Robin Dunn

Robin Dunn

Senior Principal Statistical Consultant

Novartis Pharmaceuticals

Biography

I am a Senior Principal Statistical Consultant at Novartis Pharmaceuticals. I work in the Advanced Exploratory Analytics (AEA) group of the Advanced Methodology & Data Science (AMDS) team. I provide statistical modeling and methodological support at the trial and project level across therapeutic areas, with a particular focus on Immunology, Neuroscience, and Oncology. My recent research areas include causal inference in randomized controlled trials, covariate adjustment for efficient inference, and finite-sample valid uncertainty quantification methods (universal inference, conformal predictive inference).

I completed my PhD in Statistics at Carnegie Mellon University in July 2021. I was fortunate to work with Professors Larry Wasserman (co-advisor), Aaditya Ramdas (co-advisor), and Sivaraman Balakrishnan on universal likelihood ratio testing. The universal LRT provides tests with valid type I error control in any setting where we can write a likelihood ratio (or upper bound the null maximum likelihood). My research also extended conformal predictive inference to two-layer hierarchical settings. My thesis is available here.

Education

  • PhD in Statistics, 2021

    Carnegie Mellon University

  • MS in Statistics, 2017

    Carnegie Mellon University

  • BA in Mathematics, 2016

    Kenyon College

Publications

Universal Inference Meets Random Projections: A Scalable Test for Log-concavity

Shape-constrained density estimation poses a middle ground between fully nonparametric and fully parametric density estimation. Log-concavity is a common choice of shape constraint. Using universal inference, we develop the first test for log-concavity that is provably valid. Validity holds in finite samples.
April 2024 Accepted at Journal of Computational and Graphical Statistics
Preprint PDF Code

Distribution-Free Prediction Sets for Two-Layer Hierarchical Models

We extend conformal prediction methods to a two-layer hierarchical setting. Conformal prediction constructs valid prediction sets in finite samples, even if the predictive model is incorrect.
October 2023 Journal of the American Statistical Association
Preprint PDF Code

Gaussian Universal Likelihood Ratio Testing

The universal LRT provides finite-sample valid hypothesis tests and confidence sets in any setting for which we can compute the likelihood ratio. We present the first in-depth exploration of the size, power, and relationships between several universal LRT variants. We illustrate the benefits of the universal LRT in a test of a non-convex doughnut-shaped null hypothesis.
June 2023 Biometrika
Preprint PDF Code

Risk Scoring for Time to End-stage Knee Osteoarthritis: Data from the Osteoarthritis Initiative

We construct the first risk score for end-stage knee OA, using an end-stage definition that depends on symptomatic and radiographic criteria. We note particular implications for clinical trial patient selection.
August 2020 Osteoarthritis & Cartilage
Preprint PDF Dataset Reviewer editorial

A Flexible Pipeline for Prediction of Tropical Cyclone Paths

We construct prediction bands for tropical cyclone paths. We also develop a publicly available computational pipeline for this task.
June 2019 ICML Workshop on Climate Change: How can AI help?
PDF Code

Teaching

Instructor

Summer 2019: 36-315: Statistical Graphics and Visualization. Syllabus.

Developed course materials, presented lectures, led labs, and held office hours.
Topics: choosing and interpreting graphics, mastering ggplot, interactive graphics with Shiny.

Advising / Mentoring

Summer 2018: Summer Undergraduate Research Experience graduate advisor.

Mentored three undergraduate students to identify research directions on data science for justice. Provided guidance on R tools (ggplot, data.table, shapefiles) and statistical models (generalized linear models, random forests, spatiotemporal ETAS models). Students presented their final work at a departmental seminar and at the poster session of the American Statistical Association’s 2018 StatFest.

Teaching Assistant

Carnegie Mellon University

  • Fall 2019: 36-462/662: Data Mining.
  • Spring 2019: 36-402/608: Advanced Methods for Data Analysis.
  • Fall 2018: 36-315: Statistical Graphics and Visualization.

Kenyon College

  • Spring 2014 – Spring 2016 (4 semesters): STAT 206: Data Analysis.
  • Fall 2014: ECON 101: Principles of Microeconomics.
  • Fall 2013: ECON 102: Principles of Macroeconomics.

Honors and Awards

  • 2019 PhD Teaching Assistant of the Year, Carnegie Mellon University Department of Statistics & Data Science.
  • 2016 Kenyon College valedictorian. Highest Honors on Mathematics Honors curriculum. Distinction on Senior Exercise.
  • 2016 Gertrude M. Cox Scholarship, ASA Committee on Women in Statistics and Caucus for Women in Statistics.
  • 2016 Reginald B. Allen Prize, Kenyon College Department of Mathematics & Statistics academic honor.
  • 2016 NSF Graduate Research Fellowship, National Science Foundation.
  • 2015 Goldwater Scholar, Barry Goldwater Scholarship and Excellence in Education Foundation.
  • 2013 CAUSE Undergraduate Statistics Class Project Competition (third place), Consortium for the Advancement of Undergraduate Statistics Education.