STATISTICS · MACHINE LEARNING · SOCIOLOGY
STATISTICS · MACHINE LEARNING · SOCIOLOGY
I'm currently at Oxford studying mathematics and the history of science on a Rhodes Scholarship.
My research is motivated by two sets of applied questions:
I am pretty pragmatic in the methods I use to try to solve these problems. I've used neural nets and graph methods, custom Bayesian models and simple least squares. Often existing methods don't quite fit my applied problem, so I also work on methods to try to make machine learning and statistics more flexible, computationally tractable, and reliable for research purposes.
At the moment I am working with Kosuke Imai on non-parametric approaches for shape constrained heterogeneous treatment effect estimation, Victor Chernozhukov on quasi-Bayesian methods in econometrics, and the at the Center for Science of Science & Innovation on how science is funded.
Some of my research also comes out of my time spent working for Data for Progress, where I try to use computers to push American politics left.
Before starting at Oxford I was an undergrad at Stanford studying Sociology and Computer Science. My honors thesis in sociology focused on differences between the theory and practice and the diffusion of both in criminal procedure, advised by John Meyer. I also worked on regulatory genomics in Anshul Kundaje's group and social networks and socioeconomic segregation with David Grusky and Jure Leskovec.
You can find a more complete resume here, and you can reach me at njwfish [at] gmail.com.