A new cross-school initiative at Tufts aims to change how we teach data science in STEM and non-STEM disciplines alike
Big data permeates our lives these days, in everything from health care to marketing to voting. But without context and a story to tell, data falls short in driving substantive change. Data scientists and engineers must be able to explain their work with narratives that are true to the data, while policymakers must understand data fundamentals in order to build persuasive cases for effective programs.
“No one ever made a decision because of a number alone,” says Shafiqul Islam, a professor in the School of Engineering’s Department of Civil and Environmental Engineering and at The Fletcher School. “Numbers can be used to justify what we do, but numbers do not make decisions. We can’t simply rely on numbers in decision-making—we must also consider prevailing social and political conditions.”
A new Tufts initiative—BigData@Tufts: Educating Policy-Savvy Data Experts and Data-Proficient Decision Makers—aims to do just that. An interdisciplinary Tufts team will teach engineers, scientists, and policymakers to extract actionable information from data.
Led by Islam and four co-principal investigators (PIs) from the School of Engineering, the School of Arts and Sciences, and The Fletcher School, more than 140 Tufts graduate students will be trained through the initiative over the next five years. The initiative is funded by a $3 million National Science Foundation (NSF) Research Traineeship/Harnessing the Data Revolution (NRT-HDR) award.
Each year, teams of students and faculty will consider a complex data question related to resources like food and water. Working with data professionals from both the private and public sectors, students funded through the grant and other interested graduate students will learn technical, team-building, and professional skills through studying practical scenarios. For instance, the focus might be on the lead contamination of the water supply in Flint, Mich., focusing on factors such as material corrosion of lead pipes and moral corrosion of societal and political systems.
“Students will gain invaluable experience in analyzing the intersections of policy and data in real-world scenarios,” says Jonathan Lamontagne, an assistant professor in the Department of Civil and Environmental Engineering and co-PI on the grant, who studies water resources and decision-making under uncertainty. “As climate change continues to affect natural resources, we need future planners, data professionals, and policymakers who are able to translate data into actionable information.”
With problem-focused immersion, data-centric projects—as well as symposia—and modular course elements designed to promote flexibility, the BigData@Tufts initiative will take an interdisciplinary approach to teaching a new generation of science and policy leaders.
Rather than requiring extra classes for students, initiative leaders plan to create a series of scalable teaching modules that can be offered as standalone short modules or incorporated into existing courses, on topics including machine learning, communication with broad audiences, and data visualization and visual analytics. The modules won’t just benefit Tufts students—they will also be transferrable to other universities.
It’s not only about words and persuasive verbal arguments. Co-PI Remco Chang, an associate professor in the Department of Computer Science, will be teaching a module on database design and tools for visual analytics. “Visualization can be a tool to augment human intelligence—helping analysts find patterns in large amounts of data. In addition, visualization is increasingly becoming a critical tool for communication,” says Chang.
Students will learn how visualization can help disseminate information and enable people and policymakers to understand complex situations—like the transmission of SARS-CoV-2 during the COVID-19 pandemic—and make the most informed decisions possible, he notes.
“For many of us, we were educated in a certain way, and we’re now retrofitting or adding on to our toolkits,” says co-PI Abani Patra, director of Tufts’ Data Intensive Studies Center (DISC) and Stern Family Professor in the departments of Computer Science and Mathematics. “We want our students to build their skillsets from day one, so they can go far beyond what we have been able to do.”
Tufts’ strong commitment to advancing the fundamentals of data science got a recent boost from the launches of DISC and the T-TRIPODS Institute, both of which are part of the BigData@Tufts initiative. Multiple Tufts schools have recently launched academic programs in data science and data analytics. The BigData@Tufts team draws on that history and unites Tufts expertise in data science with that in water resources, environmental studies, food and energy, learning, and pedagogy, among others.
Such an interdisciplinary approach is particularly apt for data science, which as an emerging area incorporates tools and techniques from probability and statistics, computer science, engineering, mathematics, and many other interfacing disciplines.
“It’s terrific to see an interdisciplinary team of scholars taking on this kind of educational experimentation and innovation, trying to expand the sorts of possibilities we offer students and for learning about the important area of data science,” says co-PI David Hammer, director of the Institute for Research on Learning and Instruction, professor and chair in the Department of Education, and professor in the Department of Physics and Astronomy.
In the end, it comes down to a creative blend of numbers and narratives. “Often numbers are objective but not very persuasive,” says Islam. “We have to communicate numbers in a way that is true to the narrative we are telling. What we want our students to do, really, is not only to understand the complexity of the numbers, but also tell stories of the numbers in a compelling way.”
Lynne Powers can be reached at lynne.powers@tufts.edu.
