dowhy.gcm.ml package

Submodules

dowhy.gcm.ml.autogluon module

dowhy.gcm.ml.classification module

Functions and classes in this module should be considered experimental, meaning there might be breaking API changes in the future.

class dowhy.gcm.ml.classification.ClassificationModel[source]

Bases: PredictionModel

abstract property classes: List[str]

abstract predict_probabilities(X: array) → ndarray[source]

class dowhy.gcm.ml.classification.SklearnClassificationModel(sklearn_mdl: Any)[source]

Bases: SklearnRegressionModel, ClassificationModel

property classes: List[str]

clone()[source]: Clones the prediction model using the same hyper parameters but not fitted. :return: An unfitted clone of the prediction model.

predict_probabilities(X: array) → ndarray[source]

dowhy.gcm.ml.classification.create_ada_boost_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_extra_trees_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_gaussian_nb_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_gaussian_process_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_hist_gradient_boost_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_knn_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_logistic_regression_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_polynom_logistic_regression_classifier(degree: int = 3, **kwargs_logistic_regression) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_random_forest_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.classification.create_support_vector_classifier(**kwargs) → SklearnClassificationModel[source]

dowhy.gcm.ml.prediction_model module

class dowhy.gcm.ml.prediction_model.PredictionModel[source]

Bases: object

Represents general prediction model implementations. Each prediction model should provide a fit and a predict method.

abstract clone()[source]

Clones the prediction model using the same hyper parameters but not fitted.

Returns:: An unfitted clone of the prediction model.

abstract fit(X: ndarray, Y: ndarray) → None[source]

abstract predict(X: ndarray) → ndarray[source]

dowhy.gcm.ml.regression module

Functions and classes in this module should be considered experimental, meaning there might be breaking API changes in the future.

class dowhy.gcm.ml.regression.InvertibleExponentialFunction[source]

Bases: InvertibleFunction

evaluate(X: ndarray) → ndarray[source]: Applies the function on the input.

evaluate_inverse(X: ndarray) → ndarray[source]: Returns the outcome of applying the inverse of the function on the inputs.

class dowhy.gcm.ml.regression.InvertibleFunction[source]

Bases: object

abstract evaluate(X: ndarray) → ndarray[source]: Applies the function on the input.

abstract evaluate_inverse(X: ndarray) → ndarray[source]: Returns the outcome of applying the inverse of the function on the inputs.

class dowhy.gcm.ml.regression.InvertibleIdentityFunction[source]

Bases: InvertibleFunction

evaluate(X: ndarray) → ndarray[source]: Applies the function on the input.

evaluate_inverse(X: ndarray) → ndarray[source]: Returns the outcome of applying the inverse of the function on the inputs.

class dowhy.gcm.ml.regression.InvertibleLogarithmicFunction[source]

Bases: InvertibleFunction

evaluate(X: ndarray) → ndarray[source]: Applies the function on the input.

evaluate_inverse(X: ndarray) → ndarray[source]: Returns the outcome of applying the inverse of the function on the inputs.

class dowhy.gcm.ml.regression.SklearnRegressionModel(sklearn_mdl: Any)[source]

Bases: PredictionModel

General wrapper class for sklearn models.

clone()[source]: Clones the prediction model using the same hyper parameters but not fitted. :return: An unfitted clone of the prediction model.

fit(X: ndarray, Y: ndarray) → None[source]

predict(X: array) → ndarray[source]

property sklearn_model: Any

dowhy.gcm.ml.regression.create_ada_boost_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_elastic_net_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_extra_trees_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_gaussian_process_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_hist_gradient_boost_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_knn_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_lasso_lars_ic_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_lasso_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_linear_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_linear_regressor_with_given_parameters(coefficients: ndarray, intercept: float = 0, **kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_polynom_regressor(degree: int = 2, **kwargs_linear_model) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_random_forest_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_ridge_regressor(**kwargs) → SklearnRegressionModel[source]

dowhy.gcm.ml.regression.create_support_vector_regressor(**kwargs) → SklearnRegressionModel[source]

Module contents

This module defines implementations of PredictionModel used by the different FunctionalCausalModel implementations, such as PostNonlinearModel or AdditiveNoiseModel.