Reflecting on Spring 2020

May 11, 2020

Despite the difficult circumstances, student researchers in the UC Berkeley Data Science Discovery Research Program continue to push forward and work to come up with innovative solutions to the problems and issues they face, both with the situation of the current working environment and with the projects they work on. Members of two such teams, Data for Social Good (Elizabeth Leong, Jessica Rodriguez, Natasha Hellebrandt) and Student Success Analytics Platform (Sonika Arora, Viraj Kumar, Irene Ju, Andrei Caprau, Rohil Kanwar), reflected on the program, their projects, and their progress.

Data For Social Good intends to help the Data For Social Good Foundation, as student researcher Jessica elucidated, to “diversify the electorate by utilizing data and technology to help voters often marginalized by campaigns (low propensity voters, enthnic minorities, voters affected by language barriers, etc.) have access to resources that will help them become more aware of campaigns and politics that directly affect them. We primarily do this by working on our canvassing app, helping decide which features are best for clients, what data visualizations are easiest to interpret, and what data is most relevant to our purpose.”

The three researchers on this team work on condensing research into easy to read infographics, making stylistic changes to the existing website, adding core functionality, data integration and visualization, geocoding our data for effective use. Their unique experiences working with clients compared to classmates are certainly different; “It’s been a great learning experience to learn how to communicate the ideas in my head into a way that is digestible and understandable to the client.  It’s especially important to fully understand the “Why” so that each part of the application development process is driven by a motive and a goal.  I’ve realized that it’s a continuously iterative process that requires reanalyzing and rethinking about the problem in different situations and with different users,” researcher Elizabeth added.

Their work is key to serve the communities that may not otherwise be targeted; Natasha, the third researcher on the team, explained that we must “activate voters, because an active electorate could make a world of difference to certain communities. Trying to communicate this information to the public is a task I see now as being extremely important. Encouraging voters—especially low propensity ones— to show up to the polls is an ongoing fight but one we must carry out!”

Jacqueline Chang, Project/Policy Analyst in the Letters and Science Office of Undergraduate Advising and Discovery Project Partner, provided insightful context as to how the idea for the Student Success Analytics Platform came to be and how it could engender positive impact amongst both advisors and students. She explained, “In a college as big as L&S, we have over 20,000 students and about 27 advisors, so you can imagine how it’s nearly impossible to engage every student as often as we’d hope to. The vision is to use analytics as a tool to help proactively identify students who might benefit from academic advising at a certain point in their academic journeys, but who may not otherwise make an advising appointment until later down the road when they’ve run into more serious issues that might have been avoided.” She also noted that it’s important to be sensitive and careful, as predictive analytics models are simply tools and not ‘crystal balls,’ so “we don’t want to simply create self-fulfilling prophecies or reinforce negative biases.” 

The primary goal is to start with students who receive holds on their accounts from accumulating 75 units without having declared a major. As Jacqueline discussed, “the intent of this hold is to help make sure students are on a viable path to graduation, but we’ve seen in our office that oftentimes this hold causes unnecessary anxiety for students, as well as extra stress for both L&S Advisers and Undergraduate Major Advisers. So, we hope that analytics will help us reach out to and engage students who might be struggling to declare a major early enough that they can avoid ever getting a hold placed on their account.”

As a researcher in the project, Sonika works to interview L&S advisors, assess their needs, clean student data and conduct analysis; eventually, the team would like to develop the model to help predict when students might need advising help. She described that the most essential thing she learned from is teamwork, as “working with many different people and being open to ideas is very important when it comes to executing a project of this level. With a tremendous amount of data involved, it is imperative that we work as a team to manage the data and make sure we handle it safely.”

Jacqueline also emphasized that “it’s really inspiring to see this kind of collaboration happening; I’m a Berkeley alum so I could be kind of biased, but I know how smart and capable these students are. It’s wonderful we can tap into not only the technical expertise but the perspectives and experiences of student researchers. I think this also really aligns with the Chancellor’s initiatives to improve the undergraduate experience, both because it provides a Discovery opportunity for our Data Science students, and also because it goes towards improving academic advising for the largest college on our campus. It’s a win-win-win all around.”

Current support for the Student Success Analytics Platform is provided by the Khimji Family Foundation and alumni Frank Williams. If you wish to provide funding to expand this program, please consider making a donation to the College of Letters and Science.