TiVy: Time Series Visual Summary for Scalable Visualization.

Area of Science:

• Data Visualization
• Time Series Analysis
• Pattern Recognition

Background:

• Visualizing multiple time series is crucial for understanding large-scale processes but faces scalability and clarity challenges.
• Existing methods often result in visual clutter due to numerous small multiples or overlapping lines, especially over long time spans.

Purpose of the Study:

• To introduce TiVy, a new algorithm for summarizing time series data through sequential pattern extraction.
• To develop an interactive visualization tool for real-time rendering of large-scale time series.
• To address the scalability and visual clutter issues in time series visualization.

Main Methods:

• TiVy transforms time series into symbolic sequences based on visual similarity using Dynamic Time Warping (DTW).
• It groups similar subsequences (of varying lengths) aligned in time, based on frequent sequential patterns.
• An interactive visualization system is presented for real-time rendering.

Main Results:

• The TiVy algorithm effectively extracts clear and accurate patterns from time series data.
• It achieves a significant speed-up compared to straightforward DTW clustering.
• Demonstrates efficiency in exploring hidden structures within massive time series datasets.

Conclusions:

• TiVy provides an uncluttered visual summary of time series, improving superposition and reducing the need for excessive small multiples.
• The algorithm offers a scalable and efficient solution for analyzing large-scale time series data.
• TiVy facilitates the discovery of hidden patterns and structures in complex time series.