Docs

std.causal

DAG operations, do-calculus rules, back-door identification, and numeric effect estimation.

(import std.causal)

A graph is a list of edges. Top-level std.causal operates on directed edges (u -> v); the ADMG modules add bidirected (u <-> v) edges.

DAG queries

Symbol	Description
`(dag:nodes g)`	List of node symbols.
`(dag:parents g v)`	Parents of `v`.
`(dag:children g v)`	Children of `v`.
`(dag:ancestors g v)`	Ancestors of `v` (excluding `v`).
`(dag:descendants g v)`	Descendants of `v` (excluding `v`).
`(dag:non-descendants g v)`	Nodes that are not descendants of `v`.
`(dag:has-path? g u v)`	True when there is a directed path.
`(dag:valid? g)`	Structural validity check.
`(dag:cyclic? g)`	True when the graph has a cycle.
`(dag:topo-sort g)`	Topological order.

Symbol	Description
`(dag:add-edge g e)`	Add an edge.
`(dag:remove-edge g e)`	Remove an edge.
`(dag:flip-edge g e)`	Reverse an edge.
`(dag:mutilate-do g vars)`	Remove all incoming edges to `vars` (do-mutilation).
`(dag:mutilate-see g vars)`	Remove all outgoing edges from `vars` (see-mutilation).

Symbol	Description
`(dag:d-connected? g x y z)`	True when X and Y are d-connected given Z.
`(dag:d-separated? g x y z)`	True when X and Y are d-separated given Z.

Symbol	Description
`(do-rule1-applies? g y x z w)`	Insertion/deletion of observations.
`(do-rule2-applies? g y x z w)`	Action/observation exchange.
`(do-rule3-applies? g y x z w)`	Insertion/deletion of actions.

Symbol	Description
`(dag:satisfies-backdoor? g x y z)`	Test the back-door criterion.
`(dag:adjustment-sets g x y)`	Enumerate valid adjustment sets.
`(dag:adjustment-sets-observed g x y observed)`	Restrict to observed variables.

Symbol	Description
`(do:identify g y x)`	Returns an estimand AST or a failure witness.
`(do:identify-details g y x)`	Same with diagnostic details.
`(do:identify-observed g y x observed)`	Identify using only observed variables.
`(do:identify-details-observed g y x observed)`	Same with diagnostic details.
`(do:adjustment-formula->string formula)`	Pretty-print an adjustment formula.

Symbol	Description
`(do:estimate-effect data y x z)`	Adjustment-based ATE estimate.
`(do:conditional-mean data y x z)`	Conditional mean E[Y \| X=x, Z=z].
`(do:marginal-prob data vars values)`	Empirical joint probability.

Module	Description
std.causal.adjustment	GAC, front-door, IV.
std.causal.identify	ID/IDC algorithms.
std.causal.estimate	Modern ATE estimators.
std.causal.learn	Structure learning.
std.causal.counterfactual	Counterfactual queries.
std.causal.mediation	Natural direct/indirect effects.
std.causal.transport	Transportability of effects.
std.causal.admg	ADMG operations.
std.causal.render	Graph rendering.

Source: stdlib/std/causal.eta