HierarchicalCausalModel

class HierarchicalCausalModel

Bases: object

A class that wraps HCM functionality.

Initialize the HCM.

Methods Summary

add_edge(u, v, **kwargs)	Add an edge.
add_observed_node(node)	Add an observed node.
add_subunits(subunit_nodes)	Annotate the given nodes as the subunit graph.
add_unobserved_node(node)	Add an unobserved node.
copy_hcm()	Return a copy of the HCM.
delete_edge(u, v)	Delete an edge.
delete_node(node)	Delete a node.
edges()	Get all edges.
from_lists(*[, observed_subunits, ...])	Create a hierarchical causal model from the given node and edge lists.
get_direct_unit_descendants(subunit_node)	Return the set of direct unit descendants of the given subunit variable in the HCM.
get_observed()	Return the set of observed variables (both unit and subunit) in the HCM.
get_parents(node)	Return the set of parent/predecessor variables of the given variable in the HCM.
get_subunit_graph()	Return the subunit subgraph of the input HCM.
get_subunits()	Return the set of subunit variables in the HCM.
get_units()	Return the set of unit variables in the HCM.
get_unobserved()	Return the set of unobserved variables (both unit and subunit) in the HCM.
is_node_observed(node)	Check if the node is observed.
nodes()	Get all nodes.
predecessors(node)	Get predecessors.
successors(node)	Get successors.
to_admg(*[, return_hcgm])	Return a collapsed hierarchical causal model.
to_hcgm()	Convert an HCM to a hierarchical causal graphical model (HCGM) with promoted Q variables.
to_hscm()	Convert the input HCM to an explicit hierarchical structural causal model (HSCM).
to_pygraphviz()	Get a pygraphviz object.

Methods Documentation

add_edge(u: str | Variable | QVariable, v: str | Variable | QVariable, **kwargs: Any) → None

Add an edge.

add_observed_node(node: str | Variable | QVariable) → None

Add an observed node.

add_subunits(subunit_nodes: Iterable[str | Variable | QVariable]) → None

Annotate the given nodes as the subunit graph.

add_unobserved_node(node: str | Variable | QVariable) → None

Add an unobserved node.

copy_hcm() → HierarchicalCausalModel

Return a copy of the HCM.

delete_edge(u: str | Variable | QVariable, v: str | Variable | QVariable) → None

Delete an edge.

delete_node(node: str | Variable | QVariable) → None

Delete a node.

edges() → list[tuple[Variable, Variable]]

Get all edges.

classmethod from_lists(*, observed_subunits: Sequence[str | Variable | QVariable] | None = None, unobserved_subunits: Sequence[str | Variable | QVariable] | None = None, observed_units: Sequence[str | Variable | QVariable] | None = None, unobserved_units: Sequence[str | Variable | QVariable] | None = None, edges: Sequence[tuple[str | Variable | QVariable, str | Variable | QVariable]] | None = None) → HierarchicalCausalModel

Create a hierarchical causal model from the given node and edge lists.

Parameters:

• observed_subunits – a list of names for the observed subunit variables

• unobserved_subunits – a list of names for the unobserved subunit variables

• observed_units – a list of names for the observed unit variables

• unobserved_units – a list of names for the unobserved unit variables

• edges – a list of edges

Returns:

a hierarchical causal model with subunit variables in the SUBUNITS_KEY subgraph

get_direct_unit_descendants(subunit_node: str | Variable | QVariable) → set[Variable]

Return the set of direct unit descendants of the given subunit variable in the HCM.

get_observed() → set[Variable]

Return the set of observed variables (both unit and subunit) in the HCM.

get_parents(node: str | Variable | QVariable) → set[Variable]

Return the set of parent/predecessor variables of the given variable in the HCM.

get_subunit_graph() → DiGraph

Return the subunit subgraph of the input HCM.

get_subunits() → set[Variable]

Return the set of subunit variables in the HCM.

get_units() → set[Variable]

Return the set of unit variables in the HCM.

get_unobserved() → set[Variable]

Return the set of unobserved variables (both unit and subunit) in the HCM.

is_node_observed(node: str | Variable | QVariable) → bool

Check if the node is observed.

nodes() → list[Variable]

Get all nodes.

predecessors(node: str | Variable | QVariable) → list[Variable]

Get predecessors.

successors(node: str | Variable | QVariable) → list[Variable]

Get successors.

to_admg(*, return_hcgm: bool = False) → NxMixedGraph

Return a collapsed hierarchical causal model.

Parameters:

return_hcgm – if True, returns the intermediate hierarchical causal graphical models (HCGM) with subunits and promoted Q variables

Returns:

a mixed graph

Raises:

NotImplementedError – currently cannot handle unobserved subunit variables

to_hcgm() → HierarchicalCausalModel

Convert an HCM to a hierarchical causal graphical model (HCGM) with promoted Q variables.

to_hscm() → HierarchicalStructuralCausalModel

Convert the input HCM to an explicit hierarchical structural causal model (HSCM).

to_pygraphviz() → pygraphviz.AGraph

Get a pygraphviz object.