holisticai.bias.mitigation.LPDebiaserBinary#

class holisticai.bias.mitigation.LPDebiaserBinary(*args, **kargs)[source]#

Linear Programmin Debiaser [1] is a postprocessing algorithms designed to debias pretrained classifiers. The algorithm use constraints such as Equalized Odds and Equalized Opportunity.

Parameters

constraintstr: Strategy used to evalute the cost function The available contraints are: “EqualizedOdds”, “EqualizedOpportunity”

Examples

>>> from holisticai.bias.mitigation import LPDebiaserBinary
>>> mitigator = LPDebiaserBinary()
>>> mitigator.fit(y, y_proba, group_a, group_b)
>>> test_data_transformed = mitigator.predict(y_pred, group_a, group_b)

References

fit(y: ndarray, group_a: ndarray, group_b: ndarray, y_pred: ndarray | None = None, y_proba: ndarray | None = None)[source]#

Compute parameters for Linear Programming Debiaser. For binary classification y_pred or y_proba can be used. For Multiclass classification only y_pred must be used.

Parameters

yarray-like: Target vector
y_predarray-like: Predicted label vector (num_examples,).
y_probamatrix-like: Predicted probability matrix (num_examples, num_classes). The probability estimates must sum to 1 across the possible classes and each matrix value must be in the interval [0,1].
group_aarray-like: Group membership vector (binary)
group_barray-like: Group membership vector (binary)

Returns

Self

fit_transform(y: ndarray, group_a: ndarray, group_b: ndarray, y_pred: ndarray | None = None, y_proba: ndarray | None = None)[source]#

Fit and transform

Parameters

yarray-like: Target vector
y_predarray-like: Predicted vector (nb_examlpes,)
y_probamatrix-like: Predicted probability matrix (num_examples, num_classes). The probability estimates must sum to 1 across the possible classes and each matrix value must be in the interval [0,1].
group_aarray-like: Group membership vector (binary)
group_barray-like: Group membership vector (binary)

Returns

dict: A dictionary of new predictions

transform(group_a: ndarray, group_b: ndarray, y_pred: ndarray | None = None, y_proba: ndarray | None = None)[source]#

Apply transform function to predictions and likelihoods

Parameters

y_predarray-like: Predicted vector (nb_examlpes,)
y_probamatrix-like: Predicted probability matrix (num_examples, num_classes). The probability estimates must sum to 1 across the possible classes and each matrix value must be in the interval [0,1].
group_aarray-like: Group membership vector (binary)
group_barray-like: Group membership vector (binary)

Returns

dict: A dictionary of new predictions