PWK

class mlquantify.neighbors.PWK(alpha=1, n_neighbors=10, algorithm='auto', metric='euclidean', leaf_size=30, p=2, metric_params=None, n_jobs=None)

Probabilistic Weighted k-Nearest Neighbour (PWK) quantifier.

Targets prior probability shift. PWK is an aggregative Classify-and-Count quantifier — it shares the standard fit / predict / aggregate interface of CC — but its classifier is a k-nearest-neighbour rule modified for quantification (PWKCLF): each neighbour’s vote is re-weighted by a class-specific factor (controlled by alpha) so the count is not dominated by the majority class. Unlike the other aggregative quantifiers, PWK therefore takes no external estimator parameter: the modified k-NN is intrinsic to the method.

Parameters:

alphafloat, default=1

Imbalance-correction exponent. 1 applies the standard inverse-size weighting; higher values further amplify minority-class neighbours.

n_neighborsint, default=10

Number of nearest neighbours considered for each test instance.

algorithm{‘auto’, ‘ball_tree’, ‘kd_tree’, ‘brute’}, default=’auto’

Neighbour-search algorithm.

• 'auto' : pick the best of the below from the fitted data (default).

• 'ball_tree' : ball-tree index; good in higher dimensions.

• 'kd_tree' : k-d tree index; fast in low dimensions.

• 'brute' : exhaustive search; exact, best for small data.

metricstr, default=’euclidean’

Distance metric for the neighbour search.

leaf_sizeint, default=30

Leaf size for the tree-based algorithms (speed/memory trade-off).

pint, default=2

Power parameter for the Minkowski metric (1 = Manhattan, 2 = Euclidean).

metric_paramsdict or None, default=None

Additional keyword arguments for the metric function.

n_jobsint or None, default=None

Number of parallel jobs for the neighbour search.

Attributes:

estimatorPWKCLF

The underlying weighted k-NN classifier (built from the parameters above; not an argument).

estimator_PWKCLF

The fitted classifier.

classes_ndarray of shape (n_classes,)

Class labels seen during fit.

See also

CC

Plain classify-and-count baseline.

ACC

Adjusted count for binary prior shift.

Notes

PWK is a classify-and-count method whose only quantification-specific ingredient is the imbalance re-weighting; it needs no separate scorer and handles multiclass directly, but inherits k-NN’s sensitivity to feature scaling and dimensionality. Because it subclasses CC, aggregate (with its optional classes argument) is available too.

References

References

[1]

Barranquero, J., Díez, J., & del Coz, J. J. (2013). Quantification-Oriented Learning Based on Reliable Classifiers. Pattern Recognition, 48(2), 591–604.

Examples

>>> from mlquantify.neighbors import PWK
>>> from sklearn.datasets import make_classification
>>> X, y = make_classification(n_samples=200, random_state=42)
>>> q = PWK(alpha=1.5, n_neighbors=5).fit(X, y)
>>> q.predict(X)
{0: ..., 1: ...}

aggregate(predictions, classes=None)

Aggregate predictions into class prevalence estimates.

Parameters:

predictionsndarray of shape (n_samples,) or (n_samples, n_classes)

Estimator predictions on test data. Can be probabilities (n_samples, n_classes) or class labels (n_samples,).

classesarray-like of shape (n_classes,) or None, default=None

Class labels the output must report, in order. When given, every class appears in the result even if absent from predictions (with prevalence 0). When None, the classes seen during fit are used; if the quantifier is unfitted, they are inferred from the predictions.

Returns:

ndarray of shape (n_classes,)

Class prevalence estimates.

Examples

>>> from mlquantify.counting import CC
>>> import numpy as np
>>> q = CC()
>>> predictions = np.random.rand(200)
>>> q.aggregate(predictions)
{0: ..., 1: ...}

classify(X)

Classify test instances using the underlying weighted k-NN estimator.

Returns hard class labels produced by PWKCLF without any prevalence-level aggregation.

Parameters:

Xarray-like of shape (n_samples, n_features)

Test feature matrix.

Returns:

labelsndarray of shape (n_samples,)

Predicted class label for each test instance.

Examples

>>> from mlquantify.neighbors import PWK
>>> from sklearn.datasets import make_classification
>>> X, y = make_classification(n_samples=200, random_state=42)
>>> q = PWK(alpha=1.5, n_neighbors=5).fit(X, y)
>>> labels = q.classify(X[:5])

fit(X, y, estimator_fitted=False, *args, **kwargs)

Fit the quantifier using the provided data and estimator.

get_metadata_routing()

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:

routingMetadataRequest

A MetadataRequest encapsulating routing information.

get_params(deep=True)

Get parameters for this estimator.

Parameters:

deepbool, default=True

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:

paramsdict

Parameter names mapped to their values.

predict(X)

Predict class prevalences for the given data.

save_quantifier(path: str | None = None) → None

Save the quantifier instance to a file.

set_fit_request(*, estimator_fitted: bool | None | str = '$UNCHANGED$') → PWK

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config). Please check the 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.

Parameters:

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

Metadata routing for estimator_fitted parameter in fit.

Returns:

selfobject

The updated object.

set_params(**params)

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:

**paramsdict

Estimator parameters.

Returns:

selfestimator instance

Estimator instance.