amunozj/pyCFOFiSAX

CFOF approximation via iSAX trees - Python

CFOF score computation via iSAX with Anotations

iSAX implementation based on the PhD work of Lucas

Manual installation

Run pip install -r requirements.txt or python3 -m pip install -r requirements.txt.

The tslearn package requires Cython, numba and llvmlite (pip will automatically download these packages). However for Cython, it is necessary to have a C compiler (as described on this website), and the package llvmlite for this project, tested only with Python 3.7 and 3.8 that we recommend.

Otherwise, the use of conda simplifies the installation (as recommended on the site of tslearn), because it does not require the installation of C compiler and Cython and llvmlite packages.

Documentation

The documentation is available here (in French) : https://pycfofisax.readthedocs.io/fr/main/

Thanks

Development of :

• CFOF: A Concentration Free Measure for Anomaly Detection, par Fabrizio Angiulli,
• iSAX: Indexing and Mining Terabyte Sized Time Series, par Jin Shieh et Eamonn Keogh,
• iSAX 2.0: Indexing and Mining One Billion Time Series, par Alessandro Camerra, Themis Palpanas, Jin Shieh et Eamonn Keogh.
• Scoring Message Stream Anomalies in Railway Communication Systems, par Lucas Foulon, Serge Fenet, Christophe Rigotti et Denis Jouvin

Use of the code of:

• tslearn de Romain Tavenard et al. (2017).

Use of the packages of:

• Ahmad, S., Lavin, A., Purdy, S., & Agha, Z. (2017). Unsupervised real-time anomaly detection for streaming data. Neurocomputing, Available online 2 June 2017, ISSN 0925-2312