浙江大学学报（理学版）2025，Vol.52Issue(6)：706-718,13.DOI:10.3785/j.issn.1008-9497.2025.06.005
带有测量误差的零膨胀负二项模型的统计推断及其应用
Statistical inference of zero-inflated negative binomial model with measurement error and its applications
摘要
Abstract
In the process of data collection,measurement errors are common issues,neglecting these errors often results in bias in the analysis outcomes.This paper develops a zero-inflated negative binomial model that accounts for measurement error in the response variable.It discusses the identification challenges associated with the model,proposes a latent variable representation of the zero-inflated negative binomial model,and conducts statistical inference using an enhanced Bayesian method.Through repeated simulation experiments,the effectiveness of the proposed method is validated by comparing its performance with traditional methods that do not account for measurement error,as well as with the zero-inflated Poisson model.Finally,utilizing human breast cancer genome data,we empirically analyzed the relationship between cancer stage and the probability of signaling pathway activation in cancer patients,finding that the zero-inflated negative binomial model with measurement error yields the best performance.This paper contributes to the theoretical framework of counting models and measurement error theory,providing new methodological support for addressing data characterized by excessive discreteness and measurement error.关键词
测量误差/零膨胀负二项模型/计数数据/模型可识别性/贝叶斯框架Key words
measurement error/zero-inflated negative binomial model/counting data/model identifiability/Bayesian framework分类
数理科学引用本文复制引用
赵霞,徐泽轩,钟玉洁..带有测量误差的零膨胀负二项模型的统计推断及其应用[J].浙江大学学报（理学版）,2025,52(6):706-718,13.基金项目
国家自然科学基金面上项目(12271330) （12271330）
上海市哲学社会科学规划项目(2023ZJB001). （2023ZJB001）
