By Allison Leon
Geographic information systems (GIS) and spatial data science continue to transform social science research, and UChicago College students explored those areas of study through a new summer quarter course, Introduction to GIS and Spatial Analysis for Social Scientists.
As described in the College Catalog, the course goal is to “learn how to think about spatial aspects of research questions, as they pertain to how the data are collected, organized and transformed, and how these spatial aspects affect statistical methods.” That description, translated by Marynia Kolak, Assistant Director for Health Informatics in the Center for Spatial Data Science, who taught the class, is about “teaching what GI Science and spatial analysis are for social scientists and through a social science perspective. It's a new way of teaching – instead of just focusing on software and how-to's, we're diving into the theory and concepts behind it.”
While some students came with previous GIS background and had a sense of how it applies to their interests, the course allowed all to further plumb the possibilities of GIS.
“I had some experience with GIS before taking this course. I took Intro to GIS and Intermediate GIS with cartography, and I wanted to take this class because it's very much related to the work that I do in public policy,” said Tiye Stephens, a rising fourth year in the College.
Sam Joyce, a rising second year, added, “I was on the Calumet Quarter, which is one of the offerings through the Chicago Studies Program. As part of the Calumet Quarter, I took a class called Urban Design with Nature that relied a lot on GIS to examine the built environments of some of the neighborhoods on the South Side. That gave me experience working with GIS, but because it wasn't a full GIS course, we were mostly just using it for specific applications, so I didn't have the full introduction to GIS that I've been able to get through this course.”
The course explored methods such as spatial data integration (spatial join), transformations between different spatial scales (overlay), the computation of “spatial” variables (distance, buffer, shortest path), geovisualization, visual analytics, and the assessment of spatial autocorrelation (the lack of independence among spatial variables) to answer research questions social scientists are currently investigating. The class used open source software to create models based on these answers, and was composed of a lecture and lab section.
At first glance, a lecture for this course looked much like a lecture in any other class. With laptops out, the students focused on the projected screen, but the class was anything but just a lecture. Throughout each session, students examined maps and infographics and pulled up maps of their own. At the end of each day, students submitted a discussion on Canvas, which often required them to work in groups to respond to questions Kolak posed. For example, on the first day of lecture, students worked in groups to look up various maps to discuss what the map conveyed successfully, and what its limitations were.
“The exercise pointed to the importance of choice in creating a model, and how any model is almost always a simplified version of the data. How it is chosen to be represented will always impact what the viewer can interpret from it, “ said Kolak.
Labs were guided code introductions to a method that will have been introduced in a previous lecture and often based on current discussions occurring in the larger field of spatial analysis, using real data to model real social issues.
For one lab, students worked on creating maps from origin and destination data for Divvy bikes. The lab focused on applying methods from graph theory, a branch of mathematics that deals with networks. Graphs are symbolic representations of a network through a series of links and nodes. The lab compiled data from several million rides to see where in Chicago Divvy bikes were most commonly used. The maps gave way to conversation concerning accessibility. Labs like these provide a foundation for students to be able to apply the methods and code in their own endeavours.
Stephens said, “I liked how open ended the assignments were, because sometimes I would email you about an assignment, and you'd respond, ‘Why are you doing all of that?’ But it's given me a lot of room to explore on my own.”
Notably, a key component of labs was bugs. In nearly every lab, Kolak inserted a bug into a line of code that could disrupt the function of the code and challenged students to find the bug and resolve it.
“Much of spatial data science is learning how to make mistakes in a good way,” Kolak said. “You can't survive coding unless you're making mistakes and learning, and trying to figure out what's best for now. If you're a spatial data scientist, you're wrangling constantly different pieces of code, different pieces of data, with all the tools that are available.”
“I've really enjoyed the labs. Once you have an understanding of how this code works to generate this specific outcome, it's easy to go back and tweak a few things and make it work for what you want to do.”, said Joyce. “For example, when I was making density maps for my research, I was able to use code that we used in one of the labs, and put it at specific ranges. That really cut down a lot on time.”
Many of the students in the course also had a paid internship with a division faculty member for the summer quarter. Those included Kara Fischer, working with Forrest Stuart, Sociology, on the “Mapping Criminalization” project, Maximilian Site, working with Kate Cagney, Sociology, on the “Activity Space, Social Interaction and Health Trajectories in Later Life” project, Charlotte Scott and Jessica Robinson, working with Robert Vargas (Sociology) on the “Evolution of Urban Homicide Dynamics” project, Jordi Vasquez, working with Monica Nalepa and the “Long-term Consequences of Former Authoritarian Legacies” project, Samuel Joyce and Tiye Stephens, working with Luc Anselin and Kolak on the “Quantifying the Effects of the Structural Determinants of Health in Cities” project, and Jerry Shi, working with Luc Anselin and Julia Koschinsky on the “Next Generation Health and Human Service Infrastructure Platforms” project.
Vasquez, a rising third year majoring in political science and history, said, “In class, we were exposed to all these different tools, like GeoDa, and R, and QGIS, and stuff like that. For me, in my project, I've been basically building off of what was already there, and trying to complete the data set. Coming out of the class, I'm seriously considering doing the geography minor, as it's definitely something that has interested me more. I hadn't really thought about bringing that into political science, just trying to intuit it, but the class and the internship that I'm doing opened my mind up to that idea. It's definitely something that I think would be a good skill to have.”
A gift from Peter Hammack, AB’94 (Statistics), supported the internships. Hammack said, “I’m excited that this opportunity was available for students to gain some technical skills, which I think is important, and then actually go and work on something in the real world with those skills. That’s how I learned and understood even some basic things like linear regression, by actually looking at data sets and trying to understand things.”
As an introduction to GIS, Kolak hopes to introduce students to a set of skills they will be able to carry on throughout their respective fields, as well as pass on her contagious enthusiasm for the subject.
“This class, it could be the only GIS spatial analysis class that anyone takes, so the goal is to give them the resources possible so you can swim on your own in the ocean. That being said, I'm really excited. It seems like many students are going to keep on taking courses.”