ETIA.CausalLearning.algorithms.causalnex_algorithm package

Submodules

ETIA.CausalLearning.algorithms.causalnex_algorithm.NoTears module

class NoTearsAlgorithm(algorithm='notears', verbose=False)[source]

Bases: object

Implements the NOTEARS algorithm for learning causal structures from data.

prepare_data(Data, parameters=None)[source]

Prepares data for the NOTEARS algorithm.

check_parameters(parameters, data_info)[source]

Checks if the provided parameters are valid for the NOTEARS algorithm.

structure_model_to_matrix(StructureModel)[source]

Converts a causal structure model to an adjacency matrix.

run(data, parameters, prepare_data=True)[source]

Runs the NOTEARS algorithm on the provided data and parameters.

prepare_data(Data, parameters=None)[source]

Prepares data for the NOTEARS algorithm. This function can be adapted to include specific data preparation steps for NOTEARS.

Parameters:

  • Data (object) – The dataset to be used in the algorithm.

  • parameters (dict, optional) – Additional parameters for data preparation, if any. Default is None.

Returns:

Prepared dataset and additional preparation info, if any.

Return type:

tuple

check_parameters(parameters, data_info)[source]

Checks if the provided parameters are valid for the NOTEARS algorithm.

Parameters:

  • parameters (dict) – Parameters to be used in the algorithm.

  • data_info (dict) – Information about the dataset, such as data type and time info.

Returns:

True if parameters are valid, raises ValueError otherwise.

Return type:

bool

Raises:

ValueError – If an invalid parameter value is provided, such as a threshold outside the range [0, 1].

structure_model_to_matrix(StructureModel)[source]

Converts a StructureModel to an adjacency matrix representation.

Parameters:

StructureModel (causalnex.StructureModel) – A StructureModel object representing the learned structure.

Returns:

matrix – A numpy array representing the adjacency matrix of the DAG, where 2 represents an edge from source to target, and 3 represents a reverse edge.

Return type:

numpy.ndarray

run(data, parameters, prepare_data=True)[source]

Runs the NoTears algorithm to learn a causal structure from the data.

Parameters:

  • data (object) – The dataset to be used in the algorithm.

  • parameters (dict) – The parameters for the algorithm.

  • prepare_data (bool, optional) – If True, prepares the data before running the algorithm. Default is True.

Returns:

A tuple containing the learned structure as a MEC graph (pywhy graph) and library results.

Return type:

tuple

Raises:

ValueError – If parameters are invalid or other issues arise during the algorithm run.