pusion.model.configuration module

class pusion.model.configuration.Configuration(method, problem='GENERIC', assignment_type='GENERIC', coverage_type='GENERIC')

Bases: object

The Configuration forms the main parameter of the decision fusion framework. Based on this, the framework is able to check the compatibility of a respective decision fusion method to the given decision outputs. A configuration may be defined by the user or auto-detected by the framework.

Parameters

  • method – An explicit method provided by the framework which should be applied on the input data set. See pusion.Method for possible options.

  • problem – Input problem type. See pusion.util.constants.Problem for possible options.

  • assignment_type – The class assignment describes memberships to each individual class for a sample. A class assignment type is either crisp or continuous. Crisp assignments are equivalent to labels and continuous assignments represent probabilities for each class being true. See pusion.util.constants.AssignmentType for possible options.

  • coverage_type – The classification coverage states for each input classifier, which classes it is able to decide. A classifier ensemble may yield a redundant, complementary or complementary-redundant coverage. See pusion.util.constants.CoverageType for possible options.

get_tuple()
Returns

A tuple of method, problem, assignment type and coverage type.

get_pac()
Returns

A tuple of problem, assignment type and coverage type. This tuple is also referred to as PAC.