Skip to contents

Changelog

Source: NEWS.md

hurdlebb 0.1.0

  • Initial release.
  • Two-part hurdle beta-binomial model via CmdStan.
  • Four model variants: base, weighted, SVC, SVC + weighted.
  • Survey-weighted pseudo-posterior inference.
  • Sandwich variance estimation and Cholesky posterior recalibration.
  • Average marginal effects with extensive/intensive margin decomposition.
  • LOO-CV model comparison and posterior predictive checks.
  • Three built-in synthetic datasets based on NSECE 2019.
  • Five vignettes covering full analysis workflow.