Between every second-year graduate student and graduation stands one final challenge: Capstone.
During their last semester, students are enrolled in a required course where they are expected to take what they’ve learned in prior classes and put it to the test. This course, Capstone, guides students to complete an end-to-end and advanced data science project that encompasses ideation to data acquisition to model development to the build-out of a final minimum viable data and machine learning product or rigorous research paper. Top projects are showcased at the end of each semester.
Experiential learning opportunities, like Capstone, are a hallmark of the Master of Information and Data Science. In fact, almost every class includes hands-on immersive group projects that prepare students for real world data science work. These classes require collaboration and problem solving and feature tools, environments, and processes that are practiced in industry.
But, why is this class so significant? Does the work even matter in the end?
The answer, it seems, lies in the process.
“[Capstone]’s an amazing learning opportunity for both students and faculty,” said Joyce Shen, MIDS Continuing Lecturer and Capstone Course Lead, “It’s an interdisciplinary sandbox for innovative minds to come together and apply data science, machine learning, ethics, data privacy, project and product management, team work, leadership, strategic communication, creative thinking, and critical thinking to problems highly relevant to our society and in a way that is impactful and responsible.” All Capstone projects, in fact, explore and address contemporary issues and real-life topics, such as illegal fishing, sinkholes, aging, foster care, wildfires, misinformation, drug discovery, AI fairness, and more.
Designing for Data Science
Throughout the years, such projects in the data science program have left lasting impressions on both students and faculty. “The capstone program, the project presentations, the project defense, and the bonds fostered through completing that final project are still very fresh with me,” alumnus Taiwo Raphael Alabi (MIDS ’18) stated.
In the past couple of years alone, students such as Alabi have produced unforgettable proofs of concept that provided insights and state-of-art data science approaches in addressing pressing problems in our world. One such project was ElectBot AI, the Hal R. Varian award-winning project that provided crucial information about political candidates and encouraged potential voters to cast their ballots. The design directly addressed the ongoing issue of low voter turnout, seeking to help voters make informed decisions without information overload. Another such project, Hate Crime Index project, explored important factors influencing hate crime risk per county per year.
Capstone projects have also increasingly been focusing on challenges related to sustainability and climate change. In fact, the year 2022 saw more than twenty different data science projects centered around or adjacent to climate change, looking into illegal fishing, sinkholes, fires, and more. 2021 had a little over ten; 2018 had about five. In 2019, the Capstone project FairAir identified solutions to help disadvantaged communities have access to air quality data. In 2021, a project focused on enabling local regulators and methane-emitting facility owners to identify abnormal methane levels in their region. In 2022, BikeShare Wizard, an app that provides availability to Boston bike share users, provides a more eco-friendly way to travel. While the project does not directly mention sustainability as a key outcome, it is undeniable that the topic remains on the minds of many students in capstone classes and is appearing more frequently in presentations. A 2023 project was Home Energy Steward, which applied an advanced machine learning technique called reinforcement learning to help utility companies and households optimize energy costs through energy source and utilization management.
Nevertheless, data science capstone projects have always been varied in focus, reflecting the issues affecting our lives at a specific point in time. As claims of fake news gained mainstream momentum after 2016, there was a rise in projects focusing on the topic. After COVID-19 hit the United States, the same pattern emerged. To many, researching these timely issues allows for an opportunity to, as student Eric (Yue) Ling (MIDS ’23) puts it best, “develop skills and create something impactful with like-minded peers.” Eric and his team worked on a Capstone project that designed novel machine learning architecture to help generate more fair datasets.
At the end of the day, while Capstone projects mark the end of a student’s time here at the School of Information, they also represent the beginning of a new chapter in their data science, cybersecurity, or information science careers.
“Capstone has been an integral part of my MIDS experience, capping off a truly well-rounded program that ensures we as data scientists are able to put the technical work we do in the context of real-world problems,” said student Melissa McGee (MIDS ’23).
“[This class] is not just about a project,” Joyce Shen added, “It’s an interdisciplinary learning experience that allows students to explore all corners of data science and some really hard questions.”
Ultimately, the Capstone experience is one in which students must band together to tackle problems larger than themselves. By creating ethical and responsible tools using advanced data science and machine learning, students aim to make the work easier and transparent for users, researchers, policy makers, and activists to take action. In fact, a good number of students continue their projects beyond graduation, working to ensure that their projects continue to be useful and continue to make a difference in the world.
That mixture of interdisciplinarity, dedication, and innovative spirit is what makes the I School’s Capstone experience so unique and makes every Capstone project and the final showcase worth seeing year after year.