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

MixtureAICCriterion

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

The Akaike information criterion of a fitted mixture, 2 * p - 2 * logLikelihood.

Delegated to Statistic.akaikeInfoCriterion, as BIC is to Statistic.bayesianInfoCriterion. The library's guard that the log-likelihood be finite cannot fire from here: evaluate above rejects an unusable log-likelihood and assigns the bounded worst value before compute is reached.

Constructors

MixtureAICCriterion

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(): MixtureAICCriterion

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