I develop the OpenTURNS library for uncertainty quantification.
We recently discussed the use of OpenTURNS to provide new features to statsmodels (at https://github.com/statsmodels/statsmodels/pull/7254
), and I suggested to use OT features. It seems that the OpenTURNS licence (LGPL) does not match statsmodels requirements: why is that?
If the licence was not an issue, what specific algorithms would be useful for statsmodels according to you? I guess that most probabilistic algorithms (copula, parameter estimation, non parametric estimation, etc...) would be useful, design of experiments (LHS, optimized LHS, quasi-Monte-Carlo, etc...) and also more advanced algorithms (high density regions, probabilistic transformations, stochastic processes, etc...). Is that correct?