skclean.simulate_noise.BCNoise

class skclean.simulate_noise.BCNoise(classifier, noise_level, random_state=None)

Boundary Consistent Noise- instances closer to boundary more likely to be noisy. In this implementation, “closeness” to decision boundary of a sample is measured using entropy of it’s class probabilities. A classifier with support for well calibrated class probabilities (i.e. predict_proba of scikit-learn API) is required.

Only supports binary classification for now. See [MVRN18] for details.

Parameters
  • classifier (object) – A classifier instance supporting sklearn API.

  • noise_level (float) – percentage of labels to flip

  • random_state (int, default=None) – Set this value for reproducibility

Methods

__init__(classifier, noise_level[, random_state])

Initialize self.

fit_transform(X[, y])

Fit to data, then transform it.

get_params([deep])

Get parameters for this estimator.

set_params(**params)

Set the parameters of this estimator.

simulate_noise(X, y)

transform(X)