Closed sandcha closed 6 years ago
For now, and beyond the particular uses of OpenFisca by its contributors, it seems there are some main positioning paths (that are not exclusive):
It looks like is in the latter where the publication of aggregated data, test cases and samples could be more appropriate and useful, as it would be the case of the analysis of impact of reforms.
That being said, I'm not sure what would be a condition for generated data to be published and considered as open data. Representativeness ? Fit for purpose ?
Thanks for your thoughts @benjello and @pachevalier
@MattiSG dixit ;
La première fois que j'avais évoqué la possibilité de faire des datasets à partir des paramètres, ça datait de décembre 2014 (auprès d'Emmanuel). J'en avais parlé à nouveau l'an dernier à data.gouv.fr, il n'y avait pas de ressources dispo pour collaborer, mais un début d'intérêt.
To continue the discussion, some clues:
For the latter, question is whether we should create those samples ourselves or push for the INSEE to do it.
cc @benjello @pachevalier
We'll probably work on this this Friday in the context of Fridaylabs at Etalab
Variables are a choice of implementation. Parameters can be useful but should be organized to be usable. Somehow the legislation explorer with an export button would be better than a dump. Situations: I do not see the usefulness of sharing them.
Parameters can be useful but should be organized to be usable
We'll get there soon.
Situations: I do not see the usefulness of sharing them.
I see a collection of hypothetical data as a good starting kit, not sure anymore about synthetical data though.
So results of the investigation :
Note: build a library of hypothetical data on https://data.gouv.fr "opportunistically"
Relates to openfisca/openfisca-france#888
Définition of done : To-do liste actionable pour la publication, ou pas.