Skip to content

Interface: AteEstimateResponse

@kortexya/reasoninglayer


@kortexya/reasoninglayer / AteEstimateResponse

Defined in: src/types/causal.ts:825

The certified doubly-robust ATE (or ATT) estimate.

ate: number

Defined in: src/types/causal.ts:827

The one-step / AIPW point estimate.


certificate: DoublyRobustCertificateDto

Defined in: src/types/causal.ts:829

The doubly-robust certificate: which estimator properties the engine verified.


ciLower: number

Defined in: src/types/causal.ts:831

Lower bound of the 95% Wald confidence interval.


ciUpper: number

Defined in: src/types/causal.ts:833

Upper bound of the 95% Wald confidence interval.


optional estimateTermId?: string | null

Defined in: src/types/causal.ts:839

Psi-term id of the persisted estimate (certified requests only): the estimate lives in the tenant’s knowledge base, linked to its identification certificate by a term reference.


optional identificationCertificateId?: string | null

Defined in: src/types/causal.ts:843

The identification certificate this estimate was gated on, when certified.


optional identificationProvenance?: string

Defined in: src/types/causal.ts:849

How identification was established: certified (gated on a persisted identification certificate) or caller_asserted (legacy — the caller chose the covariates unchecked).


method: string

Defined in: src/types/causal.ts:851

The estimation method.


n: number

Defined in: src/types/causal.ts:853

Number of observations used.


optional overlap?: OverlapDiagnosticDto | null

Defined in: src/types/causal.ts:855

Positivity / overlap diagnostic over the fitted propensities.


optional selectedEif?: string | null

Defined in: src/types/causal.ts:862

The efficient-influence-function variant the declared model structure selected: full_aipw, known_propensity, known_outcome_regression, or randomized_covariate_adjusted. null for the ATT endpoints (a distinct estimand).


optional sensitivity?: SensitivityDto | null

Defined in: src/types/causal.ts:867

Sensitivity to unmeasured confounding (E-value); null when the outcomes are constant (the E-value is then undefined).


stdError: number

Defined in: src/types/causal.ts:869

Standard error from the EIF sandwich variance.