This repository contains the implementation of the kNN_BPP algorithm from the paper Handling Class Imbalance in k-Nearest Neighbor Classification by Balancing Prior Probabilities by Jonatan M. N. Gøttcke and A. Zimek. The paper is presented at the 14th International Conference on Similarity Search and Applications, SISAP 2021.

Furthermore the repository contains the implementations of the DIRECT CS kNN presented in the 2013 paper Cost-Sensitive Classification with k-Nearest Neighbors and the 2020 paper Cost-sensitive KNN classification by Qin. et al. and Zhang et al.

And the CW-kNN algorithm from the Class Based Weighted K-Nearest Neighbor over Imbalance Dataset by V. Pudi and H. Dubey is implemented.

The implementations follow the Scikit-learn classifier style which most users are familiar with.

1. Git clone the repository.
2. Ensure you have Python 3.8.+ installed
3. Install the required dependencies

   pip install -r requirements.txt

4. Try the example

   python example.py

See the example.py file on how to use the implementation in further detail.

@inproceedings{DBLP:conf/sisap/GottckeZ21,
  author    = {Jonatan M{\o}ller Nuutinen G{\o}ttcke and
               Arthur Zimek},
  title     = {Handling Class Imbalance in k-Nearest Neighbor Classification by Balancing
               Prior Probabilities},
  booktitle = {{SISAP}},
  series    = {Lecture Notes in Computer Science},
  volume    = {13058},
  pages     = {247--261},
  publisher = {Springer},
  year      = {2021}
}

DOI: 10.1007/978-3-030-89657-7_19