SMOTE for compositional data

SMOTE-CD implements an adaptation of the Synthetic Minority Oversampling TEchnique in the case where the labels of the dataset are compositional data.

You can install and start using SMOTE-CD using pip

python -m pip install smote-cd

Documentation

You can find a detailed documentation here:

• Install
  • User installation
  • Dependencies
• User Guide
  • Introduction
    • Compositional data
    • Imbalance
  • Dataset generation
  • Oversampling
  • Undersampling
• API reference
  • SMOTE for compositional data (smote_cd)
    • smote_cd.oversampling_multioutput
    • smote_cd.random_undersampling
  • Data generation (smote_cd.dataset_generation)
    • smote_cd.dataset_generation.generate_betas
    • smote_cd.dataset_generation.generate_dataset

Acknowledgements

This project was financed by the BigCEES project (E2S/UPPA) and the Macquarie University.