How to read these slides?

Navigate via arrow keys between slides Overview

Click on the menu bar items to navigate to chapters

Click here for PDF version
SCIENCE PASSION TECHNOLOGY

AustroVis AnoScout

A Time Series Anomaly Detection Sandbox

Julian Rakuschek

25.02.2025

Introduction

Anomalies in time series (Selection)

Anomaly = Unexpected Pattern

Check every time series by hand?

Let algorithms do the work!

Can we build a "sandbox" to find good ones?

Algorithm Types

Two types of anomaly extraction

Scoring
Classification

Two types of anomaly extraction

Scoring Anomaly score per timestamp
Output: Time Series Unsupervised algorithms
Classification

Two types of anomaly extraction

Scoring Anomaly score per timestamp
Output: Time Series Unsupervised algorithms
Classification Binary decision per segment
Users define normal behavior One Class Classification

Two types of anomaly extraction

Scoring: Users have no knowledge about the data

Classification: Users can define normal behavior and segments

Type is bound to bucket

Custom algorithm repertoire per bucket

AnoScout

The AnoScout Pipeline

Exploring Anomalies

Exploring anomalies is achieved through a linked view, users may further provide feedback on the importance of an anomaly to satisfy a specific information need.

Clustering Anomalies

The anomalies can be arranged by similarity in the scatterplot, such that similar anomalies are grouped together. This enables the user to discover patterns in the dataset.

Conclusion

Main features of AnoScout summarized

  1. "Playground" for testing various algorithms.
  2. Exploration pipeline for anomalies in time-oriented data.
  3. Using user labels to fine-tune the system.

From Thesis to VIS

Plan: Short Paper Track

The committee also welcomes papers describing new systems or tools that offer practical value.

Open Questions

  • Multivariate Time Series
  • Evaluation
  • Real World Use Cases

Questions?