Bayesian Nonparametric Conditional Density Modeling in Causal Inference and Clustering with a Heavy-Tail Extension.
Current package version: see the Version field in DESCRIPTION.
CRAN package DOI: 10.32614/CRAN.package.CausalMixGPD.
# install.packages("remotes")
remotes::install_github(
"arnabaich96/CausalMixGPD",
build_vignettes = TRUE,
INSTALL_opts = c("--html")
)- Added support for lognormal
thresholdlink-distributions in GPD workflows, including spliced backend behavior. - Improved initialization for stability with covariate-aware threshold/link seeding and stronger latent label starts.
- Added CRP retry logic for rare all-
-Infinitialization failures during MCMC startup. - Updated wrappers/tests/manuscript examples to align with the new threshold-link and initialization behavior.
- Function reference
- CRAN package DOI:
10.32614/CRAN.package.CausalMixGPD
Use citation("CausalMixGPD") and report the exact package version used. Cite the CRAN package DOI for the R package, and cite the Aich-Bhattacharya methodology paper DOI for the statistical method.
- Performance acceptance tests and benchmark scripts are under
tests/perf/. - Main acceptance test entrypoint:
tests/testthat/test-performance-acceptance.R.
Use DPMIXGPD_TEST_LEVEL=ci for CI-level acceptance checks and
DPMIXGPD_TEST_LEVEL=full for full seeded-equivalence runs.