KSLCore/ksl.utilities.distributions.fitting.mixture.scoring/MixtureBICCriterion

MixtureBICCriterion

class MixtureBICCriterion(worstValue: Double = defaultWorstValue) : LogLikelihoodCriterion(source)

The Bayesian information criterion of a fitted mixture.

The number of free parameters is taken from the candidate, which charges one fewer parameter than there are components for the mixing weights, plus every estimated component parameter including any shift.

Constructors

MixtureBICCriterion

constructor(worstValue: Double = defaultWorstValue)

Functions

compute

protected open override fun compute(candidate: MixtureCandidate, data: DoubleArray, logLikelihood: Double): Double

Computes the criterion from a log-likelihood already known to be finite.

newInstance

fun newInstance(): MixtureBICCriterion

A fresh instance, for use where a criterion must not be shared.