“Every weekend I go down to the river and try to learn the first trick that everyone does. When you see it, it doesn’t look impressive but it’s very scary. I need all the knee pads and elbow pads before I am comfortable doing it.”
Graduating stats concentrator Yash Nair is not afraid to tackle new challenges, whether it’s landing that first skateboarding trick or officially becoming a stats concentrator only a few months ago (it helped that he took many relevant classes along the way!). Yash capped off his undergraduate experience by receiving the Undergraduate Thesis Prize at our Statistics Commencement Celebration on May 26th for his thesis on “Randomization Tests for Adaptively Collected Data”, supervised by Professor Lucas Janson. In a recent Zoom conversation with Yash, we talked about his award-winning thesis, program highlights, and tips for incoming concentrators.
Stats: Tell us about your thesis project: How did you select your project and what questions were you asking?
Yash: “Sure, I can give you some background on my project and the motivation behind it. In clinical trials, to figure out if a treatment is effective, the gold standard is the randomized control trial, but it’s not always ethical to assign a treatment based on essentially flipping a coin. In these trials, you often learn new information over time, like patient A has certain characteristics that make them respond better to a treatment than patient B. Instead of assigning the treatment randomly (to either patient A or B), you would assign the treatment adaptively to patient A. But, when you assign treatments adaptively, it’s not clear whether the positive result is because of the treatment itself or because it’s a treatment based on previous data. When decisions are built on previous data, there are a lot of complicated dependencies that make it harder to know what influences the outcome of the patient. My project with Lucas was to figure out in a regime where you are assigning treatments adaptively, how can you still infer which treatments are working or if their impact is changing over time?”
“My thesis was more focused on methodology than any particular application but finding the answer to this question might have some useful applications. For example, a mobile health application that learns that a patient’s response to a treatment is changing over time could provide the patient with a useful intervention.”
Stats: What were some of the highlights as a concentrator in the program? Are there certain people, courses or parts of the program that really shaped your experience?
Yash: “My turning point was meeting Lucas; he’s the main reason why I’ve decided to continue studying statistics in grad school [Yash will start the PhD program in Statistics at Stanford next fall]. Originally, I had thought of myself as more of a CS person, but after meeting Lucas, I was open to taking more stats courses, and I ultimately became a stats concentrator. He was incredibly supportive as an advisor and was very patient with the mistakes that I inevitably made during the project.”
“Also, the quality of teaching in the Stats Department is unique; every time I take a stats course, I feel lucky because the professors and lecturers are absolutely incredible. Some courses that had the biggest impact on me and my ability to do research were the intro grad stats series: 210 [Probability I], 211 [Statistical Inference I], 212 [Probability II], and 213 [Statistical Inference II], taught by Professors Joe Blitzstein, Lucas Janson, Morgane Austern, and Pragya Sur. The faculty were very supportive - willing to meet with students and help them outside of class - and were just generally very excited about research. In some of these classes, we would talk about how these techniques can be used in disciplines ranging from genomics to clinical trials to the sciences at large – showing how the methods we learn are used to address real-world problems.”
“Another highlight was how collaborative the classes were. During 'pencil problems,' the prof. would stop the class and allow students to talk through a problem together as a group. This would A) help students with problem-solving, and B) help students to get to know each other. When I TF’d in 110, I found that students were very collaborative and were interested in content, not obsessed with grades. Stats seems to draw both undergrads and grads from many different departments, which also contributes to the community and brings in other perspectives.”
Stats: What advice would you give to future concentrators?
Yash: “The biggest thing for me was to be open to trying and learning new things (I only became a stats concentrator a few months ago). My path to stats was a long and winding one. I would say don’t be afraid to try new things; don’t be afraid to change paths, especially if you’re not sure what you’re really interested in.”
We think that Yash has certainly followed his own advice! The Statistics Department wishes him well on his adventures (both research and skateboarding-related) at Stanford.