摘要
在两部分模型框架内讨论了不同缺失机制对协变量选择的影响;对连续协变量的缺失机制考虑了一般的情形,并对不同缺失情形下的模型建立选择程序.马尔可夫链蒙特卡洛抽样方法被用于后验抽样.统计推断建立在后验样本的经验分布基础上.随机模拟和CHFS数据分析展示了方法的有效性和实用性.
Under the two-part model framework,a Bayesian procedure is developed to explore the effects of different missing mechanisms on variable selection.We consider a general setting for missing mechanism of the continuous covariates and develop a model selection procedure to evaluate model fit under different missing scenarios.Markov Chains Monte Carlo sampling method is adopted to draw observations from the posterior.Statistical inference are carried out based on the empirical distribution of the simulated observations.Empirical results including simulation studies and CHFS data analysis are presented to illustrate the merits of the proposed methodology.
作者
康晴
夏业茂
KANG Qing;XIA Yemao(College of Science,Nanjing Forestry University,Nanjing 210037,China)
出处
《应用数学》
北大核心
2025年第3期872-885,共14页
Mathematica Applicata
基金
国家自然科学基金(114761)。
