Frequently Asked Questions

Can I use stream anomaly models for batch anomaly detection tasks?

The stream anomaly detection models can be used for batch anomaly detection tasks. Yet, the models are mainly designed for the tasks when a model sees an instance only once and removes from memory. Since models cannot always access all data, their performance may suffer. However, they are better fit to real-world problems when data arrives as time passes. You may also see PyOD library for batch learning models.

Can I use batch learning model for stream learning tasks?

Batch anomaly detection models from PyOD library can be used by reference windowing via pysad.models.reference_window_model or fitting to initial instances via pysad.models.one_fit_model. An example of integration is provided in Github.

How can I contribute to this project?

Please see Contributing for details.

I found a bug. What should I do?

You may contribute to this framework by sharing these bugs or feature ideas for improvements. You should open an issue on our Github repository. You are also welcome to contribute via opening a pull request. Please see Contributing for more information.