Skip to content

tartaruszen/oddstream

Repository files navigation

Outlier Detection in Data STREAMs (oddstream)

Project Status: WIP � Initial development is in progress, but there has not yet been a stable, usable release suitable for the public. license


Last-changedate packageversion


Work in Progress: Do not use this package

This is a modified Python reimplementation of the R package oddstream. This implementation uses an LSTM Auto Encoder for feature extraction, instead of handcrafted features. Handcrafted time series feature extraction will be added in a near future.

Installation

python setup.py install

Usage

See example notebook.

References

  • Talagala, P. D., Hyndman, R. J., Smith-Miles, K., Kandanaarachchi, S., & Muñoz, M. A. (2018). Anomaly Detection in Streaming Nonstationary Temporal Data. Working paper 4/18, Department of Econometrics & Business Statistics, Monash University.

About

Outlier detection in Data STREAMs

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published