Open sdvillal opened 9 years ago
Probably the best place for this is a plugin + a function to decorate all of the required objects (e.g. random variables).
Also the sklearn wrapper would probably benefit from its own plugin + an optional function to monkey-patch required objects.
At the moment we cannot really make probability distributions in scipy.stats to self-identify.
Vanilla introspection on these objects fails to separate id keys from object state, so manual code is in order.