知识卡片
预测随机性与参数不确定分开
用有限模拟人群的结果均值之差,还包含人群组成与结果噪声造成的蒙特卡洛误差。应提高积分精度或使用条件均值、共享抽样等方式,并区分这些随机性与真正的参数后验不确定性,才适合解释效应区间。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- · 查看原书依据
Finally, to estimate E(IE= 1) and E(IE= 0), we just need take the means of each posterior predictive sample dataset simulated from P(IE= 1) and P(IE= 1), respectively. This will yield 1,000 samples of posterior predictive values of the ATE. Variation between the samples reflects posterior uncertainty about the ATE.
参考来源
- 位置:《Causal AI》第11章《构建因果推断工作流》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-11.xhtml)
- 结论依据:摘录中关于预测随机性与参数不确定分开的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:Finally, to estimate E(IE= 1) and E(IE= 0), we just need take the means of each posterior predictive sample dataset simulated from P(IE= 1) and P(IE= 1), respectively. This will yield 1,000 samples of posterior predictive values of the ATE. Variation between the samples reflects posterior uncertainty about the ATE.