Global community detection for giant social networks is intractable. In recent work, we proposed a highly efficient algorithm that solves a local version of the community detection problem, and also established statistical guarantees for this procedure. (Publication 4)
Extending this idea, we describe a theory for sampling via connectivity for a general class of random graph models. (Publication 5)
I wrote some slides to introduce the ideas of Diaconis and Sturmfels (1998) along with some basic computational algebraic geometry. These are intended for a 30-minute talk.
For those interested in learning more about this new field (which is at the intersection of statistics and algebraic geometry) I recommend the book "Lectures on Algebraic Statistics" by Drton, Sturmfels, and Sullivant.
We characterized the extremal asymptotics of a particular exponential random graph model (ERGM), the edge-degeneracy ERGM. (Publication 2)
I am interested in improving the quality of publicly-released Census data for the Decennial Census as well as non-census surveys. In particular, I analyzed the statistical properties of a class of privacy-inducing algorithms. (Publication 1)
I gave a talk in Washington, D.C., at the NSF-Census Research Network Fall 2015 meeting; here are the slides.
This research was partially funded by Stephen E. Fienberg and William F. Eddy through NSF grant SES 1130706.
I was a statistics research intern at Lilly during the summer of 2017. There, I developed a Bayesian model for predicting which diseases a new drug may be effective in treating. In simultaneously modeling various diseases, this work extends existing meta-analytic methodology for predicting the outcome of untested comparisons between drugs for a single disease.
I was a statistics research intern at Bell Labs during the summer of 2016. There, I modeled routines in human location data.
My directory listing where you can download code, resources, etc.
My GitHub profile.
Check out my new project, Learn You a Data Analysis!
I wrote the code to generate the Department of Statistics' T-shirt: data-ify your logo or wordmark!
I am a member of the Networkshop research group at CMU.
Updated August 2017