Swupnil Sahai

Statistics Ph.D. Student

Columbia University

swupnil@stat.columbia.edu

Research

Featured Work

I am working with Sarah Cowan, Andrew Gelman, and Tian Zheng on a sociology project estimating the social prevalence of stigmatized populations (e.g. women who've had an abortion, men with prostate cancer, etc.) by asking survey questions of the form "How many X do you know?"


I am also working with Andrew Gelman and David Schiminovich on an astronomy project determining whether the relative amounts of infrared and ultraviolet radiation in various regions of the universe are indicative of radiation from the Big Bang. In the process, we have implemented a parallel, scalable algorithm for approximate posterior inference of hierarchical Bayesian models using Expectation Propagation.

In November 2012, a classmate and I successfully predicted the U.S. presidential election by utilizing a Bayesian framework. We predicted every state correctly, with the exception of Florida. Our paper explaining the methodology and results can be found here.


In January 2013, I was featured on the front page of the San Francisco Chronicle in an article describing the personalities of and challenges faced by the 27 triple majors at UC Berkeley. The article can be found here.


Since December 2014, I have been developing Swing™, the world's first social network and analytics expert for tennis players. In June 2016, I was awarded a scholarship to attend Apple's World Wide Developers Conference to discuss my app with Apple designers and engineers. In August 2016, Swing™ launched in public beta, which you can download here.