Primary Submission Category: Causal Discovery
Integrating Time Series Analysis and Causal Discovery for Enhanced Sequential Data Understanding
Authors: Suat Babayigit,
Presenting Author: Suat Babayigit*
Understanding sequential data and uncovering causal relationships are critical challenges in modern data analysis. This study integrates time series analysis with causal discovery methods to address limitations such as non-stationarity, irregular sampling, and multivariate causality in existing approaches. A novel method combining Fourier analysis and Granger causality is proposed, enhancing the strengths of each technique while mitigating their weaknesses. The method was evaluated using real-world datasets across domains, yielding robust performance in identifying causal structures within complex time series data. Results highlight the potential for improved accuracy and interpretability in sequential data modeling. This work contributes to advancing the fields of time series analysis and causal inference, offering practical insights for applications in healthcare, finance, and environmental science.