Arize Phoenix TS
    Preparing search index...

    Options shared by precision, recall, and F-score computations.

    interface PrecisionRecallFScoreOptions {
        average?: AverageType;
        beta?: number;
        positiveLabel?: ClassificationLabel;
        zeroDivision?: number;
    }
    Index

    Properties

    average?: AverageType

    Aggregation strategy across classes. Ignored when positiveLabel is set (or auto-detected).

    "macro"
    
    beta?: number

    Weight of recall relative to precision in the F-score. Must be > 0.

    1
    
    positiveLabel?: ClassificationLabel

    When set, compute binary precision/recall/F exclusively for this label (one-vs-rest). If not set, average is at its default "macro", and the labels are the numeric set {0, 1}, the positive label defaults to 1. Otherwise, multi-class averaging is used. The auto-detection is skipped whenever a non-default average is configured, so an explicit average is never silently overridden by the shape of the data.

    zeroDivision?: number

    Value substituted for a metric when it is undefined (e.g. 0/0).

    0