/Ensemble-Outlier-Analysis

Library for outlier analysis using ensemble algorithms

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

Outlier Ensemble Analysis

Summary

This repository contain the implementation of 5 ensemble models: HICS, OUTRES, LODA, Mahalanobis Kernel and Trinity. These implementations are used to make some experiments for my bachelor degree work. The theoretical description of the models are taken from the papers and books:

  1. https://ieeexplore.ieee.org/document/6228154
  2. https://link.springer.com/article/10.1007/s10994-015-5521-0
  3. https://ieeexplore.ieee.org/document/5767916
  4. https://www.springer.com/gp/book/9783319547640

The whole explanation in Spanish is in the repository: https://github.com/nacheteam/TFG

Structure

The structure of the repository is the following one:

  • Datasets: this is the folder with the datasets from ODDS.
  • Examples: this folder contains one example of use for each of the models.
  • Experiments: this folder contains the two experiments performed on my work as well as the results and the script to generate the figures and results.
  • Models: this folder contains the actual implementation of the 5 ensemble models

Datasets picked from:

Shebuti Rayana (2016). ODDS Library [http://odds.cs.stonybrook.edu]. Stony Brook, NY: Stony Brook University, Department of Computer Science.