# MultiLabelClassificationMetrics Multi-label classification metrics. ``` struct MultiLabelClassificationMetrics where Label : Hashable ``` ## Topics ### Creating the distribution [`init(some Sequence<(classification: ClassificationDistribution, labels: Set)>, strategy: MultiLabelClassificationMetrics.ThresholdSelectionStrategy) throws`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/init(_:strategy:)) Creates multi-label classification metrics for classifications and ground truth labels. [`init(some Sequence<(classification: ClassificationDistribution, labels: Set)>, strategy: MultiLabelClassificationMetrics.ThresholdSelectionStrategy, labels: Set) throws`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/init(_:strategy:labels:)) Creates multi-label classification metrics for classifications and ground truth labels. [`init(classifications: some Sequence>, groundTruth: some Sequence>, strategy: MultiLabelClassificationMetrics.ThresholdSelectionStrategy) throws`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/init(classifications:groundTruth:strategy:)) Creates multi-label classification metrics for classifications and ground truth labels. [`init(classifications: some Sequence>, groundTruth: some Sequence>, strategy: MultiLabelClassificationMetrics.ThresholdSelectionStrategy, labels: Set) throws`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/init(classifications:groundTruth:strategy:labels:)) Creates multi-label classification metrics for classifications and ground truth labels. [`init(confidenceThresholds: [Label : Float])`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/init(confidenceThresholds:)) Creates empty multi-label classification metrics. [`enum ThresholdSelectionStrategy`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/ThresholdSelectionStrategy) A strategy for selecting a confidence threshold. ### Getting the properties [`var confidenceThresholds: [Label : Float]`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/confidenceThresholds) A dictionary of label and confidence thresholds. [`var exampleCount: Int`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/exampleCount) The number of examples used to compute the metrics. [`var labels: Set`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/labels) The classifier labels. [`var meanAveragePrecision: Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/meanAveragePrecision) The mean average precision. ### Computing and scoring [`func count(of: Label) -> Int`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/count(of:)) Returns the number of times a label appeared in the ground truth collection. [`func f1Score(for: Label) -> Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/f1Score(for:)) Computes the F1 score from predicted and ground truth values. [`func falseNegativeCount(of: Label) -> Int`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/falseNegativeCount(of:)) Returns the number of times a true label was not predicted. [`func falsePositiveCount(of: Label) -> Int`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/falsePositiveCount(of:)) Returns the number of times the predicted label did not match the true label. [`func precisionScore(for: Label) -> Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/precisionScore(for:)) Computes the precision score for a class label. [`func recallScore(for: Label) -> Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/recallScore(for:)) Computes the recall score for a class label. [`func trueNegativeCount(of: Label) -> Int`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/trueNegativeCount(of:)) Returns the number of times a label was not in the predicted or ground truth collections. [`func truePositiveCount(of: Label) -> Int`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/truePositiveCount(of:)) Returns the number of times the predicted label matched the true label. ### Updating the metrics [`func add(some Sequence<(classification: ClassificationDistribution, labels: Set)>)`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/add(_:)) Updates the metrics with more pairs of classifications and ground truth labels. [`func add(classifications: some Sequence>, groundTruth: some Sequence>)`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/add(classifications:groundTruth:)) Updates the metrics with more classifications and ground truth labels. ### Computing the precision [`static func meanAveragePrecisionScore(some Sequence<(classification: ClassificationDistribution, labels: Set)>) -> Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/meanAveragePrecisionScore(_:)) Computes the mean average precision. [`static func meanAveragePrecisionScore(some Sequence<(classification: ClassificationDistribution, labels: Set)>, labels: Set) -> Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/meanAveragePrecisionScore(_:labels:)) Computes the mean average precision. [`static func meanAveragePrecisionScore(classifications: some Sequence>, groundTruth: some Sequence>) -> Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/meanAveragePrecisionScore(classifications:groundTruth:)) Computes the mean average precision. [`static func meanAveragePrecisionScore(classifications: some Sequence>, groundTruth: some Sequence>, labels: Set) -> Float`](/documentation/CreateMLComponents/MultiLabelClassificationMetrics/meanAveragePrecisionScore(classifications:groundTruth:labels:)) Computes the mean average precision. ## Relationships ### Conforms To [`SendableMetatype`](/documentation/Swift/SendableMetatype) [`Sendable`](/documentation/Swift/Sendable) --- Copyright © 2026 Apple Inc. All rights reserved. | [Terms of Use](https://www.apple.com/legal/internet-services/terms/site.html) | [Privacy Policy](https://www.apple.com/privacy/privacy-policy)