How much food is enough?

How much food is enough? Food insecurity is a part of today’s landscape, and looming larger by the moment. Here’s how Denison students are helping address the problem.
issue 04 | winter 2016
UnCommon Ground - How much food is enough?

Think data analysis, and you might picture statisticians sifting through rivers of neatly ordered numbers (think lab results, smartphones, sensor data—the ocean of information that is the internet) for insights into everything from epidemics to online shopping trends. But as some data analytics students recently learned, there’s another, equally challenging problem: no data at all.

Funded by a grant from the National Science Foundationsupported program PIC Math (Preparation for Industrial Careers in Mathematical Sciences), Matt Neal, professor of computer science, mathematics, computational science, and data analytics, devised a prototype for the capstone course in Denison’s new data analytics major: teams of students working to solve real-world problems suggested by alumni from different fields.

One of those alumni, Nic Covey ’04, is president of Denison’s Alumni Council and a vice president at Nielsen, the global information company that monitors consumer behavior around the world. He also directs Project 8, a joint effort of Nielsen and the United Nations to anticipate the needs of the 8 billion people who could inhabit this planet by 2025. Covey expanded the time horizon to give Denison students his challenge: How can we estimate the world’s food needs in 2050, when, the U.N. predicts, fully 9 billion people will be wandering Earth?

“That’s a weird problem,” said Neal. On the one hand, he explained, it’s extremely open-ended: Food need (how much people need to consume) is different from food demand (how much they choose to consume at a given price), so simply writing a definition was tricky. On the other hand, a person’s food need depends on a number of variables, from height and weight to body mass index and exercise level. Yet for many countries, those data simply don’t exist.

So Neal’s students got creative. First, they scoured the existing literature and compiled as much data as they could from countries where it was, in fact, available. Then they used that data to extrapolate the information they needed for countries where the data didn’t exist.

They realized that the limited nature of the data they had would make it difficult to apply standard statistical methods. So they turned to so-called nonparametric ones, sophisticated mathematical techniques that yield more accurate results when dealing with small or poor-quality data sets.

Given the uncertainty involved, it’s unlikely that Neal’s students will have come up with the definitive answer to Covey’s question. But just by thinking outside the box, they might have helped push the conversation on projected global food needs to a new level. And they have demonstrated what kinds of data will be required to arrive at more accurate predictions in the future.

Numbers at Work

The other six students in this class addressed different data issues. They were paired with two alums, renowned criminologist Lawrence Sherman ’70 and State Auto Insurance Product Manager Barnaby Nardella ’02. Sherman partnered his group with a colleague working for police, fire, and emergency services in Australia’s Northern Territory. Together, they analyzed police reports to predict the severity and frequency of domestic violence incidents in the area. And, as with many real-world problems, finding an answer wasn’t cookie-cutter. It took almost the whole semester to convert the data into a workable form, but the students’ diligence allowed them to make some interesting predictions that could inform future policing strategies. It also taught the students a thing or two about messy real-world problem-solving.

The second team, paired with Nardella, sought predictors of customer risk using a data set containing fender-bender information (accident rates, severity, etc.). Students started by writing computer code to extract and order the data based on ZIP codes. Then they further organized accident rates by variables, including traffic, weather, and population characteristics, to predict the likelihood of such accidents in the future. According to Neal, this analysis is exactly the type of task and math that data analysts in the auto industry perform regularly. The group presented its results at MathFest along with their classmates in August.

This fall, Denison launched a new data analytics major to prepare students for a number of professions, including business, marketing, finance, public health, government, education, social policy, research, law, and medicine. To learn more, visit:

Published December 2016