SimiPixel / automatic_label_correction

Automatic Label Correction

Installation

pip install git+https://github.com/SimiPixel/automatic_label_correction.git

Usage

from sklearn.datasets import load_iris
X, y = load_iris(return_X_y=True)

from ALC import NearestNeighbourCorrection as NNC  
nnc = NNC()

y_corrected = nnc.fit_transform(X, y)

Artifically falsify labels of Iris dataset. Correction factor represents how many false labels are corrected

• Correction factor = 1: All labels are corrected
• Correction factor = 0: Same number of incorrect labels than before applying any correction alg.

API

• ALC.NearestNeighbourCorrection

• ALC.ClusterCorrection

• ALC.AutomaticDataEnhancement

• ALC.BinaryClusterCorrection

• ALC.utils

  • ALC.utils.falsify: Artifically falsify labels
  • ALC.utils.kfold
  • ALC.utils.OneHot
  • ALC.utils.convert_labels: Convert labels into different representation

• ALC.evaluate

  • ALC.evaluate.correction_factor
  • ALC.evaluate.accuracy