My research projects are motivated primarily by questions in demography, developmental economics, sociology, machine learning, and conservation efforts.
Much of my work focuses on using sampled, incomplete, and noisy network data to learn features of unobserved networks and analyze how networks influence our behavior. I am interested in modeling standard sampled data types, such as respondent-driven sampling or aggregated relational data, to understand its uses and limitations. I also think about new data-driven ways of collecting network data to best address the researcherâ€™s questions. Much of this work is inspired by interdisciplinary collaboration. Another part of my work focuses on learning the geometric properties of data, particularly network data, and using these properties to improve downstream tasks in machine learning.
Some of my current projects focus on
See my Google Scholar page and NBER profile for an up-to-date list of publications.
Consistently estimating network statistics using aggregated relational data with Emily Breza, Arun Chandrasekhar, Tyler McCormick, and Mengjie Pan. Paper Summary. Code.
Spectral goodness of fit tests for complete and partial network data with Bolun Liu and Tyler McCormick. Paper Summary. Code.
Identifying the latent space geometry of network models through analysis of curvature with Arun Chandrasekhar and Tyler McCormick. Submitted. Paper Summary. Code.
Formal Definitions of Conservative PDFs with Clark Taylor. Submitted.
Verifying the predicted uncertainty of Bayesian estimators. Proc. SPIE 10645, Geospatial Informatics, Motion Imagery, and Network Analytics VIII, 106450E (27 April 2018)