criterion
fun criterionProfile(criteria: List<MixtureCriterionIfc> = MixtureModeler.defaultReportedCriteria(catalogSize)): Map<String, Map<Int, MixtureCriterionValue>>(source)
Every reported criterion's value at every number of components that was fitted.
One criterion ranks the candidates, but no single criterion is trustworthy here — they disagree, and they disagree in ways a user should see rather than have resolved for them.
An approximation worth stating. The candidate scored at each count is the one the ranking criterion selected, not the one each criterion would have selected for itself. At a fixed count and partition the criteria share their likelihood term and differ only in penalty, so they rank identically unless parameter counts differ; the reported value is always the criterion's own value on the candidate named.
Parameters
criteria
the criteria to report