By Harikrishnan Nb, BITS Pilani, Goa Campus

Abstract 

Discovering causal direction from temporal observational data is particularly challenging for symbolic sequences, where functional models and noise assumptions are often unavailable. We propose Dictionary Based Pattern Entropy (DPE), a novel framework that infers both the direction of causation and the specific subpatterns driving changes in the effect variable. DPE integrates Algorithmic Information Theory and Shannon Information Theory, interpreting causation as the emergence of compact, rule based patterns in the candidate cause that systematically constrain the effect. The framework constructs direction specific dictionaries and quantifies their influence using Shannon entropy, enabling a principled link between deterministic pattern structure and stochastic variability. Causal direction is inferred via a minimum uncertainty criterion, selecting the direction exhibiting stronger and more consistent pattern driven influence. We demonstrate the efficacy of DPE on various simulated datasets, highlighting its interpretability advantages over existing methods.