uqmodels.modelization.ML_estimator package

Submodules

uqmodels.modelization.ML_estimator.baseline module

Implementation of usual prediction wrappers.

class uqmodels.modelization.ML_estimator.baseline.GBRQ_UQEstimator(list_estimators: List_Estimators | None = None, list_alpha: list = [0.025, 0.5, 0.975], type_UQ='quantile', name: str = 'GBRQ_UQEstimator', pretuned: bool = False, random_state: int | None = None)

Bases: QuantileUQEstimator

Uncertainty quantification approch based on Quantile Gradient Boosting : Instatiation OF QuantileUQEstimators using GBRQ scikilearn models.

Attributes/properties:

name (str): Name of the UQ method for future reference. type_UQ (str): Method family among the main categories of UQUQEstimators. list_estimators (Estimator): List of quantile estimator. list_alpha : list of alpha confidence level for each quantile estimators. pretuned (bool): Whether to disable parameter model tuning.

Mean Methods:

fit: Fit the list of quantile estimators. predict:(pred,UQ) Predict for each quantile estimators

fit(X: Array, y: Array, skip_format=False, **kwargs) → None

Fit GBRQ_UQEstimator list using QuantileUQEstimator fit methods. :param X: Features :param y: Target values

predict(X: Array, skip_format=False, **kwargs)

Perform the quantile estimation using QuantileUQEstimator predict methods.

Parameters:

X – Features

Returns:

Median prediction or None if Median quantile estimators is not in list_estimators UQ : List of quantiles estimatiors

Return type:

pred

set_fit_request(*, skip_format: bool | None | str = '$UNCHANGED$') → GBRQ_UQEstimator

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to fit.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, skip_format: bool | None | str = '$UNCHANGED$') → GBRQ_UQEstimator

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in predict.

Returns:

self – The updated object.

Return type:

object

class uqmodels.modelization.ML_estimator.baseline.GPR_UQEstimator(name: str = 'Gaussian_Process_UQ', kernel: Kernel | None = None, gp_alpha: float = 1e-06, drop_ratio: float = 0.0, rescale: bool = False, random_state: int | None = None)

Bases: MeanVarUQEstimator

Uncertainty quantification approch based on Gaussian Process Regressor. Instatiation OF MeanVarUQEstimator using GP scikilearn model. Warning GP has UQ limitation : see “The pitfalls of using Gaussian Process Regression for normative modeling”

fit(X, y, skip_format=False, **kwargs)

Fit procedure of Gaussian process

Parameters:

• X – Training vectors.

• y – Target values.

predict(X: Array, skip_format=False, **kwargs)

Perform the prediction task of the forecasting and uncertainty models on X.

Parameters:

X – Samples on which to perform the prediction.

Returns:

A tuple containing the forecast, the predicted lower and

upper quantiles and var.

set_fit_request(*, skip_format: bool | None | str = '$UNCHANGED$') → GPR_UQEstimator

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to fit.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, skip_format: bool | None | str = '$UNCHANGED$') → GPR_UQEstimator

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in predict.

Returns:

self – The updated object.

Return type:

object

class uqmodels.modelization.ML_estimator.baseline.REGML_UQEstimator(estimator=None, estimator_var=None, pretuned: bool = False, type_UQ: str = 'var', use_biais: bool = True, name: str = 'REGML_UQEstimator', var_min: float = 1e-06, rescale: bool = False, random_state: int | None = None)

Bases: MeanVarUQEstimator

Uncertainty quantification approch based on ML regression of bias and variance Instanciation of specific Pred-Biais-Var Regression scheme for uncertainty quantification

fit(X: Array, y: Array, skip_format=False, **kwargs) → None

Train y_lowerh forecasting and UQ models on (X,Y).

get_params(**kwargs) → Dict_params

Get parameters for this estimator.

Parameters:

deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:

params – Parameter names mapped to their values.

Return type:

dict

predict(X: Array, skip_format=False, **kwargs)

Perform the prediction task of the forecasting and UQ models on features (X).

Parameters:

X – Samples on which to perform the prediction.

Returns:

Prediction and UQmeasure

Return type:

pred, UQ

set_fit_request(*, skip_format: bool | None | str = '$UNCHANGED$') → REGML_UQEstimator

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to fit.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, skip_format: bool | None | str = '$UNCHANGED$') → REGML_UQEstimator

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in predict.

Returns:

self – The updated object.

Return type:

object

uqmodels.modelization.ML_estimator.random_forest_UQ module

class uqmodels.modelization.ML_estimator.random_forest_UQ.RF_UQEstimator(estimator=RandomForestRegressor(), pretuned=False, type_UQ='var', use_biais=True, rescale=True, n_jobs=4, beta=0.1, var_min=1e-05, random_state=None)

Bases: UQEstimator

Uncertainty quantification approch based on “local” sub-sampling UQ estimation from Random forest neighboorhood extraction

RF_extraction(X)

Random-forest subsampling statistics extraction “

Parameters:

X ([array]) – Features of elements.

Output:

Statistics array of shape (n_obs,dim): ‘Pred’ : Random forest forecast values ‘Biais’ : RF Biais computed as the sum of Oob Tree biais

‘UQ’Shape of UQ depends of the type_UQ !!!

IF mode=var:

UQ = Var RF variance computed as the sum of esiduals’ variance of the leaves

IF mode=2-var:

UQ = (Var_bot,Var_top), !!! 2-uple of 2D array !!! Var_bot : bot variance stimation (partial sum) as sum of negative residuals’ variance of the leaves Var_top : top variance stimation (partial sum) as sum of positive residuals’ variance of the leaves

IF mode=quantile:

UQ = (Q_bot,Q_top), !!! 2-uple of 2D array !!! ‘Q_bot’ : Bot quantile estimation (partial sum) as ponderation of leaves’ bot quantile ‘Q_top’ : Top quantile estimatuon (partial sum) as ponderation of leaves’ top quantile

Other advanced statistics. ‘Biais_oob’ : part of Biais ‘Var_E’ : Part of total variance (Law of total variance) ‘E_Var’ : Part of total variance (Law of total variance) ‘Var_oob’ : Part of total variance (related to biais)

fit(X, y, sample_weight=None, skip_format=False, **kwargs)

Train scikit RF model and then estimate and store leafs variance to uncertainty quantification purpose :param X: Features of the training set :type X: array :param y: Targets/Observations of the training set :type y: array

predict(X, beta=None, skip_format=False, **kwargs)

Predict both forecast and UQ estimations values :param X: Features of the data to forecast :type X: [type] :param beta: Miss-coverage target :type beta: [type]

Returns: pred ([array]): Forecast values UQ ([type]): UQmeasure

set_fit_request(*, sample_weight: bool | None | str = '$UNCHANGED$', skip_format: bool | None | str = '$UNCHANGED$') → RF_UQEstimator

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to fit.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

• sample_weight (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for sample_weight parameter in fit.

• skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in fit.

Returns:

self – The updated object.

Return type:

object

set_predict_request(*, beta: bool | None | str = '$UNCHANGED$', skip_format: bool | None | str = '$UNCHANGED$') → RF_UQEstimator

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

• True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

• False: metadata is not requested and the meta-estimator will not pass it to predict.

• None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

• str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a pipeline.Pipeline. Otherwise it has no effect.

Parameters:

• beta (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for beta parameter in predict.

• skip_format (str, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED) – Metadata routing for skip_format parameter in predict.

Returns:

self – The updated object.

Return type:

object

uqmodels.modelization.ML_estimator.random_forest_UQ.get_params_dict(estimator=None, pretuned=False, type_UQ='var_A&E', use_biais=True, rescale=True, n_jobs=4, beta=0.05, var_min=1e-05, n_estimators=125, max_depth=15, min_impurity_decrease=1e-05, ccp_alpha=1e-05, max_features=0.9, max_samples=0.7, random_state=None, min_samples_leaf=5, min_samples_split=5, **kwargs)

Provide a dict of paramaters to build an RF_UQEstimator :param estimator: RandomForestRegressor with meta-parameters :type estimator: _type_, optional :param pretuned: bool flag that freeze estimator. Defaults to False. :type pretuned: bool, optional :param type_UQ: nature of UQmeasure. Defaults to ‘var’. :type type_UQ: str, optional :param use_biais: use oob biais correction. Defaults to True. :type use_biais: bool, optional :param rescale: use rescale procedure. Defaults to True. :type rescale: bool, optional :param n_jobs: number of jobs used for parallelization purpose. Defaults to 4. :type n_jobs: int, optional :param beta: miss coverage targets in case of type_UQ = quantile :type beta: float, optional :param var_min: minimal variance. Defaults to 0.00001. :type var_min: float, optional

Returns:

dict_parameters

Module contents