Carnegie Mellon University
August 05, 2020

Graduates Bring Power to the Solar Industry

Students tackle the energy market while completing their master’s degrees.

By Stacy Kish

Students completing their Master of Statistical Practice (MSP) at Carnegie Mellon University build real-world experience through capstone projects in industry and nonprofit organizations. Sabrina Zhu and two of her colleagues selected one project, which aimed to improve the forecasting in the power system. 

“It’s like the stock market,” said Sabrina Zhu, a 2020 graduate of the MSP program in the Department of Statistics and Data Science at the Dietrich College of Humanities and Social Sciences.  “Energy price is affected by many factors, such as the generation systems, different types of power being generated, capacity of the transmission lines, and so on. A lot of these factors are bound by the physical characteristics of the transmission grid. I was drawn to the complexity of this project.”

Prior to the 1990s, electricity demand was monitored on a daily basis and power supply was modified to meet regional demands. According to Zhu, this approach was not terribly efficient. In 1992, the Energy Policy Act opened the electricity market by broadening choices for utilities and creating new rate-making standards.

Zhu and her colleagues joined SESCO Enterprises, LLC, a FERC-registered power marketer, on a project aimed to improve energy market predictions.

SESCO does not participate in power generation processes. Rather the company examines subtle fluctuations in the power market based on different variables related to power usage to improve predictions and gain an advantage.

In the current energy market, multiple organizations, like SESCO, make bids based on real-time conditions. These virtual trades affect how power in a region is generated to ensure energy is released that will meet local needs in an efficient and timely fashion. 

To examine this process, Zhu and her colleagues built a regression model using historical data to simulate market conditions. They compared their model output to actual market data to evaluate the predictive power of their approaches. While the results were uneven, Zhu believes SESCO can build off of their work.

“Undoubtedly, fresh graduates and their employers often find that there’s a gap between solving hypothetical problems under the academic setting and solving real-life problems under the industrial setting,” said Jiang Shu, manager of quantitative trading at SESCO Enterprises, LLC. “We did not expect [the team] to solve the problem completely, instead, we hoped they can be creative and thoughtful on designing their approach. They have exceeded all expectations.” 

As part of the capstone, the team was required to present their on-going progress to the class. According to Zhu, this requirement forced her and her teammates to really think about the project and adjust their work to improve their results. It also provided professional training that honed their communication skills.

“It’s not like an assignment in class,” said Zhu. “It was exciting because SESCO gave us [a real-world problem] and the opportunity to explore the unknown with data science.” 

Zhu worked on the SESCO capstone project with her colleagues Andrew Follmann and Justin Kim.

“These capstone projects are a large part of what makes the MSP program so special,” said Jamie McGovern, Special Faculty, director of the Master of Statistical Practice Program.  “Students can truly understand what it means to work in data science on actual industry problems, with real data, with a real industry client, while industry partners can tap into the fresh perspectives provided by these highly capable and expertly-trained modern data scientists.”