Sensitivity and validation#

Use these objects after estimation. Diagnostics inspect fitted artifacts, sensitivity methods vary assumptions, and inference primitives construct derived uncertainty summaries.

Sensitivity#

`cleverly.assessment.SensitivityFacade`	Access sensitivity analyses supported by a fitted causal result.
`cleverly.sensitivity.PositivityReport`	Overlap diagnostics for a fitted TMLE.
`cleverly.sensitivity.SensitivityBounds`	Bias-adjusted bounds under an assumed confounder strength.
`cleverly.sensitivity.SensitivityElements`	The ingredients of the bias bound for one estimand.
`cleverly.sensitivity.BenchmarkResult`	Confounder strengths calibrated against covariates that were observed.
`cleverly.sensitivity.ConfounderStrengthGrid`	Declare treatment and outcome perturbation strengths.
`cleverly.sensitivity.ObservedConfounderCalibration`	Record one model-dependent observed-covariate calibration.
`cleverly.sensitivity.SimulatedConfoundingCell`	Store one strength pair's estimate or structured failure.
`cleverly.sensitivity.SimulatedConfoundingResult`	Store a qualitative simulated-confounding stress surface.
`cleverly.sensitivity.EValue`	E-values for a point estimate and the confidence limit nearest the null.
`cleverly.sensitivity.positivity_report`	Compute overlap diagnostics for a fitted TMLE.
`cleverly.sensitivity.truncation_curve`	Re-estimate across a grid of truncation bounds.
`cleverly.sensitivity.omitted_variable_bounds`	Bias-adjusted bounds and robustness values for one estimand.
`cleverly.sensitivity.robustness_value`	The confounding strength that would explain the effect away.
`cleverly.sensitivity.benchmark`	Calibrate `cf_y` and `cf_d` against observed covariates.
`cleverly.sensitivity.evalue.evalue`	E-value for a fitted result.
`cleverly.sensitivity.missingness_tilt`	Estimate under a range of departures from missingness-at-random.
`cleverly.sensitivity.tipping_gamma`	The tilt at which the conclusion tips.
`cleverly.sensitivity.simulated_confounding`	Refit a selected parameter across a simulated common-cause strength grid.

Validation#

`cleverly.assessment.DiagnosticsFacade`	Access diagnostics supported by a fitted causal result.
`cleverly.validation.ScoreCheck`	Whether the targeting step solved the score equations it was supposed to.
`cleverly.validation.ScoreCheckRow`	The score diagnostic for one estimand or one targeted family.
`cleverly.validation.NuisanceDiagnostics`	Fit quality for every nuisance model in a TMLE fit.
`cleverly.validation.NuisanceModelReport`	Fit quality for one nuisance model.
`cleverly.validation.RepeatSpreadRow`	Split sensitivity for one parameter in a repeated cross-fitted result.
`cleverly.validation.GaussianNoise`	Declare one Gaussian noise law for a generated outcome.
`cleverly.validation.GaussianIndependentOutcome`	Declare an outcome independent of treatment and adjustment variables.
`cleverly.validation.GaussianAdjustmentOutcome`	Declare `f(W) + effect * A + epsilon` with Gaussian noise.
`cleverly.validation.RelativeGaussianNoise`	Declare mean-zero numeric noise relative to a sampled variable.
`cleverly.validation.BootstrapMeasurementError`	Declare bootstrap measurement-error validation for adjustment variables.
`cleverly.validation.EmpiricalInclusionRule`	Declare a two-sided empirical inclusion rule for generated outcomes.
`cleverly.validation.EmpiricalRefitRecord`	One successful empirical refit.
`cleverly.validation.RefutationTest`	The outcome of one refutation test.
`cleverly.validation.RefutationResult`	All refutation tests run on a fit.
`cleverly.validation.CoverageStudy`	Repeatedly fit an estimator on fresh data and measure bias and coverage.
`cleverly.validation.StudyResult`	The full output of a `CoverageStudy`.
`cleverly.validation.ReplicationRecord`	One estimand from one successful repeated-sampling draw.
`cleverly.validation.ReplicationFailure`	A replication whose fit failed numerically, kept so the study can be diagnosed.
`cleverly.validation.ReplicationCheckpoint`	Storage that lets an interrupted `CoverageStudy` resume.
`cleverly.validation.DirectoryCheckpoint`	Keep each finished replication of a `CoverageStudy` as a file in a directory.
`cleverly.validation.score_check`	Check that targeting solved the score equation.
`cleverly.validation.nuisance_diagnostics`	Fit diagnostics for every nuisance model in the fit.
`cleverly.validation.refute`	Run refutation tests against a fitted model.

Inference primitives#

`cleverly.inference.BootstrapResult`	Bootstrap draws for every estimand, plus how many replicates failed.
`cleverly.inference.SimultaneousBands`	Joint confidence bands across several estimands.
`cleverly.inference.delta_method`	Influence curve of `function` applied to several estimands.
`cleverly.inference.influence_covariance`	Covariance matrix across several estimands' influence curves.
`cleverly.inference.influence_variance`	Variance of an estimator from its influence curve.
`cleverly.inference.multiplier_critical_value`	The `1 - alpha` quantile of the max-t statistic.
`cleverly.inference.simultaneous_bands`	Joint confidence bands over a set of estimates.
`cleverly.inference.run_bootstrap`	Run `n_replicates` full-refit bootstrap replicates.