必需做Causal Inference的根源在于:
The “fundamental problem of causal inference” (Holland, 1986) is that we cannot expose an observational unit to treatment 1 and, at the same time, to treatment 0
其实这个问题,在工业界,譬如CTR,CVR估计等问题中,都是普遍存在的,但是一定程度上被很多人忽略了。
工业界常用场景:
1、AB testing
2、营销优化
3、模型去偏(Causal Based Method),Selection Bias
4、评估去偏,MNAR问题。Unbiased Offline Evaluation
关于:Conditional Effect[1][2]
关于:Counterfactual [3]
Refer
1、Probability and Causality. Conditional and Average Total Effects
2、causal inference与model的一些理解:
total Conditional effect可不用线性模型,直接积分掉?
https://www.zhihu.com/question/266812683/answer/895210894
网友评论