OCPDet is an open-source Python package for online changepoint detection, implementing state-of-the-art algorithms and a novel approach, using a scikit-learn style API.

This package is the outcome of my Master Thesis at Imperial College London within the MSc in Statistics, Department of Mathematics.

Algorithms implemented in ocpdet are

• CUSUM: Cumulative Sum algorithm, proposed by Page (1954)
• EWMA: Exponentially Weighted Moving Average algorithm, proposed by Roberts (1959)
• Two Sample tests: Nonparametric hypothesis testing for changepoint detection, proposed by Ross et al. (2011)
• Neural Networks: Novel approach based on sequentially learning neural networks, proposed by Hushchyn et al. (2020) and extended to online context (Master Thesis)

pip install ocpdet

• CUSUM.ipynb
• EWMA.ipynb
• TwoSample.ipynb
• NeuralNetwork.ipynb

Here is a suggestion to cite this GitHub repository:

Victor Khamesi. (2022). ocpdet: A Python package for online changepoint detection in univariate and multivariate data. (Version v0.0.5). Zenodo. https://doi.org/10.5281/zenodo.7632721

And a possible BibTeX entry:

@software{victor_khamesi_2022,
  author       = {Victor Khamesi},
  title        = {ocpdet: A Python package for online changepoint detection in univariate and multivariate data.},
  month        = oct,
  year         = 2022,
  publisher    = {Zenodo},
  version      = {v0.0.5},
  doi          = {10.5281/zenodo.7632721},
  url          = {https://doi.org/10.5281/zenodo.7632721}
}

The non-software content of this project is licensed under a Creative Commons Attribution 4.0 International License, and the software code is licensed under the BSD-2 Clause license.