Zehang Richard Li

PhD student
Department of Statistics
University of Washington

Padelford Hall C14-G
Seattle, WA 98195
Phone: +1 (206) 902 7951
Email: lizehang (at) uw.edu

Github: @richardli
Google Scholar: Zehang Richard Li


News

08/2016 Presentation at JSM, Chicago
03/2016 Skype presentation at WHO VA Working Group meeting, Geneva
03/2016 openVA package released to CRAN

about     publications     vita     misc    

I am currently a forth year PhD student in the Department of Statistics at the University of Washington, working under the supervision of Dr. Tyler McCormick. I was a research intern at Microsoft Research during summer 2016.

I am broadly interested in statistical, mostly Bayesian, methods for large and noisy social science and demography data. In particular, my current research focuses mainly on characterizing dependency structures using Bayesian graphical models while accounting for different types of informative prior information.

Current Projects
Bayesian Latent Graphical Model for Mixed and Missing Data. Joint work with Tyler McCormick (UW) and Sam Clark (OSU).
Discovering Product Competitions from Weekly Sales Data. Joint work with Matt Goldman (MSR) and Matt Taddy (MSR & Chicago).
Spatial-temporal Smoothing of Child Mortality Rates in Africa from DHS Surveys. Joint work with Jon Wakefield (UW), Sam Clark (OSU), et.al.

Selected Projects
Probabilistic Cause-of-death Assignment using Verbal Autopsy A Bayesian framework to flexibly quantify the uncertainties in the noisy data from multiple sources in verbal autopsy analysis. Joint work with Tyler McCormick (UW), Sam Clark (OSU), Clara Calvert (LSHTM), and colleagues at HDSS sites.
openVA: A Comprehensive R package for VA Analysis openVA provides a wrapper class for a suite of VA packages: InSilicoVA, InterVA4, Tariff and upcoming NBC package. It implements all the major VA algorithms currently been used, and provides simple syntax to analyze, report, and visualize VA data. Joint work with Tyler McCormick (UW) and Sam Clark (OSU).
Sparse Motifs: Discovering Structure in Massive Graphs A parsimonious model that captures local structure in large-scale telephone networks through motif statistics. By utilizing a Bayesian factorization algorithm, we could characterize dependence structure in these network statistics and learn the adoption process of network goods. Joint work with Tyler McCormick (UW) and Joshua Blumenstock (Berkeley).