Discovering Hidden Earthquake Cycles Through Bayesian Mixture Models

This article demonstrates how Markov Chain Monte Carlo methods reveal that earthquake frequency isn't governed by a single statistical process, but alternates between distinct high and low-activity regimes. By moving beyond point estimates to full Bayesian posterior distributions, the approach provides rigorous uncertainty quantification for inferred seismic patterns. The practical implementation using Metropolis-Hastings sampling shows how Bayesian mixture models can uncover hidden structure in decades of real-world data.