I have a burning question: why don̵;t we use simulation and create a /tests for ? (i.e. where we know and control all the laws of and know exactly the confounders)

Would that be a better benchmark than e.g. Infant Health and Development Program as used in https://arxiv.org/pdf/1705.08821.pdf ? (where we do not really know the ground truth confounders)



Source link
thanks you RSS link
( https://www.reddit.com/r//comments/9e1j0v/d_why_not__a_benchmark_dataset_for_causal/)

LEAVE A REPLY

Please enter your comment!
Please enter your name here