pyrca.graphs.causal package

class pyrca.graphs.causal.PC(config)

Bases: CausalModel

The standard PC algorithm.

config_class: alias of PCConfig

class pyrca.graphs.causal.PCConfig(domain_knowledge_file=None, run_pdag2dag=True, max_num_points=5000000, alpha=0.01)

Bases: CausalModelConfig

The configuration class for the PC algorithm.

Parameters

domain_knowledge_file (Optional[str]) – The file path of the domain knowledge file.
run_pdag2dag (bool) – Whether to convert a partial DAG to a DAG.
max_num_points (int) – The maximum number of data points in causal discovery.
alpha (float) – The p-value threshold for independent test.

domain_knowledge_file: str = None

run_pdag2dag: bool = True

max_num_points: int = 5000000

alpha: float = 0.01

class pyrca.graphs.causal.GES(config)

Bases: CausalModel

The greedy equivalence search (GES) algorithm for causal discovery.

config_class: alias of GESConfig

class pyrca.graphs.causal.GESConfig(domain_knowledge_file=None, run_pdag2dag=True, max_num_points=5000000, max_degree=5, penalty_discount=100)

Bases: CausalModelConfig

The configuration class for the GES algorithm.

Parameters

domain_knowledge_file (Optional[str]) – The file path of the domain knowledge file.
run_pdag2dag (bool) – Whether to convert a partial DAG to a DAG.
max_num_points (int) – The maximum number of data points in causal discovery.
max_degree (int) – The allowed maximum number of parents when searching the graph.
penalty_discount (int) – The penalty discount (a regularization parameter).

domain_knowledge_file: str = None

run_pdag2dag: bool = True

max_num_points: int = 5000000

max_degree: int = 5

penalty_discount: int = 100

class pyrca.graphs.causal.FGES(config)

Bases: CausalModel

The fast greedy equivalence search (FGES) algorithm for causal discovery.

config_class: alias of FGESConfig

causal = None

static initialize()

static finish()

class pyrca.graphs.causal.FGESConfig(domain_knowledge_file=None, run_pdag2dag=True, max_num_points=5000000, max_degree=10, penalty_discount=80, score_id='sem_bic_score')

Bases: CausalModelConfig

The configuration class for the FGES algorithm.

Parameters

domain_knowledge_file (Optional[str]) – The file path of the domain knowledge file.
max_num_points (int) – The maximum number of data points in causal discovery.
max_degree (int) – The allowed maximum number of parents when searching the graph.
penalty_discount (int) – The penalty discount (a regularization parameter).
score_id (str) – The score function name, e.g., “sem_bic_score”.

domain_knowledge_file: str = None

max_num_points: int = 5000000

max_degree: int = 10

penalty_discount: int = 80

score_id: str = 'sem_bic_score'

class pyrca.graphs.causal.LiNGAM(config)

Bases: CausalModel

The non-gaussian linear causal models (LiNGAM): https://github.com/cdt15/lingam.

config_class: alias of LiNGAMConfig

class pyrca.graphs.causal.LiNGAMConfig(domain_knowledge_file=None, run_pdag2dag=True, max_num_points=5000000, lower_limit=0.1, n_sampling=-1, min_causal_effect=0.01)

Bases: CausalModelConfig

The configuration class for the LiNGAM algorithm.

Parameters

domain_knowledge_file (Optional[str]) – The file path of the domain knowledge file.
run_pdag2dag (bool) – Whether to convert a partial DAG to a DAG.
max_num_points (int) – The maximum number of data points in causal discovery.
lower_limit (float) – The lower limit of the causal effect scores for constructing causal graphs.
n_sampling (int) – The number of bootstrapping samples (for bootstrapping only). If n_sampling > 0, bootstrapping will be applied.
min_causal_effect (float) – The threshold for detecting causal direction (for bootstrapping only).

domain_knowledge_file: str = None

run_pdag2dag: bool = True

max_num_points: int = 5000000

lower_limit: float = 0.1

n_sampling: int = -1

min_causal_effect: float = 0.01