dodiscover.ci.Oracle

class dodiscover.ci.Oracle(graph, included_nodes=None)

Oracle conditional independence testing.

Used for unit testing and checking intuition.

Parameters:

graphnx.DiGraph | Graph

The ground-truth causal graph.

Methods

test(df, x_vars, y_vars[, z_covariates])

Conditional independence test given an oracle.

test(df, x_vars, y_vars, z_covariates=None)

Conditional independence test given an oracle.

Checks conditional independence between ‘x_vars’ and ‘y_vars’ given ‘z_covariates’ of variables using the causal graph as an oracle. The oracle uses d-separation statements given the graph to query conditional independences. This is known as the Markov property for graphs [1][2].

Parameters:

dfpd.DataFrame of shape (n_samples, n_variables)

The data matrix. Passed in for API consistency, but not used.

x_varsnode

A node in the dataset.

y_varsnode

A node in the dataset.

z_covariatesset

The set of variables to check that separates x_vars and y_vars.

Returns:

statisticNone

A return argument for the statistic.

pvaluefloat

The pvalue. Return ‘1.0’ if not independent and ‘0.0’ if they are.

References

[1]

Judea Pearl. Causality: Models, Reasoning and Inference. Cambridge University Press, USA, 2nd edition, 2009. ISBN 052189560X.

[2]

Peter Spirtes, Clark Glymour, and Richard Scheines. Causation, Prediction, and Search. Volume 81. The MIT Press, 01 1993. ISBN 978-1-4612-7650-0. doi:10.1007/978-1-4612-2748-9.

Examples using dodiscover.ci.Oracle

Basic causal discovery with DoDiscover using the PC algorithm

Basic causal discovery with DoDiscover using the PC algorithm