search

ijyliu / ECMA-31330-Project

Econometrics and Machine Learning Group Project
2 stars 1 forks source link

Logging variables #46

Closed ijyliu closed 3 years ago

ijyliu commented 3 years ago

Especially working with shares.

From https://github.com/ijyliu/ECMA-31330-Project/issues/45#issuecomment-845292075:

if results are not that nice, I think we may explore the following option: take the log of percapita spending in health, and all variables in log. May be things may be clearer there (specially because with PCA we dont care about the scale).

ijyliu commented 3 years ago

@marionoro I think you said you are working on this? Should I implement it in the empirical section?

ijyliu commented 3 years ago

Oh wait, nevermind, i see #64

For the comparson between avarage, in the simulation we should take log of the true varaible and/or rescale (I prefer by -1), and then do the comparison. The same for IV!

ijyliu commented 3 years ago

We can close this if we did it for the simulation section.

I didn't do it for the empirical section but maybe we don't need to? I'm not sure on this point, can someone clarify? @nicomarto

paul-opheim commented 3 years ago

I do not think you need to do it for the empirical section. We just wanted to do that to look at what would happen if the covariate measurements were on different scales from each other.

ijyliu commented 3 years ago

Ok, this makes sense. All empirical items seem to be rougly on the same scale- dollars