Follow Datanami:
November 24, 2014

‘Datathons’ Emerge as Training, Recruitment Tool

As a big data ecosystem begins to flourish, “datathons” are emerging as a way to attract future data scientists since participants are required during competitions to collaborate with team members from other disciplines.

The latest example comes from Ireland where a university working with a regional bank recently awarded a “big data prize” to a pair of brothers studying the commercial applications of computer science.

The brothers, Sam and Jeremy Bowles, computer science students at the Dublin Institute of Technology and University College Dublin, won a £4,000 ($6,279) AIB Datathon prize earlier this month in a competition sponsored by Allied Irish Banks.

The inaugural competition drew students from across Ireland who were given a full day to analyze a dataset to build a predictive model. Competitors used the dataset of create a movie recommendation scheme based on 100,000 movie ratings from 1,000 users.

Competitors were tasked with building a movie recommendation model using data analytics techniques borrowed from streaming video services like Netflix and Amazon.com, including the presentation of related video “you may also like.”

With those instructions, teams from across Ireland were given the rest of the day to create the movie recommendation model.

Datathons along with data mining boot camps are catching on in Europe and the United States as a way to attract computer science students and train future data scientists. For example, the University of California at Berkeley data laboratory co-sponsored a similar competition in August in connection with the American Sociological Association’s meeting in San Francisco.

The UC-Berkeley competition focused on scouring municipal data from cities like San Francisco, Chicago and New York to gain insights into urban housing issues. The competition was open to social, data and computer scientists along with municipal employees, tech startups, graduate students and even data hackers.

The U.S. competition sought contestants who were experienced in research design, data management, statistical research, textual analysis and computer science. Contestants were required to work on teams of three to five members, each member from a different discipline.

Increasingly, financial institutions and other businesses are sponsoring datathons as a way to attract future data scientists. Jeremy Bowles, the co-winner of the Irish competition, said participating in the datathon “has really showed me that it is possible to pursue a career in this field.”

The Institute for Public Knowledge at New York University has been a key backer of the growing datathon movement. “Datathons allow social scientists to test new research ideas and meet potential collaborators in a working environment without requiring a great deal of commitment,” the NYU Institute notes. “Ideally, a datathon is an intellectual testing pit full of the data and constructive criticism it might take months to sort out otherwise.”

The Institute staged its first datathon last winter focusing on the impact of climate change on New York City, a hot topic in the aftermath of Hurricane Sandy in October 2012. The event was sponsored by Johnson Research Labs, the data science and consulting firm. Microsoft Research also participated in the climate data effort.

Among the issues considered during the climate change datathon was how meteorological and urban datasets could be combined to “meaningfully examine the impact of climate change on cities.”

Recent items:

A Deluge of Data and a Dearth of Data Scientists

Data Scientists—Who Needs Them Anyway?

Datanami