Research progress on big-data-driven analysis strategies for imbalanced data of rare events
10.12173/j.issn.1005-0698.202411080
- VernacularTitle:大数据驱动的罕见事件非均衡数据分析方法研究进展
- Author:
Jiangjie ZHOU
1
;
Yutong WANG
;
Tian FENG
;
Xianglong MENG
;
Baosheng LIANG
;
Shengfeng WANG
Author Information
1. 北京大学公共卫生学院生物统计系(北京 100191)
- Publication Type:Journal Article
- Keywords:
Rare events;
Imbalanced data;
Data-driven;
Deep learning
- From:
Chinese Journal of Pharmacoepidemiology
2025;34(8):952-961
- CountryChina
- Language:Chinese
-
Abstract:
Rare events are widely prevalent in various disciplines,including rare adverse reactions to vaccines and drugs,clinical rare diseases,and low-probability clinical outcomes.The reason for research interest on such events is that their occurrence often brings incalculable and serious consequences.In the context of big data,numerous methods have emerged for rare event data analysis,including sampling based,category weighting,ensemble learning,and deep learning.This article systematically summarizes the research progress of current rare event data analysis methods,and introduces their basic principles and applicable scenarios.By analyzing the advantages and disadvantages of existing methods,the challenges of rare event research are sorted out and summarized,and potential research directions in related fields are explored to provide references for researchers.
