通过数据裂变进行经验贝叶斯估计
统计方法学
2024-10-17 v1
摘要
我们展示了数据裂变——一种从单一观测值创建合成复制品的方法——如何应用于经验贝叶斯估计。这扩展了最近针对具有多个复制品的经验贝叶斯方法至经典单复制品设置。关键洞见在于，数据裂变后，经验贝叶斯估计可被表述为一般回归问题。
引用
@article{arxiv.2410.12117,
title = {Empirical Bayes estimation via data fission},
author = {Nikolaos Ignatiadis and Dennis L. Sun},
journal= {arXiv preprint arXiv:2410.12117},
year = {2024}
}
备注
This note was prepared as a comment on "Data Fission: Splitting a Single Data Point," by James Leiner, Boyan Duan, Larry Wasserman, and Aaditya Ramdas, a discussion paper in the Journal of the American Statistical Association