sorbetto.ranking package

class sorbetto.ranking.AbstractRanking(entities, performance_ordering, name=None)[source]

Bases: ABC

See Axiom 1 in Piérard et al. [17].

draw(fig: Figure | None = None, ax: Axes | None = None, value_axis_label: str = '') → tuple[Figure, Axes][source]
property entities: Iterable
getAllAvgRanks() → ndarray[source]
abstract getAllMaxRanks() → ndarray[source]
abstract getAllMinRanks() → ndarray[source]
abstract getAllStableRanks() → ndarray[source]
getAvgRank(entity) → float[source]
abstract getEntitiesAtRank(rank: int) → list[source]
abstract getMaxRank(entity) → int[source]
abstract getMinRank(entity) → int[source]
abstract getStableRank(entity) → int[source]
property name
property performance_ordering
abstract property values: ndarray
class sorbetto.ranking.RankingInducedByScore(entities, score, name=None)[source]

Bases: AbstractRanking

See Axiom 1 in Piérard et al. [17].

draw(fig: Figure | None = None, ax: Axes | None = None, value_axis_label: str = '') → tuple[Figure, Axes][source]
getAllMaxRanks() → ndarray[source]
getAllMinRanks() → ndarray[source]
getAllStableRanks() → ndarray[source]
getEntitiesAtRank(rank: int) → list[source]
Parameters:

rank – an integer between 1 and the number of entities.

Returns:

The list of all entities e such that min_rank(e) <= rank <= max_rank(e)

getMaxRank(entity) → int[source]
getMinRank(entity) → int[source]
getStableRank(entity) → int[source]
property values: ndarray
class sorbetto.ranking.RankingScore(importance: Importance, constraint=None, name: str | None = None, abbreviation: str | None = None, symbol: str | None = None)[source]

Bases: AbstractScore

drawInROC(fig, ax, priorPos: float, show_values_map: bool = True, show_iso_value_lines: bool = True, show_colorbar: bool = True, show_no_skills: bool = True, show_priors: bool = True, show_unbiased: bool = True) → None[source]

See https://en.wikipedia.org/wiki/Receiver_operating_characteristic

Parameters:

  • fig (_type_) – _description_

  • ax (_type_) – _description_

  • priorPos (float) – prior of the positive class \(\pi_+ \in (0,1)\)

  • show_values_map (bool, optional) – _description_. Defaults to True.

  • show_iso_value_lines (bool, optional) – _description_. Defaults to True.

  • show_colorbar (bool, optional) – _description_. Defaults to True.

  • show_no_skills (bool, optional) – _description_. Defaults to True.

  • show_priors (bool, optional) – _description_. Defaults to True.

  • show_unbiased (bool, optional) – _description_. Defaults to True.

static equivalent(p1: TwoClassClassificationPerformance, p2: TwoClassClassificationPerformance) → Conic[source]

Computes, on the Tile with the default parameterization, the locus of performance orderings for which the performances p1 and p2 are equivalent. This locus is a curve, a conic section.

Parameters:
Returns:

the conic section.

Return type:

Conic

static getAccuracy() → RankingScore[source]
static getBalancedAccuracy(priorPos: float) → RankingScore[source]
static getCriticalSuccessIndex() → RankingScore[source]
static getCzekanowskiBinaryIndex() → RankingScore[source]
static getDetectionRate(priorPos: float) → RankingScore[source]
static getDiceSorensenCoefficient() → RankingScore[source]
static getF(beta=1.0) → RankingScore[source]
static getIntersectionOverUnion() → RankingScore[source]

Intersection over Union (IoU). Synonyms: Jaccard index, Jaccard similarity coefficient, Tanimoto coefficient, similarity, critical success index (CSI), threat score.

static getInverseF(beta=1.0) → RankingScore[source]
static getInverseIntersectionOverUnion() → RankingScore[source]
static getInverseJaccard() → RankingScore[source]
static getInversePrecision() → RankingScore[source]
static getInverseRecall() → RankingScore[source]
static getJaccard() → RankingScore[source]
static getNegativePredictiveValue() → RankingScore[source]

Negative Predictive Value (NPV). Synonym: inverse precision

getPencilInROC(priorPos) → PencilOfLines[source]
static getPositivePredictiveValue() → RankingScore[source]

Positive Predictive Value (PPV). Synonym: precision

static getPrecision() → RankingScore[source]
static getProbabilityFalseNegativeComplenent(priorPos: float) → RankingScore[source]
static getProbabilityFalsePositiveComplenent(priorPos: float) → RankingScore[source]
static getProbabilityTrueNegative(priorPos: float) → RankingScore[source]
static getProbabilityTruePositive(priorPos: float) → RankingScore[source]
static getRecall() → RankingScore[source]
static getRejectionRate(priorPos: float) → RankingScore[source]
static getSelectivity() → RankingScore[source]
static getSensitivity() → RankingScore[source]
static getSimilarity() → RankingScore[source]
static getSkewInsensitiveVersionOfF(priorPos: float) → RankingScore[source]

The skew-insensitive version of \(\scoreFOne\). Defined in cite:t:Flach2003TheGeometry.

static getSpecificity() → RankingScore[source]
static getTanimotoCoefficient() → RankingScore[source]
static getTrueNegativeRate() → RankingScore[source]

True Negative Rate (TNR). Synonyms: specificity, selectivity, inverse recall.

static getTruePositiveRate() → RankingScore[source]

True Positive Rate (TPR). Synonyms: sensitivity, recall.

static getWeightedAccuracy(priorPos: float, weightPos: float) → RankingScore[source]
static getZijdenbosSimilarityIndex() → RankingScore[source]
property importance: Importance
isCanonical(tol=1e-08) → bool[source]

See Piérard et al. [16], Definition 1.

Submodules