知识卡片
合成数据先检查方法链路
在已知DGP及真实目标值的模拟数据上,检查估计偏差、区间覆盖、样本量和假设违背的影响,可先排除实现问题。拟合现实数据得到的模拟器仍只是一个假设世界;在其上成功不能验证现实因果图。
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
Advice: Start with synthetic data One excellent practice is to build your workflow on synthetic data, rather than real data. Simulate a synthetic dataset that matches the size and correlation structure of your data, as well as your causal and statistical assumptions about your data. For example, you can write a causal generative model of your data, and use your data to train its parameters.
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:摘录中关于合成数据先检查方法链路的定义、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:Advice: Start with synthetic data One excellent practice is to build your workflow on synthetic data, rather than real data. Simulate a synthetic dataset that matches the size and correlation structure of your data, as well as your causal and statistical assumptions about your data. For example, you can write a causal generative model of your data, and use your data to train its parameters.