第十届“中国人民大学国际统计论坛”（RUC IFS 2025）将于2025年7月4-6日在中国人民大学召开。大会邀请美国、澳大利亚等国家和地区的知名学者参会，将就“统计学发展史、数理统计、数据科学与人工智能、生物统计学前沿探究、政府统计、金融统计”等问题展开深入交流与讨论。

本次介绍特邀报告人Wang, Huixia Judy，报告主题为“Model-Free Approaches to Constructing Prediction Regions in Complex Data Settings”。

Wang, Huixia Judy

Title

Model-Free Approaches to Constructing Prediction Regions in Complex Data Settings

Abstract

Accurately predicting outcomes is essential across many fields, including biomedicine, where data-driven decisions support diagnostics and treatments. A prediction region defines a range where a future outcome is likely to fall, providing a measure of uncertainty. Traditional methods are typically model-based, relying on parametric assumptions about data distributions (e.g., Gaussian). However, real-world data often do not follow these idealized assumptions, leading to unreliable prediction regions. This talk introduces conformal prediction, a model-free, distribution-free method that provides reliable uncertainty estimates without restrictive assumptions. However, its accuracy relies on a condition called exchangeability, which is often unrealistic in practice. To address this, I will present conformal prediction techniques designed for non-exchangeable data, including: (1) Data with missing responses, common in medical studies; (2) Clustered or longitudinal data, such as repeated patient measurements; (3) Label shift scenarios, where training and application data distributions differ. To improve accuracy, we propose constructing prediction regions using highest posterior density estimation, which better handles asymmetric and multimodal distributions. This approach enhances predictions in personalized and heterogeneous scenarios, such as patient-specific risk assessments. I will share numerical results demonstrating the effectiveness of these methods.

Bio

Huixia Judy Wang received her Ph.D. in Statistics from the University of Illinois in 2006. She was a faculty member in the Department of Statistics at North Carolina State University from 2006 to 2014. She is currently Professor and Chair in the Department of Statistics at The George Washington University. She served as a Program Director of the National Science Foundation from 2018 to 2022. Her research interests include biostatistics and bioinformatics, quantile regression, semiparametric and nonparametric regression, high-dimensional inference, extreme value analysis, and spatial analysis, among others. She was a recipient of the NSF CAREER award, the Tweedie New Researcher Award from the Institute of Mathematical Statistics (IMS), and the Medallion Lectureship from IMS. She is an elected Fellow of the ASA and IMS, and an elected member of the ISI. She currently serves as a co-Editor for Statistica Sinica and as an Associate Editor for the Journal of the American Statistical Association.