Speaker:Professor Kenichi Hayashi (Department of Mathematics, Keio University)
Topic:Divergence-based robust regression for binary response and its application to causal inference
Speaker:Professor Kenichi Hayashi
(Department of Mathematics, Keio University)
Date Time:Fri. Mar 14, 2025, 10:40-11:30
Place: 4F-427, Assembly Building I
Online Seminars- Google Meet
Abstract
Regression models for binary responses are frequently used in medical research. These models assume that the data are sampled from a specific population, but this assumption is often violated in practice due to the presence of outliers. This study introduces a general framework for regression models based on robust divergences, specifically Density Power Divergence and Gamma Divergence. We discuss specific loss functions derived under certain conditions. Furthermore, we apply this robust method to the problem of treatment effect estimation in randomized clinical trials.