Welcome to Topological Data Analysis for Anomaly Detection’s documentation!

TDAAD

---

TDAAD – Topological Data Analysis for Anomaly Detection

Overview

TDAAD is a Python package for unsupervised anomaly detection in multivariate time series using Topological Data Analysis (TDA). Website and documentation: https://irt-systemx.github.io/tdaad/

It builds upon two powerful open-source libraries:

• GUDHI for efficient and scalable computation of persistent homology and topological features,

• scikit-learn for core machine learning utilities like Pipeline and objects like EllipticEnvelope.

TDAAD is inspired by the methodology introduced in:

Chazal, F., Levrard, C., & Royer, M. (2024). Topological Analysis for Detecting Anomalies (TADA) in dependent sequences: application to Time Series. Journal of Machine Learning Research, 25(365), 1–49. https://www.jmlr.org/papers/v25/24-0853.html

🔍 Features

• Unsupervised anomaly detection in multivariate time series

• Topological embedding using persistent homology

• Scikit-learn–style API (fit, transform, score_samples)

• Configurable embedding dimension, window size, and topological parameters

• Works with NumPy arrays or pandas DataFrames

Contents:

• 📖 Guidelines
  • 🚀 Quickstart
  • 📌 Usage Notes
• 🧪 Tutorial
  • Learning on a modified Ornstein–Uhlenbeck process
  • Topological embedding of a multiple time series
  • Anomaly Detection based on topological features
• 📚 Technical docs
• 💡 Theory overview
• 📑 References
• 🔄 Changelog
  • 1.6.0
  • 1.5.0
  • 1.4.0
  • 1.3.1
  • 1.3.0
  • 1.2.0
• 🔎 Tdaad package
  • tdaad package

Indices and tables

• Index

• Module Index

• Search Page

Contributors and Support

This work has been supported by the French government under the “France 2030” program, as part of the SystemX Technological Research Institute within the Confiance.ai project.

TDAAD is developed by IRT SystemX and supported by the European Trustworthy AI Association