Impact Report 2015

Inaugural National Data Science Bowl

Using predictive modeling and next-generation analytics to track ocean health

When you’re splashing in the waves at the beach, it’s hard to imagine the complex ecosystem that swirls around you. Researchers at the Hatfield Marine Science Center in Oregon, however, have been capturing underwater images of tiny plankton to understand, at a microscopic level, what’s happening beneath the waves.

Although you can barely see them, the tiny plankton are critically important. They uptake 25 percent of carbon dioxide released from burning fossil fuels every year and form the foundation for marine and terrestrial food chains. Because they are susceptible to small changes in temperature or water chemistry, plankton populations serve as an indicator for broader ocean health, with rolling impacts on the environment, economics, and even tourism. Unfortunately, critical insights are slow to reveal themselves; a day’s worth of images can take up to a year to analyze manually—far too long to be of real use in protecting the oceans, forecasting changes, or correcting problems.

Booz Allen, in partnership with Kaggle, the world's largest community of data scientists, knew that predictive modeling and next-generation analytics could give Hatfield researchers a leg up. The emerging practice of data science looks for patterns and asks the kinds of questions that make sense of the world—including the oceans. Thus the challenge for the inaugural National Data Science Bowl was created. With more than 5,000 entries, the Booz Allen-sponsored National Data Science Bowl helped scientists find new tools to understand changes in oceans’ makeup.

The 90-day competition provided seasoned pros and those new to data science alike the opportunity to use data science skills to have an impact on a global scale. Participants were tasked with examining nearly 100,000 underwater images and developing an algorithm that would enable Hatfield researchers to identify and monitor planktonic organisms at a speed and scale never before possible.

The winning entry came from the other side of the Atlantic Ocean. Team Deep Sea, from Ghent University in Belgium, developed the most effective algorithm to automatically classify more than 100,000 underwater images of plankton. The work by the seven-person team of graduate students and post-doctoral researchers automated the classification process for the first time in history. Together, the team beat more than 1,000 other teams and achieved better results than 15,000+ competing submissions in quickly and accurately distinguishing 121 distinct categories of oceanic organisms.

Through their work, they have enabled the rapid analysis of data sets with a digital size equivalent to 400,000 3-minute YouTube videos, or about 800 days’ worth of viewing.

For researchers at the Hatfield Marine Science Center—and the larger marine research community—the algorithms created through the National Data Science Bowl will be critical tools for assessing ocean health, allowing rapid assessment of their plankton populations. Already scientists at Hatfield are testing and validating the algorithms and are looking to expand on the research.

Booz Allen has built a team of data scientists focused on the “X” factor of the future—curiosity. That curiosity, in combination with technological progress, has given rise to a new community of problem solvers. While historically, major innovations have been the work of a few; the emerging data science field is inherently collaborative, driven by the first true generation of data scientists who are unprecedentedly networked and socially responsible.

Sponsorship of the National Data Science Bowl is part of Booz Allen’s ongoing commitment to support data science education and awareness, an initiative that has included the firm’s Explore Data Science training program and The Field Guide to Data Science.

“With more the 5,000 entries, the Booz Allen sponsored National Data Science Bowl helped scientist find new tools to understnad changes in the ocean's makeup”

Inaugural National Data Science Bowl

Using predictive modeling and next-generation analytics to track ocean health

Tags

Related Content