Interface: PrCurve

Defined in: ml/metrics.ts:100

A precision–recall curve and its average precision.

Properties

ap

ap: number;

Defined in: ml/metrics.ts:108

Average precision — Σ (Rₙ − Rₙ₋₁)·Pₙ (NaN with no positives).

---

baseline

baseline: number;

Defined in: ml/metrics.ts:110

Positive base rate (a no-skill classifier's precision).

---

precision

precision: Float64Array;

Defined in: ml/metrics.ts:104

Precision (y), starting at 1.

---

recall

recall: Float64Array;

Defined in: ml/metrics.ts:102

Recall (x), starting at 0.

---

thresholds

thresholds: Float64Array;

Defined in: ml/metrics.ts:106

Score threshold at each vertex (+Inf for the leading point).