KSLCore/ksl.utilities.distributions.fitting.mixture.refine/PartitionRefinerIfc

PartitionRefinerIfc

Improves an initial partition with respect to a fitting objective.

Implementations range from doing nothing to running a full classification expectation maximization. They are interchangeable so that the contribution of refinement can be measured rather than assumed, which is the point of having more than one.

Inheritors

BreakShiftRefiner

ClassificationEMRefiner

NoRefinement

Properties

name

abstract val name: String

A short name identifying the refiner, suitable as a factor level when results are recorded.

Functions

refine

abstract fun refine(sortedData: DoubleArray, initial: DataPartition, certificate: AdmissibilityCertificate, fitter: ComponentFitterIfc): RefinementResult

Refines the supplied partition.