Contents

This Page

  1. Docs
  2. actableai.causal package

actableai.causal package

Subpackages

Submodules

actableai.causal.evaluation module

actableai.causal.models module

class actableai.causal.models.AAICausalEstimator(model_params: Optional[dict] = None, scorer: Optional[list] = None, has_categorical_treatment: bool = False, has_binary_outcome: bool = False)

Bases: object

effect(X: Optional[numpy.ndarray] = None, T0: Optional[numpy.ndarray] = None, T1: Optional[numpy.ndarray] = None, alpha: float = 0.05)

Compute heterogeneous treatment effect

Parameters
  • X – (m, d_x) matrix. Features for each sample. Default to None.
  • T0 – (m, d_t) maxtrix or vector of length m. Base treatment for each sample. Default to None.
  • T1 – (m, d_t) maxtrix or vector of length m. Target treatment for each sample. Default to None.
Returns: Tuple
np.ndarray: (m, d_t) matrix. Estimated treatment effect. np.ndarray: Lower bound of the confidence interval. np.ndarray: Upper bound of the confidence interval.
fit(Y: numpy.ndarray, T: numpy.ndarray, X: Optional[numpy.ndarray] = None, W: Optional[numpy.ndarray] = None, Z: Optional[numpy.ndarray] = None, label_t: str = 't', label_y: str = 'y', target_units: str = 'ate', trials: int = 3, tune_params=None, cv: Union[str, int] = 'auto', feature_importance: bool = False, model_directory: Optional[str] = None, hyperparameters_t: Optional[dict] = None, hyperparameters_y: Optional[dict] = None, hyperparameters_auto: Optional[bool] = True, presets: str = 'medium_quality_faster_train', random_state: Optional[int] = None, mc_iters: str = 'auto', remove_outliers: bool = True, contamination: float = 0.05, num_gpus: int = 0, drop_unique: bool = True, drop_useless_features: bool = True, hyperparameter_tune_kwargs: Optional[dict] = None, dml_estimator: Optional[actableai.models.causal_inference_estimators.base.EstimatorBaseParams] = None)

Function fits a causal model with a single deterministic model.

Parameters

Fit casaul model with model and hyper-parameter search.

Parameters
  • Y (np.ndarray) – (n × d_y) matrix or vector of length n. Outcomes for each sample
  • T (np.ndarray) – (n × d_t) matrix or vector of length n. Treatments for each sample
  • X (np.ndarray, optional) – (n × d_x) matrix. Defaults to None. Features for each sample
  • W (np.ndarray, optional) – (n × d_w) matrix. Defaults to None. Controls for each sample
  • Z (np.ndarray, optional) – (n × d_z) matrix. Defaults to None. Instruments for each sample
  • target_units (str, optional) – either “ate”, “att” or “atc”. Default to “ate”
  • test_size (float) – test size in percent. Defaults to 20
  • trials (int, optional) – number of trials for hyperparameter tuning experiment
  • tune_params (dict, optional) – dictionary of tune parameters
  • max_concurrent (int, optional) – max concurcent
  • scheduler (object, optional) – tune scheduler object
  • stopper (object, optional) – tune stopper object
class actableai.causal.models.PolynomialLinearPredictor(path: Optional[str] = None, name: Optional[str] = None, problem_type: Optional[str] = None, eval_metric: Optional[Union[str, autogluon.core.metrics.Scorer]] = None, hyperparameters=None)

Bases: autogluon.core.models.abstract.abstract_model.AbstractModel

exception actableai.causal.models.UnsupportedTargetUnitsIVMethods

Bases: ValueError

exception actableai.causal.models.UnsupportedTargetUnitsMultipleTreatments

Bases: ValueError

exception actableai.causal.models.UntrainedModelException

Bases: ValueError

actableai.causal.params module

class actableai.causal.params.BaseCausalEstimatorParams

Bases: abc.ABC

abstract build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
abstract tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.CausalForestDMLSingleContTreatmentParams(model_y='RandomForestRegressor', model_t='RandomForestRegressor', criterion='mse', n_estimators=(100, 2000), min_impurity_decrease=(0.0005, 0.005), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'CausalForestDMLSingleContTreatment'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.DRLearnerSingleBinaryTreatmentParams(model_propensity='RandomForestClassifier', model_regression='RandomForestRegressor', model_final='RandomForestRegressor', min_samples_leaf=(5, 20), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'DRLearnerSingleBinaryTreatmentParams'
build_estimator(params)

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.DeepIVParams(num_instruments, num_effect_modifiers, num_treatments, num_gaussian_mixtures=(2, 12), num_keras_fit_epoch=(20, 50), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'DeepIVParams'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.DomainAdaptationLearnerSingleBinaryTreatmentParams(models='RandomForestRegressor', final_models='RandomForestRegressor', propensity_model='RandomForestClassifier', min_samples_leaf=(5, 20), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'DomainAdaptationLearnerSingleBinaryTreatmentParams'
build_estimator(params)

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLCategoricalOutcomeAGParams(label_t, label_y, model_directory, hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLCategoricalOutcomeAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLCategoricalOutcomeParams(model_y='MultiTaskElasticNetCV', model_t='RandomForestRegressor', l1_ratio=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLCategoricalOutcome'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLCategoricalTreatmentAGParams(label_t, label_y, model_directory, hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLCategoricalTreatmentAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLCategoricalTreatmentAndOutcomeAGParams(label_t, label_y, model_directory, hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLCategoricalTreatmentAndOutcomeAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLCategoricalTreatmentAndOutcomeParams(model_y='MultiTaskElasticNetCV', model_t='MultiTaskElasticNetCV', l1_ratio_y=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), l1_ratio_t=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLCategoricalTreatmentAndOutcome'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLCategoricalTreatmentParams(model_y='RandomForestRegressor', model_t='MultiTaskElasticNetCV', l1_ratio=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLCategoricalTreatment'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLSingleBinaryOutcomeAGParams(label_t, label_y, model_directory, hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLSingleBinaryOutcomeAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLSingleBinaryOutcomeParams(model_y='RandomForestClassifier', model_t='RandomForestRegressor', min_samples_leaf=(5, 20), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLSingleBinaryOutcome'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLSingleBinaryTreatmentAGParams(label_t, label_y, model_directory, hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLSingleBinaryTreatmentAGParams'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLSingleBinaryTreatmentParams(model_y='RandomForestRegressor', model_t='RandomForestClassifier', min_samples_leaf=(5, 20), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLSingleBinaryTreatment'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLSingleContTreatmentAGParams(label_t, label_y, model_directory, hyperparameters=None, random_state=123, presets='best_quality', cv=10, featurizer=None)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLSingleContTreatmentAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.LinearDMLSingleContTreatmentParams(model_y='RandomForestRegressor', model_t='RandomForestRegressor', random_state=123, cv=5, mc_iters=None)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'LinearDMLSingleContTreatment'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLCategoricalOutcomeAGParams(label_t, label_y, model_directory, polyfeat_degree=(1, 4), hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLCategoricalOutcomeAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLCategoricalOutcomeParams(model_y='RandomForestRegressor', model_t='MultiTaskElasticNetCV', polyfeat_degree=(1, 4), l1_ratio=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLCategoricalOutcome'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLCategoricalTreatmentAGParams(label_t, label_y, model_directory, polyfeat_degree=(1, 4), hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLCategoricalTreatmentAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLCategoricalTreatmentAndOutcomeAGParams(label_t, label_y, model_directory, polyfeat_degree=(1, 4), hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLCategoricalTreatmentAndOutcomeAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLCategoricalTreatmentAndOutcomeParams(model_y='MultiTaskElasticNetCV', model_t='MultiTaskElasticNetCV', polyfeat_degree=(1, 4), l1_ratio_y=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), l1_ratio_t=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLCategoricalTreatmentAndOutcome'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLCategoricalTreatmentParams(model_y='RandomForestRegressor', model_t='MultiTaskElasticNetCV', polyfeat_degree=(1, 4), l1_ratio=(0.1, 0.5, 0.7, 0.9, 0.95, 0.99, 1), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLCategoricalTreatment'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLSingleBinaryOutcomeAGParams(label_t, label_y, model_directory, polyfeat_degree=(1, 4), hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLSingleBinaryOutcomeAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLSingleBinaryOutcomeParams(model_y='RandomForestClassifier', model_t='RandomForestRegressor', min_samples_leaf=(5, 20), polyfeat_degree=(1, 4), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLSingleBinaryOutcome'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLSingleBinaryTreatmentAGParams(label_t, label_y, model_directory, polyfeat_degree=(1, 4), hyperparameters=None, random_state=123, presets='best_quality', cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLSingleBinaryTreatmentAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLSingleBinaryTreatmentParams(model_y='RandomForestRegressor', model_t='RandomForestClassifier', min_samples_leaf=(5, 20), polyfeat_degree=(1, 4), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLSingleBinaryTreatment'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLSingleContTreatmentAGParams(label_t, label_y, model_directory, polyfeat_degree=4, hyperparameters=None, presets='best_quality', random_state=123, cv=10)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLSingleContTreatmentAG'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
class actableai.causal.params.SparseLinearDMLSingleContTreatmentParams(model_y='RandomForestRegressor', model_t='RandomForestRegressor', polyfeat_degree=(1, 4), random_state=123)

Bases: actableai.causal.params.BaseCausalEstimatorParams

MODEL_NAME = 'SparseLinearDMLSingleContTreatment'
build_estimator(params={})

Returns an estimator object given model config dictionary

Parameters
params (dict, optional) – Dictionary of model parameters. Defaults to {}.
Returns
A causal estimator that supports .fit() function.
Return type
object
tune_config()

Returns a configuration dictionary of the model parameters

Returns
Dictionary of parameters. Naming conventions:
Return type
dict
exception actableai.causal.params.UnsupportedCausalScenario

Bases: ValueError

actableai.causal.params.get_model_params(is_single_treatment, has_categorical_treatment, is_single_binary_treatment, is_single_outcome, has_categorical_outcome, is_single_binary_outcome, has_effect_modifiers_and_common_causes, label_t, label_y, model_directory, hyperparameters=None, presets='best_quality', cv=10, featurizer=None, mc_iters=None)
actableai.causal.params.get_rscorer(is_single_treatment, has_categorical_treatment, is_single_binary_treatment, is_single_outcome, has_categorical_outcome, is_single_binary_outcome)

actableai.causal.predictors module

class actableai.causal.predictors.DataFrameTransformer(column_names: Optional[List[str]] = None)

Bases: sklearn.base.TransformerMixin, sklearn.base.BaseEstimator

fit(X)
fit_transform(X, y=None, **fit_params)

Fit to data, then transform it.

Fits transformer to X and y with optional parameters fit_params and returns a transformed version of X.

Parameters
  • X (array-like of shape (n_samples, n_features)) – Input samples.
  • y (array-like of shape (n_samples,) or (n_samples, n_outputs), default=None) – Target values (None for unsupervised transformations).
  • **fit_params (dict) – Additional fit parameters.
Returns

X_new – Transformed array.

Return type

ndarray array of shape (n_samples, n_features_new)

transform(X)
class actableai.causal.predictors.LinearRegressionWrapper(*, fit_intercept=True, normalize='deprecated', copy_X=True, n_jobs=None, positive=False)

Bases: sklearn.linear_model._base.LinearRegression

score(X, y, sample_weight=None)

Return the coefficient of determination of the prediction.

The coefficient of determination \(R^2\) is defined as \((1 - \frac{u}{v})\), where \(u\) is the residual sum of squares ((y_true - y_pred)** 2).sum() and \(v\) is the total sum of squares ((y_true - y_true.mean()) ** 2).sum(). The best possible score is 1.0 and it can be negative (because the model can be arbitrarily worse). A constant model that always predicts the expected value of y, disregarding the input features, would get a \(R^2\) score of 0.0.

Parameters
  • X (array-like of shape (n_samples, n_features)) – Test samples. For some estimators this may be a precomputed kernel matrix or a list of generic objects instead with shape (n_samples, n_samples_fitted), where n_samples_fitted is the number of samples used in the fitting for the estimator.
  • y (array-like of shape (n_samples,) or (n_samples, n_outputs)) – True values for X.
  • sample_weight (array-like of shape (n_samples,), default=None) – Sample weights.
Returns

score – \(R^2\) of self.predict(X) wrt. y.

Return type

float

Notes

The \(R^2\) score used when calling score on a regressor uses multioutput='uniform_average' from version 0.23 to keep consistent with default value of r2_score(). This influences the score method of all the multioutput regressors (except for MultiOutputRegressor).

class actableai.causal.predictors.SKLearnMultilabelWrapper(ag_predictor, x_w_columns: Optional[List] = None, hyperparameters: Optional[Union[List, Dict]] = None, presets: Optional[str] = 'best_quality', ag_args_fit: Optional[Dict] = None, feature_generator: Optional[autogluon.features.generators.abstract.AbstractFeatureGenerator] = None, holdout_frac: Optional[float] = None)

Bases: actableai.causal.predictors.SKLearnWrapper

add_output(train_data, y)

Method to add the output to the train_data. Necessary for AutoGLuon

Parameters
  • train_data – Training data
  • y – Groundtruth target
Returns

Original DataFrame with target added

Return type

pd.DataFrame

feature_importance() → pandas.core.frame.DataFrame

Feature importance of the model

Raises
NotFittedError – If the model hasn’t been fitted before
Returns
Feature importances
Return type
pd.DataFrame
score(X, y, sample_weight=None) → Dict[str, Any]

Score for the model

Parameters
  • X – Values on which the score is calculated
  • y – Groundtruth values for the target
  • sample_weight – Sample weight for each class
Raises

UnsupportedProblemType – If the problem_type of the TabularPredictor is not right

Returns

Score results
  • accuracy | r2: Score metric
  • y: Original data used for calculating the metric
  • y_pred: Predicted data uised for calculating the metric
  • sample_weight: Weights used for each class if usable

Return type

Dict

set_sample_weight(sample_weight=None)

Set samples weight for the model

Parameters
sample_weight – Sample weights for each class
class actableai.causal.predictors.SKLearnTabularWrapper(ag_predictor: autogluon.tabular.predictor.predictor.TabularPredictor, x_w_columns: Optional[List] = None, hyperparameters: Optional[Union[List, Dict]] = None, presets: Optional[str] = 'best_quality', ag_args_fit: Optional[Dict] = None, feature_generator: Optional[autogluon.features.generators.abstract.AbstractFeatureGenerator] = None, holdout_frac: Optional[float] = None, hyperparameter_tune_kwargs: Optional[Dict] = None, num_gpus: Optional[int] = 0)

Bases: actableai.causal.predictors.SKLearnWrapper

add_output(train_data: pandas.core.frame.DataFrame, y) → pandas.core.frame.DataFrame

Method to add the output to the train_data. Necessary for AutoGLuon

Parameters
  • train_data – Training data
  • y – Groundtruth target
Returns

Original DataFrame with target added

Return type

pd.DataFrame

feature_importance() → pandas.core.frame.DataFrame

Feature importance of the model

Raises
NotFittedError – If the model hasn’t been fitted before
Returns
Feature importances
Return type
pd.DataFrame
score(X, y, sample_weight=None) → Dict[str, Any]

Score for the model

Parameters
  • X – Values on which the score is calculated
  • y – Groundtruth values for the target
  • sample_weight – Sample weight for each class
Raises

UnsupportedProblemType – If the problem_type of the TabularPredictor is not right

Returns

Score results
  • accuracy | r2: Score metric
  • y: Original data used for calculating the metric
  • y_pred: Predicted data uised for calculating the metric
  • sample_weight: Weights used for each class if usable

Return type

Dict

set_sample_weight(sample_weight=None) → None

Set samples weight for the model

Parameters
sample_weight – Sample weights for each class
class actableai.causal.predictors.SKLearnWrapper(ag_predictor: Union[actableai.utils.multilabel_predictor.MultilabelPredictor, autogluon.tabular.predictor.predictor.TabularPredictor], x_w_columns: Optional[List] = None, hyperparameters: Optional[Union[List, Dict]] = None, presets: Optional[str] = 'best_quality', ag_args_fit: Optional[Dict] = None, feature_generator: Optional[autogluon.features.generators.abstract.AbstractFeatureGenerator] = None, holdout_frac: Optional[float] = None, hyperparameter_tune_kwargs: Optional[Dict] = None, num_gpus: Optional[int] = 0)

Bases: abc.ABC

abstract add_output(train_data: pandas.core.frame.DataFrame, y: numpy.ndarray) → pandas.core.frame.DataFrame

Method to add the output to the train_data. Necessary for AutoGLuon

Parameters
  • train_data – Training data
  • y – Groundtruth target
Returns

Original DataFrame with target added

Return type

pd.DataFrame

abstract feature_importance() → pandas.core.frame.DataFrame

Feature importance of the model

Raises
NotFittedError – If the model hasn’t been fitted before
Returns
Feature importances
Return type
pd.DataFrame
fit(X: Union[pandas.core.frame.DataFrame, numpy.ndarray], y: Union[pandas.core.series.Series, numpy.ndarray], sample_weight: Optional[List] = None) → actableai.causal.predictors.SKLearnWrapper

Fit the model

Parameters
  • X (features, n) – Features to fit the model
  • y (n) – Groundtruth to fit the model
  • sample_weight – Optional sample weight to use in case of classification
Returns

The fitted SKLearnWrapper

Return type

self

predict(X: Union[pandas.core.frame.DataFrame, numpy.ndarray]) → Union[pandas.core.frame.DataFrame, pandas.core.series.Series]

Infer the result from the data

Parameters
X – Data used to make the prediction
Returns
Prediction
Return type
np.ndarray
predict_proba(X: pandas.core.frame.DataFrame) → numpy.ndarray
Predict the probabilities of classes if classification problem else returns
the result.
Parameters
X (features, n) – Data to predict each probability class
Returns
Probabilities for each class
Return type
np.ndarray (class_number, n)
abstract score(X: Union[pandas.core.frame.DataFrame, numpy.ndarray], y: numpy.ndarray, sample_weight: Optional[Iterable] = None) → Dict[str, Any]

Score for the model

Parameters
  • X – Values on which the score is calculated
  • y – Groundtruth values for the target
  • sample_weight – Sample weight for each class
Raises

UnsupportedProblemType – If the problem_type of the TabularPredictor is not right

Returns

Score results
  • accuracy | r2: Score metric
  • y: Original data used for calculating the metric
  • y_pred: Predicted data uised for calculating the metric
  • sample_weight: Weights used for each class if usable

Return type

Dict

abstract set_sample_weight(sample_weight: Optional[Iterable] = None) → None

Set samples weight for the model

Parameters
sample_weight – Sample weights for each class
exception actableai.causal.predictors.UnsupportedPredictorType

Bases: ValueError

Raised when the predictor type is not supported.

exception actableai.causal.predictors.UnsupportedProblemType

Bases: ValueError

Raised when the problem type is not supported.

actableai.causal.tree_utils module

class actableai.causal.tree_utils.Node(left: Optional[actableai.causal.tree_utils.Node] = None, right: Optional[actableai.causal.tree_utils.Node] = None, label: Optional[str] = None, conditions: list = [], is_right_child: bool = False, num_right_ancestors: int = 0)

Bases: object

actableai.causal.tree_utils.convert_label_to_categorical(root_node, cat_name, cat_vals)

cat_name: name of the categorical variable cat_vals: mapping of numeric to categorical value

actableai.causal.tree_utils.get_cat_constraint(cat_name, cat_vals, constraints)

cat_name: name of categorical variables cat_vals: dictionary mapping numeric to categorical values constraints: list of conditions

actableai.causal.tree_utils.make_pretty_tree(tree_dot: str, cat_names: list[str], cat_vals_list: list[dict]) → str

Make a pretty tree from a tree_dot.

Parameters
  • tree_dot – String representation of the tree in dot format.
  • cat_names – List of categorical variables.
  • cat_vals_list – List of dictionaries mapping numeric to categorical values.
Returns

_description_

Return type

_type_

actableai.causal.tree_utils.negate_condition(condition: str) → Optional[str]

Converts a condition to its negation.

Parameters
condition – Condition to be negated.
Returns
Negated condition.
Return type
Optional[str]
actableai.causal.tree_utils.parse_tree_dot(tree_dot: str) → Tuple[str, dict]

Parser for tree_dot.

Parameters
tree_dot – String representation of the tree in dot format.
Returns
  • tree_dot: Parsed tree in dot format.
  • node_dict: Dictionary mapping node ids to nodes.
Return type
Tuple

Module contents

class actableai.causal.OneHotEncodingTransformer(df)

Bases: object

fit_transform(X, y=None)
transform(X)
actableai.causal.autogluon_hyperparameters()
actableai.causal.has_categorical_column(df, columns)
actableai.causal.prepare_sanitize_data(pd_table: pandas.core.frame.DataFrame, treatments: List[str], outcomes: List[str], effect_modifiers: List[str], common_causes: List[str]) → pandas.core.frame.DataFrame

Drop NAN rows for treatments and outcomes then impute effect_modifiers and common_causes

Parameters
  • pd_table – Table to prepare
  • treatments – List of treatments
  • outcomes – List of outcomes
  • effect_modifiers – List of effect modifiers
  • common_causes – List of common causes
Returns

New table sanitized and imputed

Return type

pd.DataFrame

© Copyright 2022, ActableAI Team. Created using Sphinx.