Uncovering Causal Relationships in Time Series Data: A Hybrid Approach Combining Constraint-Based and Noise-Based Algorithms

This paper introduces a novel hybrid framework that combines constraint-based and noise-based algorithms to uncover causal graphs from observational time series data. The framework consists of two classes of methods, NBCB and CBNB, which leverage the strengths of both approaches to provide robust causal inference under various assumption violations.