Most data analysis stops at correlation. My research recovers the actual causal structure and governing relationships behind data, so decisions rest on what truly drives a system rather than on coincidence. The core method, entropic regression, overcomes the outlier and noise problems that defeat conventional system identification, recovering sparse and accurate models of nonlinear dynamics directly from observations. Related work uses the optimal causation entropy principle to learn networks and Boolean functions, and applies these ideas to real problems ranging from neurological data to North American monsoon prediction. The result is a rigorous, information-theoretic toolkit for root-cause analysis, true-driver identification, and decision intelligence under uncertainty.
AFRL CRADA: Geometric Information Theory and Causation Inference (2023–2025). NSF EAGER (Co-PI): North American Monsoon Prediction Using Causality-Informed ML (2023–2024).