Hanchao Zhang

About Me

I work as a Research Data Scientist at Google, where our team uses statistical methods to analyze complex strategic questions, optimize decision-making, and solve key challenges across the company. Before joining Google, I completed my Ph.D. at at New York University, worked under the mentorship of Prof. Thaddeus Tarpey. My research involves developing statistical models that cater to high-dimensional complex data, such as functional and imaging data. This endeavor confronts two main challenges: first, defining a metric with self-consistent properties to cluster populations into subgroups based on rich and complex tensor data; second, accurately identifying subgroups that closely align with target labels of interest by developing methods like semi-supervised clustering algorithms. Prior to joining the Ph.D. program, I completed my master’s degree in Biostatistics and Data Science from Cornell University advised by Prof. Iván Díaz.

Outside of academia, I began my journey as a professional swimmer at the age of 5 and went on to achieve several regional and national championships. I’m also passionate about Chinese calligraphy, and my artwork has been exhibited in top galleries and museums including the National Art Museum of China (Beijing). In addition, I am also interested in diving, oil painting, and aeromodelling.

Research Interests

Causal Inference and Precision Medicine: causal subpopulation identification, mediation analysis, personalized recommendation based on treatment effect and generated effect modifiers
Machine Learning: fairness ML, semi-supervised learning
Manifold Learning and Metric Learning: positive semi-definite manifold learning and metric learning
Functional Data Analysis: functional regression, and clustering methods

Awards

[May. 2023] MacCracken Awards for outstanding research and activities

Publications and Projects

arXiv

K-Tensors: Clustering Positive Semi-Definite Matrices

Hanchao Zhang, Thaddeus Tarpey

arXiv (arXiv), Jun. 2023.

Website PDF GitHub Package BibTeX arXiv.org
arXiv

On the Inference of Confounders for Mediation Analysis

Hanchao Zhang, Thaddeus Tarpey

arXiv (arXiv), Jun. 2024.
AHI

Open Case Studies

Stephanie Hicks, Leah Jager, Margaret Taub, Carrie Wright, John Muschelli, Lyla Atta, Michael Breshock, Qier Meng, Alexandra Stephens, Hanchao Zhang, and etc.

the Johns Hopkins University, and Bloomberg American Health Initiative.

Website Project Page GitHub Contact

Talks

JSM

Common Principal Components Subspace for Positive Semi-Definie Matrices

Hanchao Zhang, Thaddeus Tarpey

the Joint Statistical Meeting (JSM), 2024, Portland, OR.

Poster Code Package Program Invited Session

TTHS

K-Tensors: Clustering Postive Semi-Definite Matrices

Hanchao Zhang, Thaddeus Tarpey

Thomas R. Ten Have Symposium (TTHS), Harvard University, 2023, Cambridge, MA.

Poster Code Package Program Poster Presentation

ENAR

Clustering Positive Semi-Definite Matrices: A Metric Learning Approach

Hanchao Zhang, Thaddeus Tarpey

the International Biometric Society Eastern North American Region (ENAR), 2023, Nashville, TN.

Slides Code Website Invited Talk

FDAWG

Functional Data Clustering and Generated Effect Modifier on Functional and Penalized Functional Regression

Hanchao Zhang, Thaddeus Tarpey

the Functional Data Analysis Working Group (FDAWG), Columbia University, 2022, New York, NY.

Slides Code Website Invited Talk

JSM

Optimal Transformations of High-Dimensional Functional Data for Clustering Methods

Hanchao Zhang, Thaddeus Tarpey

the Joint Statistical Meeting (JSM), 2022, Washington, D.C.

Slides Code Website Invited Talk

JSM

Optimal Linear Transformations of Functional Data for Clustering Methods

Hanchao Zhang, Thaddeus Tarpey

the Joint Statistical Meeting (JSM), 2021, Online.

Slides Code Website Invited Talk