Skip to content

Interface: LogOddsScoreDecompositionDto

@kortexya/reasoninglayer


@kortexya/reasoninglayer / LogOddsScoreDecompositionDto

Defined in: src/types/inference.ts:392

The log-odds breakdown of a solution’s score.

totalScore is the sum the ranking used; every other field is one of its parts. version identifies the scoring model, so a stored decomposition stays interpretable after the model changes.

antecedents: LogOddsAntecedentContributionDto[]

Defined in: src/types/inference.ts:416

What each antecedent contributed.


coverageTerm: number

Defined in: src/types/inference.ts:402

The coverage term’s contribution.


coverageWeight: number

Defined in: src/types/inference.ts:404

The weight applied to coverage.


creditedCount: number

Defined in: src/types/inference.ts:406

How many antecedents were credited.


logLikelihoodSum: number

Defined in: src/types/inference.ts:400

The summed log-likelihood over credited antecedents.


logPrior: number

Defined in: src/types/inference.ts:398

The prior, in log space.


nullHitRate: number

Defined in: src/types/inference.ts:410

The null model’s hit rate.


nullLlrRate: number

Defined in: src/types/inference.ts:412

The null model’s log-likelihood-ratio rate.


nullTerm: number

Defined in: src/types/inference.ts:408

The null model’s term.


optional observationCount?: number

Defined in: src/types/inference.ts:414

How many observations backed the estimate, when recorded.


totalScore: number

Defined in: src/types/inference.ts:396

The score the ranking used.


version: number

Defined in: src/types/inference.ts:394

Which scoring model produced this.