University of Wisconsin-Madison Illustrates Why Adopting GIS in Data Science Programs Works

Modern technologies have produced a flood of big data. Before this information can be used to drive better decision-making, data scientists must first make sense of it. Anticipating growth in data science demands in the labor market, the University of Wisconsin-Madison (UW-Madison) has created a dedicated undergraduate data science major with one key differentiator: a geospatial data science track that introduces students to geographic information system (GIS) technology.

Dr. Song Gao, an associate professor of geography at UW-Madison and lead for the geospatial lab, helped spearhead the adoption of spatial data within the course curriculum of the new major. Established in autumn 2020, the data science major has quickly become one of the fastest growing at the university, with more than 900 students enrolled. Gao says access to cutting-edge GIS technology and understanding how to use it in data-rich environments are key to setting students up for success.

"For data science students, adding a geospatial lens can add nuance and complexity to the topics they are already exploring, and it can help contextualize their work through data management, analysis, and mapping and visualization tools," said Gao.

Gao's research and his course, Advanced Geocomputing and Geospatial Big Data Analytics, show students how geospatial data and data science relate. He advocated for his course, alongside two other GIS prerequisite courses, to be included in the major. Gao's goal is to ensure data science students learn basic geospatial principles and advanced geospatial analytics to meet emerging workforce needs.

"The increasing availability of geospatial data across industries is why we need to introduce the technology and emerging spatial analysis methods," said Gao, adding that his course covers the life cycle of data—from storing to processing and later analyzing and visualizing.

When Gao's course was approved for data science major students, he had to consider how to make his curriculum accessible for both data science majors and traditional geography students.

"I have different groups of students with different mindsets and skills. I realized I needed to acknowledge those differences and create a curriculum that addresses their strengths and weaknesses, so they can learn from one another," said Gao.

Gao structures his course to include lectures, classroom discussions, coding experiments, and lab work. This structure creates a collaborative and hands-on learning environment for all his students. Both groups are given the same exercises and lectures, but Gao uses different module design principles for each.

"Those who are skilled in Python coding can use ArcPy and ArcGIS API for Python on Jupyter Notebooks, while those not skilled in programming can try ArcGIS Pro or ArcGIS Online at the beginning," said Gao. "By the end of the class, both groups will converge on ArcGIS Notebooks."

Gao uses the student body's diversity in class and lecture activities to create group discussions framed around broad geospatial and computational questions.

"On the one hand, the curriculum should be designed so that students are seeing and learning something new. But it also needs to encourage networking and communication, so I ask questions that call out different groups of people in the hope that students will share their thoughts and learn from one another," said Gao.

Gao references the Higher Ed Guide to Esri E-Learning for Spatial Data Science to design the overall module structure for his course. His curriculum includes three modules, each one building on the last and introducing students to new technologies like ArcGIS Notebooks, ArcGIS Online, and ArcGIS Pro. For example, in the first module, students learn about data structure and how to convert datasets from organizational formats like vector and raster.

In the second module, students grow proficient with analysis methods for processing and analyzing geospatial data in Python with technologies like ArcPy as well as open-source libraries. Using the skills from the first two modules, students then study a special topic for the third module, like remote sensing imagery, GPS trajectory, or geospatial semantics.

Gao says using online GIS tools offers more accessibility for students to experiment with data and pretrained models for emerging techniques.

"With the popularity of machine learning and GeoAI, it was important to familiarize students with these topics," said Gao. "Using ArcGIS Notebooks and pretrained models in ArcGIS Online, I can demonstrate to students those machine learning techniques for the GeoAI topics, including clustering and classification, and then for the geospatial phenomena."

Between modules and lectures, Gao also invites industry guest speakers to share with students what they do.

"I think that is important because it shows students what a data scientist's job is like and helps them understand how geospatial data science connects to the data science industry," said Gao.

Spatial data science that uses geographic location to explore spatial patterns, trends, and connections can explain the "why" behind many of today's most pressing environmental and business questions. Many of Gao's students have offered feedback, stating that taking the course provided them with the necessary skill set for real-world applications. In other instances, the course helped them find jobs.

Catherine McSorley, a product engineer at Esri, is one of those students who has benefited from learning spatial data in Gao's course. As an undergraduate at UW-Madison, McSorley majored in statistics, and during her senior year, she sought additional electives like cartography to upskill her discipline. Having enjoyed the course and seeing overlap with her studies, McSorley sought other geography courses including web mapping and Gao's course.

"A lot of my coursework was theory based. [Gao's course] looked interesting, like I could actually apply data science and analysis," said McSorley.

As a non-GIS student, McSorley appreciated the blend of enrolled students and the collaborative opportunities Gao created.

"I didn't really know geography or spatial science, but there were peers in the class I had met from other geography courses who I could rely on for help," said McSorley. "Looking back, it felt reflective of the data science world where you work with a lot of people from different backgrounds."

As part of the course, students create a final project that brings together the skills they've learned throughout. The project was a key moment outside of group discussion, where McSorley understood how to combine her statistics background with spatial science for real-world application. Inspired by a paper about patterns in taxis and ride network flows, McSorley wanted to apply the paper's concept to countries and migrant flows using a dimension reduction methodology (a favorite of hers at the time).

"I wanted to present something that was different from what my peers were doing and that played to my strengths," said McSorley.

The project and course experience overall were a highlight in her education journey.

"I felt motivated by the analysis, applications, and digging into a problem and applying novel research to a new method," said McSorley. "I hadn't felt that way about a class before. It was really the first time I felt passionate about something, which was really cool."

While McSorley would go on to pursue a master's degree in data analytics and work for a consulting company, the course stayed with her. She credits it for inspiring her to apply for her current position as a product engineer on the spatial statistics team at Esri.

Equally, she argues, it's vital that students are exposed to spatial data early on since everyone has data on something, and all data is spatial, so that they may succeed in their careers. McSorley said there's a tendency for data scientists to go through their university studies without ever encountering spatial data lessons, unless they happen to stumble on it like she did. She adds that universities could benefit from collaborating with their computer science and geography departments to encourage more cross-listed courses so students like herself can more easily upskill.

Whether it's incorporating GeoAI or high-performance computing with big data, technology will advance. Gao believes GIS is a fundamental tool for data science, and giving students geospatial fundamentals will help them succeed with whatever technological advances occur. He encourages universities to offer opportunities for students to learn the technology as a necessary tool for data acquisition, management, analysis, and visualization.

"It is really important that universities build a campus-wide interdisciplinary team to think about these technological advancements and build a relevant curriculum that is more accessible and not limited to one department," said Gao.

In looking to the future at UW-Madison, Gao hopes to expand the spatial computing and the geospatial data science track within the data science program to include more courses related to deep learning frameworks, R programming, and unmanned aerial vehicle imagery analysis. In doing so, UW-Madison graduates will be better prepared to embrace GeoAI trends, tackle high volumes of big data, and discover solutions to the problems of tomorrow.