Dr. Peter W. MacDonald

About

I am an Assistant Professor in the Department of Statistics & Actuarial Science at the University of Waterloo. I work primarily on statistical analysis and methods for multiple and dynamic networks. I also have interests in post-selection inference, and formally private and fair statistical methods for network data.

I was formerly (2023-2024) a postdoctoral scholar in the Department of Mathematics & Statistics at McGill University, under the supervision of Dr. Eric D. Kolaczyk.

I completed my PhD (2018-2023) at the University of Michigan, co-advised by Dr. Elizaveta Levina and Dr. Ji Zhu. I am also an alumnus of their research group. Previously, I received my MMath (Statistics 2017-2018) and my BMath (Statistics and Pure Mathematics 2012-2017) from the University of Waterloo.

Teaching

Winter 2025

• STAT 450 (Estimation and Hypothesis Testing)
• STAT 321 (Regression and Forecasting)

Prospective Students

I am currently recruiting a Masters student starting in the Winter 2025 semester, to work on network changepoint detection

If you are a current or prospective UWaterloo student interested in working together, email me (pwmacdon AT uwaterloo DOT ca) with a CV, a brief bio, and your research interests; we can set up an online or in-person meeting to discuss further.

Research Interests and Selected Publications

Inference for noisy dynamic networks

(with Eric Kolaczyk, McGill)

• Autoregressive networks with dependent edges (with Jinyaun Chang, Qin Fang, Qiwei Yao; pre-print on arXiv).

Fair inference for network data

(with Eric Kolaczyk and Hui Shen, McGill)

Multilayer and dynamic networks

(with Elizaveta Levina and Ji Zhu, UMich)

• Latent space models for common and individual structure in multiplex networks (paper in Biometrika, multiness R package on CRAN).
• Latent process models for functional networks (submitted, pre-print on arXiv, fase R package on CRAN).
• Mesoscale inference for multilayer networks (pre-print on arXiv)

Post-selection inference

(with Dan Kessler, UNC and Snigdha Panigrahi, UMich)

• Bayesian post-selection inference with the Group LASSO (paper in the Journal of Machine Learning Research).

Multiple testing and false discovery rate (FDR) control

(with Kun Liang and Yingli Qin, UWaterloo)

• Dynamic-adaptive procedures which control the FDR (paper in the Electronic Journal of Statistics).
• Multiple hypothesis testing with group information (book chapter in Modern Statistical Methods for Health Research).

Interdisciplinary work

(supervised by Ji Zhu, UMich; Joel Rubenstein, VA Hospital and UMich Medicine)

• Predicting incident adenocarcinoma of the esophagus or gastric cardia using machine learning of electronic health records (papers in Gastroenterology).

Links

• Publications on Google Scholar
• Pre-prints on arXiv
• Code on GitHub

Education

• PhD, University of Michigan, 2023
• MMath, University of Waterloo, 2018
• BMath, University of Waterloo, 2017

Selected awards and honors

• NSERC Postdoctoral Fellowship, 2023-24
• Department of Statistics Outstanding Dissertation Award, University of Michigan 2023
• Rackham Pre-doctoral Fellowship, University of Michigan 2022-2023
• Statistical Learning and Data Science (SLDS) Student Paper Award, JSM 2021
• Rackham International Student Fellowship, University of Michigan, 2019

Contact information

Email: pwmacdon AT uwaterloo DOT ca