Outlier detection by analysis of cluster transitions

Outlier detection method based on CLOSE.

Introduction

This repository contains the implementation of an outlier detection method based on maximum stable clustering per timestamp (CLOSE).

First create a virtual environment and run following inside it:

pip install .

After required libraries are installed go to experiments/show_me_ur_friends. This folder contains all experiments for the comparison of our method with https://link.springer.com/chapter/10.1007/978-981-15-1699-3_8.

In each folder to run experiment with our method, enter:

python sigma_outlier.py

To run an experiment with the method of Tatusch et al. enter:

python tau_outlier.py

After you executed one of the scripts, the corresponding result will appear in the results-folder.