mapie.mondrian.MondrianCP

class mapie.mondrian.MondrianCP(mapie_estimator: Union[MapieClassifier, MapieRegressor])[source]

Mondrian is a method for making conformal predictions for partition of individuals.

The Mondrian method is implemented in the MondrianCP class. It takes as input a MapieClassifier or MapieRegressor estimator and fits a model for each group of individuals. The MondrianCP class can then be used to run a conformal prediction procedure for each of these groups and hence achieve marginal coverage on each of them.

The underlying estimator must be used with cv=’prefit’ and the conformity score must be one of the following: - For MapieClassifier: ‘lac’, ‘score’, ‘cumulated_score’, ‘aps’ or ‘topk’ - For MapieRegressor: ‘absolute’ or ‘gamma’

Parameters
mapie_estimatorUnion[MapieClassifier, MapieRegressor]

The estimator for which the Mondrian method will be applied. It must be used with cv=’prefit’ and the conformity score must be one of the following: - For MapieClassifier: ‘lac’, ‘score’, ‘cumulated_score’, ‘aps’ or ‘topk’ - For MapieRegressor: ‘absolute’ or ‘gamma’

References

Vladimir Vovk, David Lindsay, Ilia Nouretdinov, and Alex Gammerman. Mondrian confidence machine. Technical report, Royal Holloway University of London, 2003

Examples

>>> import numpy as np
>>> from sklearn.linear_model import LogisticRegression
>>> from mapie.classification import MapieClassifier
>>> from mapie.mondrian import MondrianCP
>>> X_toy = np.arange(9).reshape(-1, 1)
>>> y_toy = np.stack([0, 0, 1, 0, 1, 2, 1, 2, 2])
>>> partition_toy = [0, 0, 0, 0, 1, 1, 1, 1, 1]
>>> clf = LogisticRegression(random_state=42).fit(X_toy, y_toy)
>>> mapie = MondrianCP(MapieClassifier(estimator=clf, cv="prefit")).fit(
...     X_toy, y_toy, partition=partition_toy
... )
>>> _, y_pi_mapie = mapie.predict(
...     X_toy, partition=partition_toy, alpha=0.4)
>>> print(y_pi_mapie[:, :, 0].astype(bool))
[[ True False False]
 [ True False False]
 [ True  True False]
 [ True  True False]
 [False  True False]
 [False  True  True]
 [False False  True]
 [False False  True]
 [False False  True]]
Attributes
partition_groupsNDArray

The unique groups of individuals for which the estimator was fitted

mapie_estimatorsDict

A dictionary containing the fitted conformal estimator for each group of individuals

__init__(mapie_estimator: Union[MapieClassifier, MapieRegressor])[source]
fit(X: Union[_SupportsArray[dtype[Any]], _NestedSequence[_SupportsArray[dtype[Any]]], bool, int, float, complex, str, bytes, _NestedSequence[Union[bool, int, float, complex, str, bytes]]], y: Union[_SupportsArray[dtype[Any]], _NestedSequence[_SupportsArray[dtype[Any]]], bool, int, float, complex, str, bytes, _NestedSequence[Union[bool, int, float, complex, str, bytes]]], partition: Union[_SupportsArray[dtype[Any]], _NestedSequence[_SupportsArray[dtype[Any]]], bool, int, float, complex, str, bytes, _NestedSequence[Union[bool, int, float, complex, str, bytes]]], **fit_params) → MondrianCP[source]

Fit the Mondrian method

Parameters
XArrayLike of shape (n_samples, n_features)

The input data

yArrayLike of shape (n_samples,) or (n_samples, n_outputs)

The target values

partitionArrayLike of shape (n_samples,)

The groups of individuals. Must be defined by integers. There must be at least 2 individuals per group.

**fit_params

Additional keyword arguments to pass to the estimator’s fit method that may be specific to the Mapie estimator used

predict(X: Union[_SupportsArray[dtype[Any]], _NestedSequence[_SupportsArray[dtype[Any]]], bool, int, float, complex, str, bytes, _NestedSequence[Union[bool, int, float, complex, str, bytes]]], partition: Union[_SupportsArray[dtype[Any]], _NestedSequence[_SupportsArray[dtype[Any]]], bool, int, float, complex, str, bytes, _NestedSequence[Union[bool, int, float, complex, str, bytes]]], alpha: Optional[Union[float, Iterable[float]]] = None, **predict_params) → Union[ndarray[Any, dtype[_ScalarType_co]], Tuple[ndarray[Any, dtype[_ScalarType_co]], ndarray[Any, dtype[_ScalarType_co]]]][source]

Perform conformal prediction for each group of individuals

Parameters
XArrayLike of shape (n_samples, n_features)

The input data

partitionArrayLike of shape (n_samples,), optional

The groups of individuals. Must be defined by integers.

By default None.

alphafloat or Iterable[float], optional

The desired coverage level(s) for each group.

By default None.

**predict_params

Additional keyword arguments to pass to the estimator’s predict method that may be specific to the Mapie estimator used

Returns
y_predNDArray of shape (n_samples,) or (n_samples, n_outputs)

The predicted values

y_pssNDArray of shape (n_samples, n_outputs, n_alpha)

The predicted sets for the desired levels of coverage

set_fit_request(*, partition: Union[bool, None, str] = '$UNCHANGED$') → MondrianCP

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.

New 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. Otherwise it has no effect.

Parameters
partitionstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for partition parameter in fit.

Returns
selfobject

The updated object.

set_predict_request(*, alpha: Union[bool, None, str] = '$UNCHANGED$', partition: Union[bool, None, str] = '$UNCHANGED$') → MondrianCP

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.

New 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. Otherwise it has no effect.

Parameters
alphastr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for alpha parameter in predict.

partitionstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for partition parameter in predict.

Returns
selfobject

The updated object.

Examples using mapie.mondrian.MondrianCP