By Emma Folts
In 2014, the San Francisco Bay Area was rattled by its largest earthquake in more than 25 years. Of the more than 8 million affected people, about 200 were injured, and two were killed.
Earthquakes are not a distant threat for the United States. In the next 100 years, there’s more than a 95% chance that populous parts of the West Coast will experience a damaging earthquake, according to the U.S. Geological Survey. When an earthquake strikes again, the nonprofit American Red Cross will likely respond, mobilizing its disaster relief workers to open shelters and provide lifesaving aid to those affected.
That’s why the Red Cross asked a team of Heinz College students to develop a real-time forecasting tool to more accurately and efficiently predict shelter needs. The students created the tool as part of their capstone project –– a key feature of Heinz College’s core curriculum –– and presented their findings this spring. Now, the Red Cross plans to train its staff to use the students’ product.
The Red Cross wanted the tool to provide rapid estimates and be easy for its staff and volunteers to use, as many lack formal training in data analytics. Meriem Fouad, Miguel Rivera-Lanas, Sami Ouyang, Yusuf Surya and Heng Jiang, all 2025 graduates of the data analytics track of the Master of Science in Public Policy and Management program, made up the capstone team.
The Data Analytics track of our Master of Science in Public Policy and Management is our most quantitatively rigorous public policy program.
The Data Analytics track of our Master of Science in Public Policy and Management program will train you to harness the power of data and analytic technologies to transform organizations that serve the public good.
“I hope our tool is able to help the Red Cross mobilize resources and help people as fast as possible. Ideally, that would mean people can get in shelters the first night after the earthquake,” Fouad said. “To provide aid effectively, the Red Cross needs fast confirmation that they have enough volunteers and resources to respond.”
The students’ tool does not use software and does not need to be downloaded; instead, it uses Excel and Google Colab. After an earthquake, a Red Cross staff member can open the Excel interface and enter the earthquake’s identification number, which comes from the U.S. Geological Survey. Excel will generate a snippet of code that the staff member will paste into Google Colab, where the team’s Python model runs.
After running the code, Google Colab will generate a spreadsheet with estimated shelter needs for each affected area.
“Once an earthquake happens, and the American Red Cross has the identification number of the earthquake, they can use our tool and get an estimate of shelter needs in under five minutes,” Fouad said. “All they would need to use our tool is WiFi and a web browser.”
The Python model pulls and analyzes publicly available data, so the Red Cross doesn’t need to input data manually. To deliver accurate predictions, the model analyzes the number of damaged buildings and weighs other factors that could shape shelter needs, from the loss of electrical power to the ability of an affected resident to stay with loved ones. The team reviewed existing research and gathered insights from the Red Cross to learn what factors to consider.
Heng Jiang (MSPPM-DA '25) delivers part of the team's capstone presentation.
Creating the tool posed some challenges for the students. Few earthquakes have struck the United States in recent years, so the team initially had little data to feed a predictive model. They decided to incorporate earthquake data from other countries but had to exclude those with different building regulations, architecture and seismic codes, as these factors can shape an earthquake’s impact on shelter needs.
With the help of the Red Cross, the team identified a few additional countries that are earthquake-prone and are similar to the U.S. across several metrics.
Prior to the capstone project, the Red Cross would rely on its experts to estimate shelter needs based on population information, maps and their previous experience responding to natural disasters. This approach, though, could lead to over-or under-estimating shelter needs.
Michael Whitehead, planning integration manager at the Red Cross, has sought for the nonprofit to take a more data-driven approach to responding to natural disasters. To improve the earthquake response process, Whitehead asked his team: “Why don't we get a bunch of really smart college kids, who don't know this is really hard, to work on a possible solution?”
The Red Cross met with the capstone team every other week. In the meetings, the students shared progress updates and discussed obstacles they encountered, and the Red Cross provided guidance. As the project’s point of contact, Fouad would meet briefly with one Red Cross staff member each week for additional support.
“A lot of our conversations involved the students talking about ideas, and then on our side, trying to figure out what was practical,” said Red Cross volunteer Louis Luangkesorn.
Sami Ouyang (MSPPM-DA '25) during the team's capstone presentation.
Whitehead said he was very pleased with the capstone project.
“They found a research paper from Italy on estimating shelters after earthquakes. The authors had come up with a model for determining household habitability — and the students used that in their methodology,” Whitehead said.
Support from Mariana Escallon Barrios, an assistant teaching professor of information systems at Heinz College, was a key component of the successful collaboration between the students and the Red Cross. Escallon Barrios –– who met fellow Northwestern University alumnus Luangkesorn during her doctoral studies –– was the capstone team’s advisor.
The students’ focus “was always on providing a tool with the needs of the American Red Cross in mind,” Escallon Barrios said.
“They became experts on a topic that was not familiar to all of them, they found research papers and different sources of data, and put everything together to create the tool. Week after week, they would come up with new information and ideas that would make the tool’s estimates more precise. Their final product showcased their great work as a team,” she said.
She continued: “Guiding students on real-world impactful projects is always a pleasure. I enjoy being part of their thought process and seeing the great ideas they come up with.”
Right now, Whitehead and Luangkesorn are the only people at the Red Cross who know how to use the students’ tool. Going forward, they plan to create a procedure for using the tool that integrates with the nonprofit’s existing procedures for other disasters. They intend to train staff and volunteers on the tool and run regular practices with it, Whitehead said.
“It's going to be on the shelf there, ready, and in the case of an earthquake, we’ll break the glass,” Whitehead said. “There's no earthquake season –– the San Andreas fault could happen tonight. We've always got to be ready.”
Fouad said that Heinz College’s data analytics training enabled the capstone team to find a solution for the Red Cross.
“None of us came in with any knowledge of seismic analysis, or shelter-seeking psychological behavior analysis, but we were able to do them because of our Heinz College education,” Fouad said. “I think that helps us feel more empowered to go into different domains after Heinz, and know that we will make the impact that we hope for.”