In order to make the model code more generic, and enable easier implementation of BAD for other experimental domains, then I need a slight refactor of the models.
Model base class
[x] change predictive_y to be an abstract method, such that each concrete model class implements this method instead.
[x] remove calc_decision_variable as an abstract method. This is too specific and is not needed in general. We can keep it if we like in concrete model classes for DARC if that is convenient, but it is not a necessity.
All concrete models inheriting from Model
[x] implement the predictive_y method
Dealing with choice functions
[x] import choice functions into concrete models.py and call the desired choice function in the concrete model classes.
In order to make the model code more generic, and enable easier implementation of BAD for other experimental domains, then I need a slight refactor of the models.
Modelbase classpredictive_yto be an abstract method, such that each concrete model class implements this method instead.calc_decision_variableas an abstract method. This is too specific and is not needed in general. We can keep it if we like in concrete model classes for DARC if that is convenient, but it is not a necessity.All concrete models inheriting from
Modelpredictive_ymethodDealing with choice functions
Some method renaming, for convenience
get_simulated_responsetosimulate_y_θ_initialto_sample_from_priorCleaning up