Closed Mutaz94 closed 1 year ago
it would be convenient to copy the original dataset to the model directory (other than FDATA). Is there any similar functionality to do so?
I'm not Seth, but not that I'm aware of. Can you say a bit more about why you would find that convenient?
Depending on the answer, perhaps you'd find bbr's nm_data()
or nm_join()
useful.
Hey Kyle,
Thanks for your response. When dealing with simulation-estimation a lot of data generated for different scenarios (like bootstrap) and would be convenient to place each data with associated model. It was pointed out to me today that with project example, for bootstrap, is to create a subdir inside model directory but was curious if there any functionality to held the data there. I will check out these functions. Thanks a lot!!
Hi Mutaz. This may be what you're referencing, but our example script here puts the bootstrap data in a subdirectory under data/boot/
and keeps the models in model/pk/boot/
. Let us know if there are reasons this isn't ideal.
We've considered creating some bootstrap helpers in bbr
(probably based on what's in that script I linked ^) but haven't done it yet. Definitely on the roadmap though.
It is great way to handle and keep data consistent. I was curious if there a way to through a copy of original data into the model dir for example if a control stream called 1.mod
a copy of data will reside in ../1/org.dta
. Anyway, I'm now comfortable with the way you guys provided in expo project.
Hi @seth127:
When one handle simulation-estimation, it would be convenient to copy the original dataset to the model directory (other than FDATA). Is there any similar functionality to do so? e.g. if data
data.csv
is indata\
folder, when the model executed, a copy ofdata.csv
would go to the model dir. Thanks!!