This repo contains different script used to investigate on hypothesis concerning copulas applied to imprecise probabilities.
Here is a small sum up of the different scripts:
copulas.py
simple script containing the definitions for the different copulas considerednecessity_functions.py
contains different objects for defining univariate and bivariate necessity functions (with mass etc...)robust_set_sampling.py
contains different object for defining and computing the lower envelope of "robust" credal sets (both univariate and bivariate).ordering_focal_set_importance.py
script that generate both "mass" and "robust" credal sets with all orders on focal sets, for all marginals in a given range. Does the sampling of marginals incrementally (not optimal but goes to all extremes). Outputs a dataframe that indicates the orders on focal set where the robust set is a subset of the mass set.random_ordering_focal_set_importance.py
same asordering_focal_set_importance.py
but the sampling is random (with seeding)
And the different notebooks:
Copulas.ipynb
basic notebook to do some test and some verifications. Not interestingdataframe_analysis.ipynb
Plot figures to visualize marginal necessities and index processing to be sure that we consider the same marginals when comparing different copulas.
pandas>=1.2.0
for using pd.DataFrame.merge(df, how='cross')