期刊文献+

类别归因比例的Bayes估计与极大似然估计比较

Bayesian Graphic Model Estimation of Category Attributable Fraction
下载PDF
导出
摘要 目的 比较类别归因比例的Bayes图模型估计与极大似然估计方法。方法 应用Gibbs抽样和迭代蒙特卡罗方法得到参数后验分布 ,得到类别归因比例的模型法估计 ,应用婴儿低出生体重资料分析孕妇吸烟状况的类别归因比例。结果 类别归因比例的Bayes估计与极大似然估计一致性较好。 结论 Bayes方法避免了复杂的解析和高维积分运算 ,对于复杂模型估计 ,比极大似然估计有更实际的可行性。 Objective We will compare Bayesian graphic model estimation of Category Attributable Fraction with maximum likelihood estimation of it.Methods Based on Markov Chain Monte Carlo(MCMC)theory and full conditional distribution provided by multinomial logit regression model,Bayesian graphic model by means of Gibbs sampler is used to calculate parameter value and fitted probability value,thereby one can get the model-based estimating of CAF j.Results the analysis of data demonstrate that the approach gives good approximation to maximum likelihood estimation.Conclusion Avoiding sophisticated analysis and numerical high dimensiolnal integration procedure,Bayes.Estimation is more accessible than maximum likelihood estimation.
出处 《中国卫生统计》 CSCD 北大核心 2000年第6期322-324,共3页 Chinese Journal of Health Statistics
关键词 类别归因比例 Bayes图模型 马尔科夫链蒙特卡罗 Category attributable fraction Bayesian graphic model M
  • 引文网络
  • 相关文献

参考文献2

;
使用帮助 返回顶部