causalexplain.estimators.lingam package#

Submodules#

Python implementation of the LiNGAM algorithms. The LiNGAM Project: https://sites.google.com/site/sshimizu06/lingam

Python implementation of the LiNGAM algorithms. The LiNGAM Project: https://sites.google.com/site/sshimizu06/lingam

class BootstrapMixin[source]#

Bases: object

Mixin class for all LiNGAM algorithms that implement the method of bootstrapping.

Methods

bootstrap(X, n_sampling)

Evaluate the statistical reliability of DAG based on the bootstrapping.
bootstrap(X, n_sampling)[source]#

Evaluate the statistical reliability of DAG based on the bootstrapping.

Parameters:

  • X (array-like, shape (n_samples, n_features)) – Training data, where n_samples is the number of samples and n_features is the number of features.

  • n_sampling (int) – Number of bootstrapping samples.

Returns:

result – Returns the result of bootstrapping.

Return type:

BootstrapResult

class BootstrapResult(adjacency_matrices, total_effects)[source]#

Bases: object

The result of bootstrapping.

Attributes:
adjacency_matrices_

The adjacency matrix list by bootstrapping.

total_effects_

The total effect list by bootstrapping.

Methods

get_causal_direction_counts([n_directions, ...])

Get causal direction count as a result of bootstrapping.

get_directed_acyclic_graph_counts([n_dags, ...])

Get DAGs count as a result of bootstrapping.

get_paths(from_index, to_index[, ...])

Get all paths from the start variable to the end variable and their bootstrap probabilities.

get_probabilities([min_causal_effect])

Get bootstrap probability.

get_total_causal_effects([min_causal_effect])

Get total effects list.
__init__(adjacency_matrices, total_effects)[source]#

Construct a BootstrapResult.

Parameters:

  • adjacency_matrices (array-like, shape (n_sampling)) – The adjacency matrix list by bootstrapping.

  • total_effects (array-like, shape (n_sampling)) – The total effects list by bootstrapping.

property adjacency_matrices_#

The adjacency matrix list by bootstrapping.

Returns:

adjacency_matrices_ – The adjacency matrix list, where n_sampling is the number of bootstrap sampling.

Return type:

array-like, shape (n_sampling)

property total_effects_#

The total effect list by bootstrapping.

Returns:

total_effects_ – The total effect list, where n_sampling is the number of bootstrap sampling.

Return type:

array-like, shape (n_sampling)

get_causal_direction_counts(n_directions=None, min_causal_effect=None, split_by_causal_effect_sign=False)[source]#

Get causal direction count as a result of bootstrapping.

Parameters:

  • n_directions (int, optional (default=None)) – If int, then The top n_directions items are included in the result

  • min_causal_effect (float, optional (default=None)) – Threshold for detecting causal direction. If float, then causal directions with absolute values of causal effects less than min_causal_effect are excluded.

  • split_by_causal_effect_sign (boolean, optional (default=False)) – If True, then causal directions are split depending on the sign of the causal effect.

Returns:

causal_direction_counts – List of causal directions sorted by count in descending order. The dictionary has the following format:

{'from': [n_directions], 'to': [n_directions], 'count': [n_directions]}

where n_directions is the number of causal directions.

Return type:

dict

get_directed_acyclic_graph_counts(n_dags=None, min_causal_effect=None, split_by_causal_effect_sign=False)[source]#

Get DAGs count as a result of bootstrapping.

Parameters:

  • n_dags (int, optional (default=None)) – If int, then The top n_dags items are included in the result

  • min_causal_effect (float, optional (default=None)) – Threshold for detecting causal direction. If float, then causal directions with absolute values of causal effects less than min_causal_effect are excluded.

  • split_by_causal_effect_sign (boolean, optional (default=False)) – If True, then causal directions are split depending on the sign of the causal effect.

Returns:

directed_acyclic_graph_counts – List of directed acyclic graphs sorted by count in descending order. The dictionary has the following format:

{'dag': [n_dags], 'count': [n_dags]}.

where n_dags is the number of directed acyclic graphs.

Return type:

dict

get_probabilities(min_causal_effect=None)[source]#

Get bootstrap probability.

Parameters:

min_causal_effect (float, optional (default=None)) – Threshold for detecting causal direction. If float, then causal directions with absolute values of causal effects less than min_causal_effect are excluded.

Returns:

probabilities – List of bootstrap probability matrix.

Return type:

array-like

get_total_causal_effects(min_causal_effect=None)[source]#

Get total effects list.

Parameters:

min_causal_effect (float, optional (default=None)) – Threshold for detecting causal direction. If float, then causal directions with absolute values of causal effects less than min_causal_effect are excluded.

Returns:

total_causal_effects – List of bootstrap total causal effect sorted by probability in descending order. The dictionary has the following format:

{'from': [n_directions], 'to': [n_directions], 'effect': [n_directions], 'probability': [n_directions]}

where n_directions is the number of causal directions.

Return type:

dict

get_paths(from_index, to_index, min_causal_effect=None)[source]#

Get all paths from the start variable to the end variable and their bootstrap probabilities.

Parameters:

  • from_index (int) – Index of the variable at the start of the path.

  • to_index (int) – Index of the variable at the end of the path.

  • min_causal_effect (float, optional (default=None)) – Threshold for detecting causal direction. Causal directions with absolute values of causal effects less than min_causal_effect are excluded.

Returns:

paths – List of path and bootstrap probability. The dictionary has the following format:

{'path': [n_paths], 'effect': [n_paths], 'probability': [n_paths]}

where n_paths is the number of paths.

Return type:

dict

Python implementation of the LiNGAM algorithms. The LiNGAM Project: https://sites.google.com/site/sshimizu06/lingam

class DirectLiNGAM(name, random_state=None, prior_knowledge=None, apply_prior_knowledge_softly=False, measure='pwling', th=0.0, inverse=False, absolute_values=False, verbose=False)[source]#

Bases: _BaseLiNGAM

Implementation of DirectLiNGAM Algorithm [1] [2]

References

[1]
  1. Shimizu, T. Inazumi, Y. Sogawa, A. Hyvärinen, Y. Kawahara, T. Washio,

      1. Hoyer and K. Bollen.

DirectLiNGAM: A direct method for learning a linear non-Gaussian structural

equation model.

Journal of Machine Learning Research, 12(Apr): 1225–1248, 2011.

[2]
  1. Hyvärinen and S. M. Smith. Pairwise likelihood ratios for estimation

    of non-Gaussian structural eauation models.

Journal of Machine Learning Research 14:111-152, 2013.

Attributes:
adjacency_matrix_

Estimated adjacency matrix.

causal_order_

Estimated causal ordering.

dag
feature_names
metrics

Methods

bootstrap(X, n_sampling)

Evaluate the statistical reliability of DAG based on the bootstrapping.

estimate_total_effect(X, from_index, to_index)

Estimate total effect using causal model.

fit(X)

Fit the model to X.

fit_predict(train, test[, ref_graph])

Fit the model to X and return the estimated DAG.

get_error_independence_p_values(X)

Calculate the p-value matrix of independence between error variables.
is_fitted_ = False#
metrics = None#
dag = None#
feature_names = None#
__init__(name, random_state=None, prior_knowledge=None, apply_prior_knowledge_softly=False, measure='pwling', th=0.0, inverse=False, absolute_values=False, verbose=False)[source]#

Construct a DirectLiNGAM model.

Parameters:

  • random_state (int, optional (default=None)) – random_state is the seed used by the random number generator.

  • prior_knowledge (array-like, shape (n_features, n_features),) –

    optional (default=None) Prior knowledge used for causal discovery, where n_features is the number of features.

    The elements of prior knowledge matrix are defined as follows [1]_:

    • 0 : \(x_i\) does not have a directed path to \(x_j\)

    • 1 : \(x_i\) has a directed path to \(x_j\)

    • -1No prior knowledge is available to know if either of the two

      cases above (0 or 1) is true.

  • apply_prior_knowledge_softly (boolean, optional (default=False)) – If True, apply prior knowledge softly.

  • measure ({'pwling', 'kernel'}, optional (default='pwling')) – Measure to evaluate independence: ‘pwling’ [2]_ or ‘kernel’ [1]_.

fit(X)[source]#

Fit the model to X.

Parameters:

X (array-like, shape (n_samples, n_features)) – Training data, where n_samples is the number of samples and n_features is the number of features.

Returns:

self – Returns the instance itself.

Return type:

object

fit_predict(train, test, ref_graph=None)[source]#

Fit the model to X and return the estimated DAG.

Parameters:

X (array-like, shape (n_samples, n_features)) – Training data, where n_samples is the number of samples and n_features is the number of features.

Returns:

self – Returns the instance itself.

Return type:

object

main(dataset_name, threshold=0.0, inverse=True, absolute_values=True, input_path='/Users/renero/phd/data/sachs/', output_path='/Users/renero/phd/output/RC4/sachs/compared/', save=False)[source]#

Python implementation of the LiNGAM algorithms. The LiNGAM Project: https://sites.google.com/site/sshimizu06/lingam

find_all_paths(dag, from_index, to_index, min_causal_effect=0.0)[source]#

Find all paths from point to point in DAG.

Parameters:

  • dag (array-like, shape (n_features, n_features)) – The adjacency matrix to fine all paths, where n_features is the number of features.

  • from_index (int) – Index of the variable at the start of the path.

  • to_index (int) – Index of the variable at the end of the path.

  • min_causal_effect (float, optional (default=0.0)) – Threshold for detecting causal direction. Causal directions with absolute values of causal effects less than min_causal_effect are excluded.

Returns:

  • paths (array-like, shape (n_paths)) – List of found path, where n_paths is the number of paths.

  • effects (array-like, shape (n_paths)) – List of causal effect, where n_paths is the number of paths.

predict_adaptive_lasso(X, predictors, target, gamma=1.0)[source]#

Predict with Adaptive Lasso.

Parameters:

  • X (array-like, shape (n_samples, n_features)) – Training data, where n_samples is the number of samples and n_features is the number of features.

  • predictors (array-like, shape (n_predictors)) – Indices of predictor variable.

  • target (int) – Index of target variable.

Returns:

coef – Coefficients of predictor variable.

Return type:

array-like, shape (n_features)

Module contents#