Lastly, after scratching my head for a while over the DRY (Don't Repeat Yourself) development principle and Python object inheritance, I've opted to remove the _quantities_of_interest predictive method call dictionary from the PredictMixin abstract model primitive. It adds a small amount of code redundancy as this dict must now be defined in all model subclasses, but there will nonetheless be variations in the items this dict contains depending on the model. This is the most pythonic solution for this code smell and way more maintainable than alternatives involving complexities like super().predict().
Just incorporating a few best practices at the end of the Alpha development stage.
I've added a Deprecation warning reminding users the legacy Lifetimes
fittersmodule will be deprecated during Beta development and also moved thecdnow_summarypytest fixture toconftest.pyso that it may be shared among all testing scripts.Lastly, after scratching my head for a while over the DRY (Don't Repeat Yourself) development principle and Python object inheritance, I've opted to remove the
_quantities_of_interestpredictive method call dictionary from thePredictMixinabstract model primitive. It adds a small amount of code redundancy as this dict must now be defined in all model subclasses, but there will nonetheless be variations in the items this dict contains depending on the model. This is the most pythonic solution for this code smell and way more maintainable than alternatives involving complexities likesuper().predict().