RegimeChange 0.1.0

Initial Release

Detection Methods

• Added PELT (Pruned Exact Linear Time) for optimal changepoint detection
• Added CUSUM for cumulative sum based detection
• Added Binary Segmentation for greedy recursive detection
• Added Wild Binary Segmentation for robust detection
• Added BOCPD (Bayesian Online Changepoint Detection)
• Added Shiryaev-Roberts procedure for sequential detection
• Added ensemble method combining multiple algorithms

Prior Specifications

• normal_gamma() for unknown mean and variance
• normal_known_var() for known variance models
• normal_wishart() for multivariate data
• poisson_gamma() for count data
• geometric_hazard() for constant hazard rate

Features

• Unified detect_regimes() interface for all methods
• Online detection with regime_detector() class
• Uncertainty quantification via bootstrap and Bayesian posteriors
• Multivariate changepoint detection
• Comprehensive evaluation metrics
• ggplot2-based visualization suite
• Method comparison framework

Documentation

• Introduction vignette
• Offline detection vignette
• Bayesian methods vignette

mirror server hosted at Truenetwork, Russian Federation.