« first day (5144 days earlier)      last day (22 days later) » 

Guillaume Dehaene
Guillaume Dehaene
7:48 AM
@usεr11852 I'm sorry, even after scrolling a few minutes through their site, I struggle to find their elevator pitch for the purpose of the library. Looking at their list of examples, it seems to be very wide? Do you have a specific application(s) in mind you would use it for?
 
1 hour later…
usεr11852
usεr11852
8:56 AM
Probably the authors of it can do a better job but in my mind is more like: "Uncertainty modelling Python toolkit with a focus on reliability and sensitivity applications"? I personally used it for some (ill-fated) copulas and general distribution fitting, as well as validating some of my scipy.stats work.
You are right, it is very broad. It reminds me a bit of early (10y+ ago) sklearn, which was far patchier than it is now.
 
1 hour later…
Guillaume Dehaene
Guillaume Dehaene
10:16 AM
@usεr11852 Right, that makes sense. I remember seeing so many projects about probabilistic modelling (mostly very bayesian) floating around but I don't think there's any great solution out there, sadly.
 
3 hours later…
usεr11852
usεr11852
1:22 PM
I mean, it is not trying to be numpyro and anything like that; is more like scipy.stats/sklearnrather than numpyro/tfp.
 
2 hours later…
J-J-J
J-J-J
3:12 PM
@usεr11852 Looking at the contributors and the history of the project (openturns.github.io/openturns/latest/about.html#people), it looks like a common project by a couple of big European companies (Airbus, EDF, etc.), so probably something primarily tailored to their internal needs and requirements, but they decided to share it publicly anyway.
@usεr11852 it looks nice anyway, I'll certainly explore the copula functions they seem to offer

« first day (5144 days earlier)      last day (22 days later) »