Undergraduate Students Predict 2016 Presidential Election at Inaugural UNH Hackathon
On Friday, October 21, 2016, our team along with almost 40 other students gathered at the Peter T. Paul Entrepreneurship Center (ECenter) to compete in the first ever ECenter hackathon. It was open to all students and sponsored by Datawatch and IBM Watson Analytics.
A hackathon is an idea competition, often involving technology, where teams compete to problem-solve in innovative and exciting ways (it does not involve computer hacking, a common misconception). Each hackathon has a theme and this event’s theme was the 2016 U.S. presidential election. The task was to use IBM Watson Analytics and Datawatch Monarch software tools to find insights and reach conclusions pertaining to the 2016 United States presidential election.
The hackathon spanned from 1:00 PM on Friday to 11:00 AM on Saturday. Datawatch and IBM flew in teams of experts and they kicked off the event by teaching all of us how to use their respective advanced data analysis and visualization tools. At around 4:00 PM on Friday afternoon, the teams were set loose to find a quiet place to work with the software and attack the political data.
Working in our team of four, one of the first things we noticed once we delved into the data was the strong difference in the behavior of the support base of each candidate. This led us into deeper analysis on the behavior of these support bases to find trends and project them out over time.
Our presentation and analysis focused on the makeup and behavior of each candidate’s support bases with our findings predicting that Donald Trump would win the election.
We saw that many of Trump’s supporters decided early on that they didn’t want to vote for Hillary and were not hugely affected by the hype surrounding his campaign stunts. This was in contrast to the volatility we saw in Hillary’s supporters, whose poll numbers were extremely unstable by comparison despite her overall poll lead. We projected out this trend and predicted that the poll numbers would equalize shortly before election week, which looking back is exactly what happened. Hillary’s sporadic support base ultimately prevented her from being elected.
First thing on Saturday morning, the teams were asked to submit their analysis and findings via slide deck for the judges to review. The judges were Laura Trouvais of IBM, Dan Potter of Datawatch, and Andy Smith of the UNH Survey Center. The judges had a tough decision to make and deliberated for over an hour! At the end of the event, our team, comprised of Sam Warach (’17 Paul College), TJ Evarts (’20 CEPS), Max Miller (’20 CEPS), and Brandon Allen (’18 Paul College), was chosen as the winner.
The first-place prize was an all-expense paid trip to IBM’s World of Watson conference in Las Vegas, Nevada and licenses to IBM’s Watson software package. This package includes cutting edge software such as IBM Bluemix Framework, Watson Analytics, and Watson Cognitive.
We truly enjoyed working with IBM’s products as well as the Datawatch Monarch platform. They are easy to understand and use but allow the user to do really advanced manipulations and analysis.
Attending the World of Watson conference was an incredible opportunity. We were able to meet some talented people and see some impressive work, and we look forward to seeing how Watson will help shape the future of innovation.
When not focusing on our studies, all four of us are aspiring entrepreneurs and currently working on our own start-ups. We are passionate about changing the world and recognize entrepreneurship as the vehicle to make that happen. We are excited that UNH has invested in building the Peter T. Paul Entrepreneurship Center to foster a community of innovation on campus. It’s a great resource that gives us access to professional advice, programs, and events like the hackathon to help develop our ideas.
Sam Warach (’17 Paul College), TJ Evarts (’20 CEPS), Max Miller (’20 CEPS), and Brandon Allen (’18 Paul College)
To learn more about the winners’ pursuits, see their LinkedIn profiles below: