Generalized Isolation Forest

The Generalized Isolation Forest (GIF) constitutes a novel outlier detection algorithm, which in contrast to “classical” Isolation-based approaches directly aims at estimating the underlying probability distribution to identify non-conforming observations. The algorithm has originally been proposed by Buschjäger, Honysz and Morik within this research paper:

Buschjäger, S., Honysz, PJ. & Morik, K. Randomized outlier detection with trees. International Journal of Data Science and Analytics (2020). https://doi.org/10.1007/s41060-020-00238-w

This package provides an implementation of the GIF algorithm, which is accessible from Python and C++.

Install from PyPI (recommended)

We provide Linux wheel packages for various Python versions, which can be installed like this:

pip install genif

Windows or macOS builds are currently not provided. Please resort to installation from source, if you are either using Windows or macOS.

Install Python package from source

The following requirements have to be met in order to build this package from source:

• GCC >= 5.4.0 (older versions or other compilers such as Clang or ICC may work, but have not been tested yet.)

• CMake >= 3.5.1

• OpenMP

Building is then conducted by following these steps:

1. Recursively clone this repository by issueing git clone --recurse-submodules git@github.com:philippjh/genif.git

2. Change your working directory to the root of the repository. Run pip3 install .

3. The Python package manager will now build and install the package.

Acknowledgments

Part of the work on this paper has been supported by Deutsche Forschungsgemeinschaft (DFG) within the Collaborative Research Center SFB 876 “Providing Information by Resource-Constrained Analysis”, project A1, http://sfb876.tu-dortmund.de and by the German Competence Center for Machine Learning Rhine Ruhr (ML2R, https://www.ml2r.de, 01IS18038A), funded by the German Federal Ministry for Education and Research.