How Do We See Data?

May 19, 2020

How do we see data?

Eric Field, Professor in the School of Data Science and in the School of Architecture at UVA, asks his students this exact question throughout the semester. He teaches a course called Data Visualization, an elective course that almost all MSDS students take before graduation.

“In the class, there's a lot of framing of how we look at data and information from the cognitive perspective, from a human perspective,” Field explained. “What is it that we as humans can do that AI and machine learning, the more computationally directed approaches, can't do?”

Field has been teaching a version of this course since 2003 in the School of Architecture. Year after year, this course has become more popular throughout the University. In fact,  students from over 21 departments at the University enrolled in Field’s course this (past) semester.

“This course in Architecture (SARC 5400), supported by Data Science, brings together 55 students from eight schools and 21 departments across the university, both graduate and undergraduate - Architecture, Urban Planning, Data Science, Engineering, Public Policy, Computer Science, Global Studies, Foreign Affairs, English, Economics, Statistics, Cognitive Science, and more,” Field said.

Throughout the semester, students study and create data visualizations. Field explained how students work together to examine each other’s work and make suggestions to better visualize the data.

“That's the big agenda of this whole thing is how do we see, literally see, into problems, how do we draw insights, and I use that term, draw, in a literal sense,” Field said.

The classwork all builds into the final project for the semester. Each year, Field gives his students a simple prompt. From there, students have the ability to look at a problem that interests them and can be as creative as they want in the visualization, using what they learned throughout the semester.

“Students have been studying visual approaches to problem solving - through data and information - and building dynamic visualizations of their own design,” Field said.

This year, the prompt for the final project is simply the word “viral.”

While prompted by the current COVID-19 pandemic, Field freed his students to explore a wide range of topics.

Some students have taken the prompt and looked into particular aspects of the current pandemic. Others have taken the word and explored fake news, including how it spreads. Other students are creating data visualizations about sentiment on Twitter and Facebook during this unprecedented time. 

“One student found a project at MIT where they turned the protein sequence of the virus itself into music[CM1] , and is looking to find ways of visualizing elements and trying to see if there are things that we can kind of poke on in terms of the way that the virus responds or attacks by using these abstractions and things like music and visualization, as a way to understand it,” Field said.

Daisy Mosqueda's data visualization on memes
Daisy Mosqueda's data visualization on memes

Another student, Daisy Mosqueda, is tracking memes, understanding their origin and references, and creating a data visualization that reflects this.

“I was looking to have fun and enjoy this project, so memes came to mind because I see them on social media every day, they can make me laugh uncontrollably, and I love sharing them with friends,” Mosqueda said. “I was also interested in the meme beyond the image macro- memes can also be dances, slang, politicians, and even conspiracy theories. What do memes tell us about what matters in our culture and society at different points of time? Another exciting part of my project was discovering how memes can suddenly become viral decades after they've been created.”

From policy response in COVID-19, to public transit ridership data, distancing measures, video gaming, and more, students have taken the prompt and run with it in all directions.

Field explained that he never wants to simply give students a data set to create a visual. He wants the students to dive into a topic that excites and challenges them, leaving room for discovery along the way.

“It's very open ended, to give all the students the flexibility to get something that not only are they interested in pursuing, because they're spending six weeks working on this project, but also gives them enough of an avenue and enough flexibility to get their own voice into this,” Field noted. “I'm not giving them the question to answer here. The students have to figure out what's important to them.”

Additionally, this prompt allowed students to explore something positive or all-together different from the COVID-19 pandemic.

“One of the reasons that I broadened the scope and just used the word, ‘viral,’ instead of trying to say, ‘Okay, everybody’s got to do something on coronavirus,’ is because for a lot of people, this is just too much right now. I wanted to give the students the out to be able to take a different angle and have some fun,” Field explained. 

To create a successful data visualization, Field noted that it is crucial to actually form a question before making the visual. What is it you actually want to know? From there, data visualizations can enable people from a variety of backgrounds to see a problem in a new way.

“By reframing a problem through a different formula abstraction, we can actually begin to see connections that even experts can't see,” Field said. “There's a really interesting lesson there even for the experts to sort of step back for a moment and come at the problem from a different perspective to be able to get out something that you couldn't previously understand.”

“Data is everywhere. It's used more and more every day in all aspects of society,” said Mallory Lewis, student in this class who is a rising third year student in the McIntire School of Commerce. “Not everyone can be a data scientist, but everyone needs to be able to understand the insights and conclusions they come up with. This class taught me that even the most fascinating data means nothing if you can't visualize it in a way the average person can understand. As someone who can barely code, I liked that this class wasn't about your ability to perform super complex data analysis, but your ability to communicate whatever insights you had found.