ClassificationMetricResults

class ClassificationMetricResults(data)

Bases: MetricResults[ClassificationMetricKey]

Results from computing metrics.

Initialize the result wrapper.

Attributes Summary

metrics

Methods Summary

from_scores(metrics, scores_and_true_masks)	Return an instance of these metrics from a given set of true and scores.
key_from_string(s)	Parse the metric key from a (un-normalized) string.

Attributes Documentation

Parameters:

data (Mapping[MetricKeyType, float]) –

metrics: ClassVar[Mapping[str, Type[Metric]]] = {'accuracy': <class 'pykeen.metrics.classification.Accuracy'>, 'areaunderthereceiveroperatingcharacteristiccurve': <class 'pykeen.metrics.classification.AreaUnderTheReceiverOperatingCharacteristicCurve'>, 'averageprecisionscore': <class 'pykeen.metrics.classification.AveragePrecisionScore'>, 'balancedaccuracyscore': <class 'pykeen.metrics.classification.BalancedAccuracyScore'>, 'diagnosticoddsratio': <class 'pykeen.metrics.classification.DiagnosticOddsRatio'>, 'f1score': <class 'pykeen.metrics.classification.F1Score'>, 'falsediscoveryrate': <class 'pykeen.metrics.classification.FalseDiscoveryRate'>, 'falsenegativerate': <class 'pykeen.metrics.classification.FalseNegativeRate'>, 'falseomissionrate': <class 'pykeen.metrics.classification.FalseOmissionRate'>, 'falsepositiverate': <class 'pykeen.metrics.classification.FalsePositiveRate'>, 'fowlkesmallowsindex': <class 'pykeen.metrics.classification.FowlkesMallowsIndex'>, 'informedness': <class 'pykeen.metrics.classification.Informedness'>, 'matthewscorrelationcoefficient': <class 'pykeen.metrics.classification.MatthewsCorrelationCoefficient'>, 'negativelikelihoodratio': <class 'pykeen.metrics.classification.NegativeLikelihoodRatio'>, 'negativepredictivevalue': <class 'pykeen.metrics.classification.NegativePredictiveValue'>, 'numscores': <class 'pykeen.metrics.classification.NumScores'>, 'positivelikelihoodratio': <class 'pykeen.metrics.classification.PositiveLikelihoodRatio'>, 'positivepredictivevalue': <class 'pykeen.metrics.classification.PositivePredictiveValue'>, 'prevalencethreshold': <class 'pykeen.metrics.classification.PrevalenceThreshold'>, 'threatscore': <class 'pykeen.metrics.classification.ThreatScore'>, 'truenegativerate': <class 'pykeen.metrics.classification.TrueNegativeRate'>, 'truepositiverate': <class 'pykeen.metrics.classification.TruePositiveRate'>}

Methods Documentation

classmethod from_scores(metrics, scores_and_true_masks)

Return an instance of these metrics from a given set of true and scores.

Parameters:

• metrics (Iterable[ClassificationMetric]) –

• scores_and_true_masks (Iterable[ScorePack]) –

classmethod key_from_string(s)

Parse the metric key from a (un-normalized) string.

Parameters:

s (UnionType[str, None]) – the metric key, or None to get the default key.

Return type:

ClassificationMetricKey

Returns:

the fully resolved key as a named tuple