causalexplain package

Subpackages

• causalexplain.common package
  • Submodules
    • BaseExperiment
      • BaseExperiment.__init__()
      • BaseExperiment.prepare_experiment_input()
      • BaseExperiment.experiment_exists()
      • BaseExperiment.create_estimator()
    • Experiment
      • Experiment.load()
      • Experiment.fit()
      • Experiment.save()
      • Experiment.estimator_name
      • Experiment.__init__()
      • Experiment.fit()
      • Experiment.predict()
      • Experiment.fit_predict()
      • Experiment.load()
      • Experiment.save()
    • setup_plot()
    • add_grid()
    • subplots()
    • format_graph()
    • draw_graph_subplot()
    • cleanup_graph()
    • set_colormap()
    • dag2dot()
    • values_distribution()
    • correlation_matrix()
    • hierarchies()
    • dag()
    • dags()
    • shap_values()
    • shap_discrepancies()
    • deprecated_dags()
    • score_by_method()
    • scores_by_method()
    • score_by_subtype()
    • combined_metrics()
    • latex_table_by_datatype()
    • latex_table_by_method()
    • save_experiment()
    • load_experiment()
    • valid_output_name()
    • graph_from_dot_file()
    • graph_from_dictionary()
    • graph_from_adjacency()
    • graph_from_adjacency_file()
    • graph_to_adjacency()
    • graph_to_adjacency_file()
    • graph_to_dot_file()
    • select_device()
    • graph_intersection()
    • graph_union()
    • digraph_from_connected_features()
    • correct_edge_from_prior()
    • valid_candidates_from_prior()
    • break_cycles_using_prior()
    • potential_misoriented_edges()
    • break_cycles_if_present()
    • stringfy_object()
    • get_feature_names()
    • get_feature_types()
    • cast_categoricals_to_int()
    • find_crossing_point()
    • format_time()
    • combine_dags()
    • list_files()
    • read_json_file()
  • Module contents
• causalexplain.estimators package
  • Subpackages
    • causalexplain.estimators.cam package
      • Submodules
      • Module contents
    • causalexplain.estimators.fci package
      • Submodules
      • Module contents
    • causalexplain.estimators.ges package
      • Submodules
      • Module contents
    • causalexplain.estimators.lingam package
      • Submodules
      • Module contents
    • causalexplain.estimators.notears package
      • Submodules
      • Module contents
    • causalexplain.estimators.pc package
      • Submodules
      • Module contents
    • causalexplain.estimators.rex package
      • Submodules
      • Module contents
  • Module contents
• causalexplain.explainability package
  • Submodules
    • Hierarchies
      • Hierarchies.correlations
      • Hierarchies.__init__()
      • Hierarchies.linkage_mat
      • Hierarchies.fit()
      • Hierarchies.compute_correlation_matrix()
      • Hierarchies.compute_correlated_features()
      • Hierarchies.expand_clusters_perm_importance()
      • Hierarchies.hierarchical_dissimilarities()
    • connect_isolated_nodes()
    • connect_hierarchies()
    • plot_dendogram_correlations()
    • Parameters:
    • PermutationImportance
      • PermutationImportance.device
      • PermutationImportance.__init__()
      • PermutationImportance.fit()
      • PermutationImportance.predict()
      • PermutationImportance.fit_predict()
      • PermutationImportance.plot()
      • PermutationImportance.set_predict_request()
    • RegQuality
      • RegQuality.__init__()
      • RegQuality.predict()
      • RegQuality.set_predict_request()
    • ShapDiscrepancy
      • ShapDiscrepancy.target
      • ShapDiscrepancy.parent
      • ShapDiscrepancy.shap_heteroskedasticity
      • ShapDiscrepancy.parent_heteroskedasticity
      • ShapDiscrepancy.shap_p_value
      • ShapDiscrepancy.parent_p_value
      • ShapDiscrepancy.shap_model
      • ShapDiscrepancy.parent_model
      • ShapDiscrepancy.shap_discrepancy
      • ShapDiscrepancy.shap_correlation
      • ShapDiscrepancy.shap_gof
      • ShapDiscrepancy.ks_pvalue
      • ShapDiscrepancy.ks_result
      • ShapDiscrepancy.__init__()
    • ShapEstimator
      • ShapEstimator.device
      • ShapEstimator.shap_discrepancies
      • ShapEstimator.__init__()
      • ShapEstimator.explainer
      • ShapEstimator.models
      • ShapEstimator.correlation_th
      • ShapEstimator.mean_shap_percentile
      • ShapEstimator.iters
      • ShapEstimator.reciprocity
      • ShapEstimator.min_impact
      • ShapEstimator.exhaustive
      • ShapEstimator.parallel_jobs
      • ShapEstimator.on_gpu
      • ShapEstimator.verbose
      • ShapEstimator.prog_bar
      • ShapEstimator.silent
      • ShapEstimator.fit()
      • ShapEstimator.predict()
      • ShapEstimator.adjust()
      • ShapEstimator.set_predict_request()
      • ShapEstimator.compute_error_contribution()
    • custom_main()
    • sachs_main()
  • Module contents
• causalexplain.generators package
  • Submodules
    • AcyclicGraphGenerator
      • AcyclicGraphGenerator.__init__()
      • AcyclicGraphGenerator.init_variables()
      • AcyclicGraphGenerator.generate()
      • AcyclicGraphGenerator.to_csv()
    • gmm_cause()
    • LinearMechanism
      • LinearMechanism.__init__()
      • LinearMechanism.__call__()
    • Polynomial_Mechanism
      • Polynomial_Mechanism.__init__()
      • Polynomial_Mechanism.mechanism()
      • Polynomial_Mechanism.__call__()
    • SigmoidAM_Mechanism
      • SigmoidAM_Mechanism.__init__()
      • SigmoidAM_Mechanism.mechanism()
      • SigmoidAM_Mechanism.__call__()
    • SigmoidMix_Mechanism
      • SigmoidMix_Mechanism.__init__()
      • SigmoidMix_Mechanism.mechanism()
      • SigmoidMix_Mechanism.__call__()
    • computeGaussKernel()
    • GaussianProcessAdd_Mechanism
      • GaussianProcessAdd_Mechanism.__init__()
      • GaussianProcessAdd_Mechanism.mechanism()
      • GaussianProcessAdd_Mechanism.__call__()
    • GaussianProcessMix_Mechanism
      • GaussianProcessMix_Mechanism.__init__()
      • GaussianProcessMix_Mechanism.mechanism()
      • GaussianProcessMix_Mechanism.__call__()
    • gaussian_cause()
    • noise()
  • Module contents
• causalexplain.independence package
  • Submodules
    • ConditionalIndependencies
      • ConditionalIndependencies._cache
      • ConditionalIndependencies.add()
      • ConditionalIndependencies.__str__()
      • ConditionalIndependencies.__repr__()
      • ConditionalIndependencies.__init__()
      • ConditionalIndependencies.__repr__()
      • ConditionalIndependencies.add()
    • SufficientSets
      • SufficientSets._cache
      • SufficientSets.add()
      • SufficientSets.__str__()
      • SufficientSets.__init__()
      • SufficientSets.add()
      • SufficientSets.__str__()
    • get_backdoor_paths()
    • get_paths()
    • find_colliders_in_path()
    • get_sufficient_sets_for_pair()
    • get_sufficient_sets()
    • get_conditional_independencies()
    • custom_main()
    • main()
    • dag_main()
    • get_edge_orientation()
    • estimate()
    • estimate_edge()
    • main()
    • select_features()
    • find_cluster_change_point()
    • main()
    • test()
    • GraphIndependence
      • GraphIndependence.__init__()
      • GraphIndependence.fit()
      • GraphIndependence.predict()
      • GraphIndependence.fit_predict()
      • GraphIndependence.compute_cond_indep_pvals()
    • HSIC_Values
      • HSIC_Values.__init__()
      • HSIC_Values.hsic
      • HSIC_Values.p_value
      • HSIC_Values.stat
      • HSIC_Values.independence
    • HSIC
      • HSIC.__init__()
      • HSIC.fit()
      • HSIC.p_value
      • HSIC.hsic
      • HSIC.stat
      • HSIC.independence
    • rbf_dot()
    • kernel_Delta_norm()
    • kernel_Delta()
    • kernel_Gaussian()
    • pairwise_mic()
    • fit_and_get_residuals()
    • run_feature_selection()
  • Module contents
• causalexplain.metrics package
  • Submodules
    • debug_()
    • pm_()
    • allDagsIntern()
    • allDagsJonas()
    • computePathMatrix()
    • computePathMatrix2()
    • compute_caus_order()
    • dag2cpdagAdj()
    • dSepAdji()
    • unique_rows()
    • SID()
    • get_indInAllOthers()
    • hammingDist()
    • main()
    • Metrics
      • Metrics.__init__()
      • Metrics.Tp
      • Metrics.Tn
      • Metrics.Fn
      • Metrics.Fp
      • Metrics.precision
      • Metrics.recall
      • Metrics.aupr
      • Metrics.f1
      • Metrics.shd
      • Metrics.sid
      • Metrics.to_dict()
      • Metrics.matplotlib_repr()
    • evaluate_graph()
  • Module contents
• causalexplain.models package
  • Submodules
    • MLP
      • MLP.device
      • MLP.Block
      • MLP.__init__()
      • MLP.forward()
      • MLP.predict_step()
      • MLP.training_step()
      • MLP.validation_step()
      • MLP.configure_optimizers()
      • MLP.predict()
    • DFF
      • DFF.__init__()
      • DFF.forward()
      • DFF.predict_step()
      • DFF.training_step()
      • DFF.validation_step()
      • DFF.configure_optimizers()
    • MDN
      • MDN.__init__()
      • MDN.configure_optimizers()
      • MDN.training_step()
      • MDN.validation_step()
      • MDN.common_step()
      • MDN.forward()
      • MDN.mdn_loss()
      • MDN.gaussian_probability()
      • MDN.mmd_loss()
      • MDN.add_noise()
      • MDN.g_sample()
      • MDN.sample()
    • ColumnsDataset
      • ColumnsDataset.__init__()
    • RBF
      • RBF.__init__()
      • RBF.get_bandwidth()
      • RBF.forward()
    • MMDLoss
      • MMDLoss.__init__()
      • MMDLoss.forward()
    • BaseModel
      • BaseModel.model
      • BaseModel.all_columns
      • BaseModel.callbacks
      • BaseModel.columns
      • BaseModel.logger
      • BaseModel.extra_trainer_args
      • BaseModel.scaler
      • BaseModel.train_loader
      • BaseModel.val_loader
      • BaseModel.n_rows
      • BaseModel.device
      • BaseModel.__init__()
      • BaseModel.init_logger()
      • BaseModel.init_callbacks()
      • BaseModel.init_data()
      • BaseModel.override_extras()
    • MLPModel
      • MLPModel.__init__()
      • MLPModel.train()
    • extract_weights()
    • see_weights_to_hidden()
    • see_weights_from_input()
    • plot_feature()
    • plot_features()
    • layer_weights()
    • summarize_weights()
    • identify_relationships()
    • infer_causal_relationships()
    • NNRegressor
      • NNRegressor.__init__()
      • NNRegressor.fit()
      • NNRegressor.predict()
      • NNRegressor.score()
      • NNRegressor.tune()
      • NNRegressor.tune_fit()
    • custom_main()
    • GBTRegressor
      • GBTRegressor.__init__()
      • GBTRegressor.random_state
      • GBTRegressor.fit()
      • GBTRegressor.predict()
      • GBTRegressor.score()
      • GBTRegressor.tune()
      • GBTRegressor.tune_fit()
    • custom_main()
  • Module contents
    • CausalExplain Models
    • GBTRegressor
      • GBTRegressor.__init__()
      • GBTRegressor.random_state
      • GBTRegressor.fit()
      • GBTRegressor.predict()
      • GBTRegressor.score()
      • GBTRegressor.tune()
      • GBTRegressor.tune_fit()
    • NNRegressor
      • NNRegressor.__init__()
      • NNRegressor.fit()
      • NNRegressor.predict()
      • NNRegressor.score()
      • NNRegressor.tune()
      • NNRegressor.tune_fit()

Submodules

This module contains the GraphDiscovery class which is responsible for creating, fitting, and evaluating causal discovery experiments.

class GraphDiscovery(experiment_name=None, model_type='rex', csv_filename=None, true_dag_filename=None, verbose=False, seed=42)

Bases: object

Attributes:

model

Methods

combine_and_evaluate_dags([prior])	Retrieve the DAG from the Experiment objects.
create_experiments()	Create an Experiment object for each regressor.
export(output_file)	This method exports the DAG to a DOT file.
fit_experiments([hpo_iterations, ...])	Fit the Experiment objects.
load(model_path)	Load the model from a pickle file.
plot([show_metrics, show_node_fill, title, ...])	This method plots the DAG using networkx and matplotlib.
printout_results(graph, metrics)	This method prints the DAG to stdout in hierarchical order.
run([hpo_iterations, bootstrap_iterations, ...])	Run the experiment.
save(full_filename_path)	Save the model as an Experiment object.

__init__(experiment_name=None, model_type='rex', csv_filename=None, true_dag_filename=None, verbose=False, seed=42)

Initializes a new instance of the GraphDiscovery class.

Parameters:

• experiment_name (str, optional) – The name of the experiment.

• model_type (str, optional) – The type of model to use. Valid options are: ‘rex’, ‘pc’, ‘fci’, ‘ges’, ‘lingam’, ‘cam’, ‘notears’.

• csv_filename (str, optional) – The filename of the CSV file containing the data.

• true_dag_filename (str, optional) – The filename of the DOT file containing the true causal graph.

• verbose (bool, optional) – Whether to print verbose output.

• seed (int, optional) – The random seed for reproducibility.

create_experiments()

Create an Experiment object for each regressor.

Parameters:

• dataset_name (str) – Name of the dataset

• true_dag (str) – Path to the true DAG DOT file

• regressors (list) – List of regressor types to create experiments for

• dataset_path (str) – Path to the input dataset

• output_path (str) – Path for output files

Returns:

A dictionary of Experiment objects

Return type:

dict

fit_experiments(hpo_iterations=None, bootstrap_iterations=None, prior=None, **kwargs)

Fit the Experiment objects.

Parameters:

• trainer (dict) – A dictionary of Experiment objects

• estimator (str) – The estimator to use (‘rex’ or other)

• verbose (bool, optional) – Whether to print verbose output. Defaults to False.

• hpo_iterations (int, optional) – Number of HPO trials for REX. Defaults to None.

• bootstrap_iterations (int, optional) – Number of bootstrap trials for REX. Defaults to None.

combine_and_evaluate_dags(prior=None)

Retrieve the DAG from the Experiment objects.

Parameters:

prior (List[List[str]], optional) – The prior to use for ReX. Defaults to None.

Returns:

The experiment object with the final DAG

Return type:

Experiment

run(hpo_iterations=None, bootstrap_iterations=None, prior=None, **kwargs)

Run the experiment.

Parameters:

• hpo_iterations (int, optional) – Number of HPO trials for REX. Defaults to None.

• bootstrap_iterations (int, optional) – Number of bootstrap trials for REX. Defaults to None.

save(full_filename_path)

Save the model as an Experiment object.

Parameters:

full_filename_path (str) – A full path where to save the model, including the filename.

load(model_path)

Load the model from a pickle file.

Parameters:

model_path (str) – Path to the pickle file containing the model

Returns:

The loaded Experiment object

Return type:

Experiment

printout_results(graph, metrics)

This method prints the DAG to stdout in hierarchical order.

export(output_file)

This method exports the DAG to a DOT file.

plot(show_metrics=False, show_node_fill=True, title=None, ax=None, figsize=(5, 5), dpi=75, save_to_pdf=None, layout='dot', **kwargs)

This method plots the DAG using networkx and matplotlib.

property model

Module contents

CausalExplain: A Python package for causal discovery and inference.

This package provides tools for discovering and analyzing causal relationships in data using various methods and algorithms.

class GraphDiscovery(experiment_name=None, model_type='rex', csv_filename=None, true_dag_filename=None, verbose=False, seed=42)

Bases: object

Attributes:

model

Methods

combine_and_evaluate_dags([prior])	Retrieve the DAG from the Experiment objects.
create_experiments()	Create an Experiment object for each regressor.
export(output_file)	This method exports the DAG to a DOT file.
fit_experiments([hpo_iterations, ...])	Fit the Experiment objects.
load(model_path)	Load the model from a pickle file.
plot([show_metrics, show_node_fill, title, ...])	This method plots the DAG using networkx and matplotlib.
printout_results(graph, metrics)	This method prints the DAG to stdout in hierarchical order.
run([hpo_iterations, bootstrap_iterations, ...])	Run the experiment.
save(full_filename_path)	Save the model as an Experiment object.

__init__(experiment_name=None, model_type='rex', csv_filename=None, true_dag_filename=None, verbose=False, seed=42)

Initializes a new instance of the GraphDiscovery class.

Parameters:

• experiment_name (str, optional) – The name of the experiment.

• model_type (str, optional) – The type of model to use. Valid options are: ‘rex’, ‘pc’, ‘fci’, ‘ges’, ‘lingam’, ‘cam’, ‘notears’.

• csv_filename (str, optional) – The filename of the CSV file containing the data.

• true_dag_filename (str, optional) – The filename of the DOT file containing the true causal graph.

• verbose (bool, optional) – Whether to print verbose output.

• seed (int, optional) – The random seed for reproducibility.

create_experiments()

Create an Experiment object for each regressor.

Parameters:

• dataset_name (str) – Name of the dataset

• true_dag (str) – Path to the true DAG DOT file

• regressors (list) – List of regressor types to create experiments for

• dataset_path (str) – Path to the input dataset

• output_path (str) – Path for output files

Returns:

A dictionary of Experiment objects

Return type:

dict

fit_experiments(hpo_iterations=None, bootstrap_iterations=None, prior=None, **kwargs)

Fit the Experiment objects.

Parameters:

• trainer (dict) – A dictionary of Experiment objects

• estimator (str) – The estimator to use (‘rex’ or other)

• verbose (bool, optional) – Whether to print verbose output. Defaults to False.

• hpo_iterations (int, optional) – Number of HPO trials for REX. Defaults to None.

• bootstrap_iterations (int, optional) – Number of bootstrap trials for REX. Defaults to None.

combine_and_evaluate_dags(prior=None)

Retrieve the DAG from the Experiment objects.

Parameters:

prior (List[List[str]], optional) – The prior to use for ReX. Defaults to None.

Returns:

The experiment object with the final DAG

Return type:

Experiment

run(hpo_iterations=None, bootstrap_iterations=None, prior=None, **kwargs)

Run the experiment.

Parameters:

• hpo_iterations (int, optional) – Number of HPO trials for REX. Defaults to None.

• bootstrap_iterations (int, optional) – Number of bootstrap trials for REX. Defaults to None.

save(full_filename_path)

Save the model as an Experiment object.

Parameters:

full_filename_path (str) – A full path where to save the model, including the filename.

load(model_path)

Load the model from a pickle file.

Parameters:

model_path (str) – Path to the pickle file containing the model

Returns:

The loaded Experiment object

Return type:

Experiment

printout_results(graph, metrics)

This method prints the DAG to stdout in hierarchical order.

export(output_file)

This method exports the DAG to a DOT file.

plot(show_metrics=False, show_node_fill=True, title=None, ax=None, figsize=(5, 5), dpi=75, save_to_pdf=None, layout='dot', **kwargs)

This method plots the DAG using networkx and matplotlib.

property model