Two-Stage Bayesian Segmentation with AR(1) Errors

This repository contains the Python implementation of the **Two-Stage Bayesian Segmentation Algorithm** developed for detecting structural changes in time series with high serial correlation.

The method addresses the problem of **Weak Identification** and **Likelihood Ridges** (Gustafson, 2010; Florens & Simoni, 2021) inherent in the simultaneous estimation of change-points and autoregressive parameters. By adopting a **Sequential Identification Strategy**, the algorithm decouples structure learning from noise estimation, ensuring asymptotic consistency and valid Bayesian learning.