Three new multi-disciplinary, student-inspired “big data” research projects are the winners of grant funding from the U.Va. Alumni Association’s Jefferson Trust in the University of Virginia's first Big Data Collaborative Graduate Student Research Competition.
The winning projects deal with such diverse data-heavy challenges as bringing new architecture-inspired visualization to biomedical research, ultimately resulting in a new course; improving transportation sustainability by mining existing data from traffic cameras; and measuring the effects of student efforts in online education, using massive open online courses as a laboratory.
Each graduate student in the grant-winning research projects will earn a $20,000 stipend. Undergraduates receive $5,000 for summer salary or work-study.
An additional request for proposals to fund one or two more big data projects likely will be issued in the early fall, said Rick Horwitz, associate vice president for research and biosciences, who spearheads the University’s big data initiative.“We were trying to get people to think outside of the box regarding research, and that’s what has occurred,” Horwitz said. The initiative is designed to bring faculty, graduate and undergraduate students and administrators together from across disciplines and administrative units to develop services, curricula and new research activities related to complex data sets.
“To solve society’s most pressing challenges, we need to develop new ways to acquire, analyze and make sense out of big data,” Sullivan said. “The generous grant from the Jefferson Trust will foster interactions among our students and faculty who produce, manage, analyze and visualize big data sets.”
Grants for big data research projects – with heavy student participation – were an idea that grew from the “Big Data Summit” held in May 2012 and sponsored by the Office of the President, the Office of the Vice President for Research and the U.Va. Alliance for Computational Science and Engineering.
A second summit, “Big Data 2,” will be held May 14 from 8:30 a.m. to 1:30 p.m. in Rice Hall. Like last year’s inaugural event, the sequel is designed to bring together people from around Grounds to represent a diversity of big data activities and to brainstorm and develop new ideas for conducting new and better research around the reams of data that roll in by the moment in projects across the University.
“We realized last year that people who have problems handling big data were not in touch with the people who deal with managing big data,” Horwitz said. “Our goal is get people talking to each other, and to present opportunities for our students to help lead the way on innovating data-heavy projects.”
The projects are designed to allow the students to take leadership roles as part of the interdisciplinary teams, exposing them to cross-disciplinary collaborative experiences.
The following are brief descriptions of the winning projects.
Replacing the Bar Graph
With the advent of computational techniques for analyzing complex, large-scale systems, researchers have begun generating masses of data describing how interactions in systems occur across two- and three-dimensional landscapes. But techniques for representing multi-scale data sets have been outpaced by the techniques used to generate them. As a result, young investigators increasingly need training in data visualization methods.
Biomedical engineer Shayn Peirce-Cottler in the School of Engineering and Applied Science and architect Jeana Ripple are collaborating with each other and graduate students Joseph Walpole in biomedical engineering, and William Green and Elizabeth O'Brien in architecture, along with rising third-year undergraduate biomedical engineering student Angela Jividen, on a case study that merges biology, engineering and architectural representation techniques. That study will be used to create an interdisciplinary graduate and undergraduate course for students interested in complex data analysis and representation.
Peirce-Cottler has extensive experience studying and modeling the development of blood vessel networks in tissues, which have their own three-dimensional architectures, while Ripple has expertise in computer science engineering, architecture and 3-D computer modeling. Working with their students, they will envision new methods for displaying and communicating large biological data sets that describe how blood vessels in the body alter their form and function depending on the tissue environment.
The data visualization techniques they develop will also be tested using other complex systems, including ecological systems affecting human built environments. The work will lead to a new course that would provide students with hands-on experience with the techniques and software used in the case study for a variety of potential multidisciplinary studies.
MOOCs as a Laboratory
A project to measure the impact of MOOCs will push the limits of existing data-analysis techniques and potentially present challenges that will require new ways of thinking about data analysis.
“The hope is that this work will help to identify good study habits in the online course environment, and course policies that encourage those habits,” said statistics professor Dan Spitzner, who is the project’s faculty co-sponsor, along with Curry School of Education professor Sarah Turner.
The project – initiated by Paul Diver, a graduate statistics student, and Ignacio Martinez, a graduate economics student – will record students’ efforts in a massive online environment, including time-stamped participation logs, records of transactions with online course materials and information about the geographical location and background preparation, which together provide detailed digital profiles of students.
The goal is to enhance educational achievement, looking at how much effort a student exerts, how it is distributed over time and how grading policies and the instructor’s effort at reducing uncertainty affect the effort put forth by the student.
“One of the neat dimensions of this project is that it sets forth frontier research questions while also addressing important policy questions about teaching and learning,” Turner said.
Improving Transportation Sustainability
Civil and environmental engineer Andres Clarens and electrical and computer engineer Scott Acton are working with electrical and computer engineering graduate student Emmanuel Denloye-Ito and civil and environmental engineering graduate student Conrad Gosse on a project to improve bicycle transportation by sorting through masses of traffic camera data, then using that data to create computer models that could provide insight to complex traffic patterns.
They say bicycle transportation is affected by subtle changes in lifestyle, trip origination and destination, weather, time of day, infrastructure and culture in ways that confound traditional automobile-centric transportation engineering.
Their goal is to glean new understandings of bicycle traffic flow that could eventually facilitate improvements to road and path infrastructure, ultimately encouraging more and safer bicycle ridership for commuting in urban areas. The computer model they plan to develop would be applicable to cities throughout the United States and worldwide.