scoring
Estimation results in the list of results are scored by each scoring model in the supplied set scoring models. Any estimation results within the supplied list that were not successfully estimated or had no parameters estimated will not be scored. The returned list contains instances holding the scoring results for each successfully estimated distribution.
A result is also skipped when no distribution can be constructed from its parameters. That covers two cases, and the second used to end the whole call: an unknown random variable type, for which createDistribution returns null, and a KNOWN type whose estimated parameter values are not valid for it, for which the distribution's own constructor throws — a normal with zero variance, say, which an estimator will report as a success on constant or near-constant data. One unusable candidate out of many is a candidate to drop, not a reason to fail the scoring of every other.