Topological Data Analysis for Anomaly Detection (TDAAD)

tdaad provides machine learning algorithms for analyzing timeseries data through the lense of Topological Data Analysis, and deriving anomaly scores.

The targeted input is an object X representing a multiple time series with variables columns and timestamps lines. We use the term multiple time series to describe a set of univariate timeseries that describe a system or object. Note that the package does not handle analysis of a single univariate timeseries.

The main idea of this package is to analyze time series with topological methods.

Documentation

User Manual

Documentation is available online.

Methodological Guidelines

1. Methodological Guideline for Anomaly Detection Models
2. Methodological Guideline for Time Series Anomaly Detection

Benchmarks

1. Anomaly Detection Models Use Case level document
2. Unsupervised Anomaly Detection and Explainability tools for Time Series

Demonstrators

• A pubic dataset based demonstrator is available here. It uses the Server Machine Dataset from OmniAnomaly
• A restricted demonstrator is available here. It relies on demand forecasting and anomaly detection over time-series use-cases.

Support

Support for TDAAD must be obtained by sending an email to support@confiance.ai

Ensure your email contains :

• Your name
• A link to this page
• the version you are working with
• A clear description of the problematic (bug, crash, feature or help request)
• A full description of the problem whichallow to reproduce it
• Any file or screenshort element mandatory for the full understanding of the problematic