated based on the tool VADER Sentiment Analysis, which is one of the best social media sentiment analysis tools on the web. Please see the below cited paper for further details. **Hutto, C.J. & Gilbert, E.E. (2014). VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. Eighth International Conference on Weblogs and Social Media (ICWSM-14). Ann Arbor, MI, June 2014.**

The 2015 NCAA Men’s Basketball Championship game will be played on April 6th between the Wisconsin Badgers and the Duke Blue Devils. This visualization package aims to allow fans, the media, and any other viewers to monitor social media activity (specifically Twitter) during the game to correlate with in-game events. We hope that this tool will enhance the viewing experience of the game for anyone who uses it, or at least make it a little bit more fun! You can track the public’s reactions to game events by reading actual tweets, see which regions of the country light up for certain things that happen, and even how people feel based on the sentiment of the tweets. The entire visualization is interactive so users can drill down into more depth easily.

Real time tweet distribution and sentiment on map

This map displays the geographic location of only geotagged tweets on an interactive map. Sometimes the map may be blank because there wont be any geotagged tweets now. The color of the markers indicate the sentiment associated with a particular tweet: red indicating negative sentiment, blue indicating neutral sentiment, and green indicating positive sentiment. The markers can be clicked on to view the specific tweet.

The tweets are slowly coming in. Please wait or visit March Madness to open up a window so that you can enjoy the match and see the twitter reaction side by side. It won't take much time.

Tweets (Only latest 10 tweets shown)
Tweets distribution by team

The charts will take about 25 seconds to start rolling. By clicking on either segment, the other viz will update.

Tweets sentiment Binned

Click on the segments to see other viz change.

Teams' tweet counts over time.

Team’s tweet counts over time displays a real-time run chart of the count of tweets over team separated into 30 second intervals. This visualization is linked to the donut graphs above.

Brush over the chart to see the teams' sentiment for that time period. The charts will take 25 seconds to start rolling.

Teams' sentiment over time.

Team’s sentiment over time shows the average sentiment of the tweets gathered in each 30 second interval displayed. Unfortunately, this chart is not linked with the others :-(. This chart is interactive :-) - place your cursor on the legend for each team to see that series alone.

Brush over the chart to see the teams' sentiment for that time period. The chart will take 25 seconds to start rolling.

Reset the Viz