search

mwelz / GenericML

R implementation of Generic Machine Learning Inference (Chernozhukov, Demirer, Duflo and Fernández-Val, 2020).
GNU General Public License v3.0
60 stars 14 forks source link

ATE in CLAN plots #13

Closed mgalashin closed 2 years ago

mgalashin commented 2 years ago

Hi!

The plot method is adding the ATE mean and error bars for the CLAN plots. It doesn't make much sense, since the ATE is unrelated to the CLANs (even the scale is different). It looks like it is easily fixable either by removing the layer for the plot or changing it to the average of the CLAN covariate. Btw, thank you for the great package!

Best, Misha

mwelz commented 2 years ago

Dear @mgalashin,

Thank you for your interest in our package and sorry for the late reply (teaching obligations and Easter holidays have kept me busy). I agree that having the ATE in a plot of the CLANs can sometimes be confusing, so including the ATE in the plot should at least be optional. Hence, I have added the argument ATE to plot.GenericML() in PR #14 (you can download the development version via devtools::install_github("mwelz/GenericML")) . If you set this argument to FALSE, then neither the ATE nor its confidence bounds will be added to the plot.

I hope this helps!

All the best, Max

EDIT: changed the argument's name from plot_ATE to ATE, see #15.

aalfons commented 2 years ago

@mwelz, I don't think that plot_ATE is a good name for the argument. Plotting is what the function does, so including plot_ in the argument name is redundant and requires the user to type more than necessary. Why not just call the argument ATE? It's short and clean.

mwelz commented 2 years ago

@aalfons Fully agreed and fixed in #15.

mgalashin commented 2 years ago

Thank you, @mwelz and @aalfons!