API

Modeling

Build, fit, and query Bayesian path models.

modeling.model(): Parse a specification and compile a Bayesian path model.
modeling.PathModel: A compiled Bayesian path model.
adjustment.AdjustmentModel: Facade for a backdoor-adjusted single-equation outcome model.
interpret.datagrid(): Build a covariate grid as the cartesian product of supplied columns.

Time-forward and counterfactual simulation helpers.

modeling.simulate(): Simulate data from a pathmc model with known parameter values.

Rules evaluated inside a Gaussian lag-panel scan.

policy.TreatmentPolicy: Immutable rule that assigns one numeric variable during panel simulation.
policy.TimeSchedule: One finite action per fitted time, copied and frozen at construction.
policy.current(): Read this unit’s simulated value at the decision time.
policy.previous(): Read this unit’s simulated value from the preceding period.
policy.where(): Assign yes where condition is true and no otherwise.
policy_result.PolicyRun: Population summaries for one history-dependent policy.
policy_result.PolicySimulationInfo: Simulation context; policies may differ in rules and unused initial keys.

Whole-graph validation via a permutation-based test.

falsify.falsify_graph(): Falsify a whole DAG against data via a node-permutation test.

Learn candidate structure from data and turn a DAG into a model.

discovery.TBFPC: Target-first Bayes Factor PC (TBF-PC) causal-discovery algorithm.
cpdag.same_markov_equivalence_class(): Return True if two graphs share a Markov equivalence class.
dag.dag_to_spec(): Translate a causal DAG into a pathmc DSL specification string.
dag.BuildModelFromDAG: Build a probabilistic model directly from a causal DAG and a dataset.

Extend the model DSL with domain-specific transformations.

transforms.Transform: Base class for named transforms with estimable parameters.
transforms.ParamSpec: Specification for a single transform parameter.
transforms.register_transform(): Register a custom transform for use in the pathmc DSL.

Result containers returned by PathModel and AdjustmentModel methods.

simulate.DoResult: Container for propagated posterior draws under an intervention.
simulate.EstimandResult: Posterior draws for a causal estimand (ATE, CATE, ATT, ATU).
interpret.InterpretResult: Unit-preserving posterior draws for predictions, contrasts, or slopes.
effects.EffectResult: Posterior draws for a labeled or path-based effect.
identify.ImplicationTestResult: Results of testing implied conditional independences against data.
sensitivity.SensitivityResult: Result of unmeasured confounding sensitivity analysis.
falsify.FalsificationResult: Result of a permutation-based DAG falsification test.
refute.PlaceboRefutationResult: Result of a Bayesian placebo-treatment refutation.
discovery.TestResult: Conditional-independence test statistics recorded during fitting.