Visibility graph-based segmentation of multivariate time series data and its application.

Area of Science:

• Data Science
• Time Series Analysis
• Network Science

Background:

• Multivariate time series analysis presents challenges due to high dimensionality.
• Existing segmentation methods may suffer from the curse of dimensionality and computational inefficiency.

Purpose of the Study:

• To develop an efficient and accurate method for segmenting multivariate time series.
• To overcome the curse of dimensionality in time series data.
• To provide a robust segmentation approach applicable to both synthetic and real-world data.

Main Methods:

• Dimensionality reduction using Principal Component Analysis (PCA).
• Network construction from time series data via Visibility Graph theory.
• Community detection algorithm with modularity optimization for segmentation.

Main Results:

• The proposed method effectively segments multivariate time series into distinct stages.
• Achieved high accuracy in segmentation compared to state-of-the-art models.
• Demonstrated lower time complexity (O(n^3)) with superior segmentation performance.

Conclusions:

• The integrated approach of PCA, visibility graphs, and community detection offers an efficient and accurate solution for multivariate time series segmentation.
• The method proves effective on both generated data and real-world oil futures data.
• This technique addresses the curse of dimensionality and enhances segmentation quality.