High-Dimensional Uncertainty Quantification

When the number of variables is comparable to or larger than the sample size, conventional estimators and uncertainty measures can become unstable or misleading. My work develops confidence intervals, hypothesis tests, and multiple-testing procedures for high-dimensional linear, generalized linear, and additive models, including problems involving hidden confounding, endogeneity, mediation, genetic relatedness, and risk prediction.

Authorship notation Name supervised student # equal contribution * alphabetical ordering corresponding author