Extracting group-time metric values and sample sizes

bernardodionisi / differences

difference-in-differences in Python

https://bernardodionisi.github.io/differences/latest/

GNU General Public License v3.0

93 stars 21 forks source link

Extracting group-time metric values and sample sizes #19

Open cadlaon opened 2 weeks ago

cadlaon commented 2 weeks ago

I wasn't sure if this was available in the documentation, but is this possible with the current package?

Extracting the mean variable value of a chosen group-time? For example, in a very standard canonical DiD (one control, one treatment group, and one pre, one post period), how do I extract the mean value of the target metric of the control group in the pre-period?
Extracting the sample size of the chosen group-time? The ATTgt.fit() method removes some units where they were treated since first time or never treated, so is it possible to know what sample size the model is working with in each group after performing these removals?

bernardodionisi commented 2 weeks ago

Hi, you should be able to extract the info you are looking for with a bit of work.

The info should be stored in the private ATTgt attribute _result_dict. May be mistaking on which attribute it is since I have not worked on this codebase in a while.

The number of observations should be directly retrievable from within _result_dict. The mean you may need to calculate with the information you will find in that attribute.

Maybe at some point I will expose the information for easier retirval.