Can train/test split help standard econometrics?
Imagine we are presented with the following data-set:
medium.com