February 12, 2019

New Duke program equips scholars to lead in emerging field of energy data analytics

Nicholas Institute for Environmental Policy Solutions

The last decade has ushered in astounding growth in the amount of energy-related data, along with game-changing developments in machine learning and other innovative data science techniques. The result? Unprecedented opportunities to analyze and make better decisions about how we generate, transmit, and consume energy.

At Duke University, a new program is pushing past traditional disciplinary boundaries to develop scholars prepared to seize these opportunities, deftly using data and advanced quantitative methods in pursuit of accessible, affordable, reliable, and clean energy systems.

The Energy Data Analytics PhD Student Fellows program, designed to support a cohort of four doctoral students in 2018-2019 and a second cohort of four in 2019-2020, is funded by a grant from the Alfred P. Sloan Foundation.

The Fellows Program is an effort of the university's Energy Data Analytics Lab, which is working to position Duke University as an international leader in the emerging area of energy data analytics. The Lab is a collaboration among three of Duke's signature interdisciplinary units: Duke University Energy Initiative, Rhodes Information Initiative at Duke, and Social Science Research Institute (SSRI).  

"The massive increase in data from energy system offers potentially transformational opportunities to address some of the most daunting energy challenges our world faces," observed Dr. Kyle Bradbury, managing director of the Lab. "The fellows program actively engages young scholars in that work, as they propose and tackle projects that bring together energy domain expertise with data science tools and research. They work with faculty from diverse corners of the university—beyond their own degree programs—as they learn to navigate interdisciplinary collaboration over the course of nine months."

In addition to funding equivalent to one-half of a full fellowship for an academic year, fellows receive conference travel support and data acquisition support up to $2,000, as well as priority access to virtual machines, storage, and other computational resources. The scholarship of the first two cohorts of fellows will be highlighted at a symposium hosted by Duke University in spring 2020.

A look at this year's fellows

The inaugural cohort of four fellows in 2018-2019 includes doctoral students in electrical and computer engineering, environmental policy, and computer science. This year's fellows affirm the value of a multidisciplinary approach to their research, reporting that the program has strengthened dissertation chapters, encouraged them to present their work at conferences, provided computational resources, and driven their engagement with real-world energy problems.

Bohao Huang is a Ph.D. student in electrical and computer engineering at Duke's Pratt School of Engineering. He is part of the Applied Machine Learning Lab at Duke and focuses on the translation of advanced machine learning techniques into practical solutions for challenging real-world problems. Specifically, Huang is developing algorithms to automatically extract useful energy systems information from large volumes of aerial imagery.

Huang reflected on the invigorating, interdisciplinary nature of the fellowship: "Even though the fellows and faculty are all doing research related to the energy field, the fact that we are coming from different departments gives us very unique skill sets. This helps me to expand my scope and rethink my research problems, considering them from different perspectives."

Qingran Li is a Ph.D. student in the University Program in Environmental Policy (economics track) offered jointly by Duke's Nicholas School of the Environment and Sanford School of Public Policy. Her research includes using analytical tools to understand behavioral responses to policies and developing interdisciplinary solutions to energy and environmental issues. Li is using smart meter data and behavioral surveys to develop a new algorithm to better identify residential usage patterns and estimate demand more precisely.

With financial support from the fellows program, Li traveled to the Southern Economic Association's November 2018 meeting to present preliminary findings from her research. She reported that conversations she had with others at the conference—including feedback on her research—have strengthened her sense that the fusion of data science techniques and energy applications will be increasingly crucial to successful work in energy economics.

Tianyu Wang is a PhD student in the department of computer science at Duke. His general research interests are in machine learning and applications of machine learning algorithms. Wang is using a reinforcement learning ("multi-armed bandit") approach to efficiently design the architecture of neural networks for energy domain problems such as energy demand prediction. 

The cohort format, bringing together a wide range of disciplines, skills, and perspectives, has enriched the fellowship experience for Wang. At monthly meetings, fellows alternate in presenting on their areas of expertise, from energy systems and power markets to GitHub and integrating version control tools into the research pipeline. Wang shared, "My fellow students are all very good teachers. I learn a lot from sitting with them in a discussion and they always have inspiring and sharp ideas." 

Edgar Virguez is a student in the doctoral program in environment at Duke's Nicholas School of the Environment. He is interested in contributing to the understanding of market mechanisms that facilitate the integration of variable energy resources. Virguez is designing quantitative tools that support the process of assessing policy and market approaches, promoting an increased penetration of variable energy resources in the energy matrix. 

As Virguez looks towards his future path in academia, he said he "will assign a higher value to programs that recognize the intrinsic value of multidisciplinary education," which he's come to believe is "essential to stay at the forefront of innovative training."

Brian Murray, director of the Duke University Energy Initiative, reflected, "For decades, Duke has advanced an interdisciplinary approach to putting scholarship to use in service of society. This program applies that idea to doctoral training for those exploring important problems in energy."

He noted that the job market for PhD students is broader than ever, so it makes sense for doctoral study to evolve with those changes: "This kind of cross-training will help equip scholars to make important contributions to addressing energy challenges whether they ultimately land in academia, government, nongovernmental organizations, or the private sector."   

Applications for the 2019-2020 cohort of Energy Data Analytics PhD Student Fellows are due March 1, 2019 at 11:59 p.m. PhD students from any degree program at Duke can apply. Interested students are invited to attend an information session on Feb. 22, or to direct questions to Kyle Bradbury, managing director of Duke's Energy Data Analytics Lab.