Northwestern U Holds Big Data Analytics Hackathon
Northwestern University and Teradata held their second annual data
mining hackathon on April 29.
Thirty-six students in Northwestern's Master of Science in Analytics (MSiA) program
competed in teams of two to resolve complex hypothetical problems using the Teradata Aster Discovery
Platform. The platform is "an analytic engine designed for big data sets,"
according to information from the company. Participating students used
techniques such as pattern recognition, text analysis, graph analytics and
predictive computation to solve problems in the domains of social network
analysis, fraud detection, supply chain management, text analysis and money
laundering.
Judging criteria for the competition included:
- Creativity pertaining to data sets and analytic theory proposed;
- Use case concept;
- Use case approach in execution and multi-genre analytic techniques;
and
- Ability to demonstrate and tell the story.
Teams were allowed to choose from eight publicly accessible data sets,
including NFL statistics, MLB statistics, Amazon reviews, State of the Union
addresses, airline flight data, and United States consumer bank complaints, and
they were free to choose how they analyzed the sets and what insights to search
for. Participating students received three hours of training on the Teradata
Aster Discovery Platform the day before the event, and Teradata employees were
on hand during the event to answer software-related questions.
"It's beneficial for our students to receive a data set without
having a clear question to solve," said Diego Klabjan, director of the
MSiA program, in a prepared statement. "So then they have to be creative
about finding the problem and look for a strategy to solve it."
The winning team analyzed every State of the Union address since 1790, using
the software to label each sentence as positive or negative, and then comparing
the ratios. They found that the addresses throughout history have typically
been positive, regardless of the political climate of the time. The team
compared the address data to approval ratings from Gallup polls taken directly
before and after the addresses and found that more positive addresses
correlated to higher subsequent Presidential approval ratings.
The second place team analyzed customer reviews on Amazon.com to identify
those submitted by bots. Both the first and second place teams won passes to
the annual Teradata Partners Conference and Expo, a data analytics event being
held in Anaheim, CA in October.
Northwestern University and Teradata began collaborating more than three
years ago, when they developed new business and technology curricula designed
to help students learn the science and practice of business analytics,
including data warehousing techniques and platforms, according to information
from Teradata.
About the Author
Leila Meyer is a technology writer based in British Columbia. She can be reached at [email protected].