Question about data generating process in experiments

[nehargupta]

Hi,

For the experiment provided for the metalearners in this file https://github.com/py-why/EconML/blob/main/notebooks/Metalearners%20Examples.ipynb

in section 1.1, "We use the data generating process (DGP) from Kunzel et al."

Hmm....is that true? That's not the data generating process I think I'm seeing...is it possible you're referring to an old version of the paper or have adjusted the experiment somehow? Or, if I'm wrong, would be a huge help if you can refer me to the right section of the paper. Thanks!

[kbattocchi]

Hi, I believe what we meant is that the high-level structure of the DGP is taken from the paper, but you're right that the specific instantiations of e(x), mu_0(x), and mu_1(x) don't correspond exactly to any of the specific definitions from the appendix of that paper (though the treatment effect here is the same as that from their first simulation).