Multiresolution granger causality testing with variational mode decomposition: a python software

Area of Science:

• Time Series Analysis
• Causality Inference
• Signal Processing

Background:

• Traditional Granger causality tests can overlook complex interactions in aggregated time series data.
• Analyzing causality across multiple frequency scales is crucial for understanding intricate system dynamics.

Purpose of the Study:

• To develop an advanced multiscale approach for Granger causality testing.
• To enhance the precision and granularity of causal relationship analysis.
• To provide a robust and accessible tool for complex data analysis.

Main Methods:

• Integration of Variational Mode Decomposition (VMD) with traditional Granger causality testing.
• Decomposition of time series into intrinsic mode functions (IMFs) representing distinct frequency scales.
• Application of Granger causality tests to stationary IMFs for detailed causal pattern identification.

Main Results:

• The multiscale approach effectively uncovers causal patterns hidden in aggregated data.
• Empirical studies on cryptocurrency, biomedical signals, and simulations validate the method's effectiveness.
• The VMD-integrated Granger causality demonstrates superior accuracy and precision compared to existing techniques.

Conclusions:

• The novel VMD-based multiscale Granger causality testing offers a flexible and robust framework for causal analysis.
• This methodology significantly improves the ability to reveal hidden causal interactions in complex systems.
• The user-friendly Python package enhances accessibility for researchers and practitioners across various scientific domains.