Added a pseudo-code vignette, refreshed the README with workflow details, and expanded the unit test suite for the new helpers.
Fixed code and descriptions to get rid of notes during CRAN checks.
Enabled optional parallel resampling in sb_beta()/sb_resample_groups() via
future.apply, added S3 print/summary/autoplot helpers for sb_beta
results, and documented the new behaviour in the README.
Extended the stepwise beta selectors to handle observation weights and precision-submodel search, exposing precision coefficients in the returned paths.
Added reproducible resampling caches and quality diagnostics to
sb_resample_groups()/sb_beta(), including interval-response support that
reuses pseudo-responses across correlation thresholds.
Documented interval workflows more prominently by adding sb_beta_interval(),
expanding the README/CRAN vignette guidance for selector choice and interval
stability, and clarifying comparison-helper outputs and response squeezing.
sb_beta() to run the full SelectBoost correlated-resampling loop with
beta-regression selectors, plus a vignette illustrating the workflow.simulation_DATA() to generate interval-valued Beta-regression data:
interval = "jitter" (symmetric) or "quantile" (Beta quantile intervals).fastboost_interval(); added a small vignette and unit test.delta_low/delta_high),
asymmetric quantile coverage (alpha_low/alpha_high), covariate-driven parameters (accept functions of (mu, X)),
and optional missing bounds per row (na_rate, na_side).compare_selectors_single(), compare_selectors_bootstrap() to run all selectors (AIC/BIC/AICc, GAMLSS LASSO/ENet*, GLMNET) and compute selection frequencies.plot_compare_coeff(), plot_compare_freq() heatmaps to compare selectors side by side.fastboost_interval() (interval response stability selection), C++ IRLS speedups, and prestandardize option for betareg_glmnet().
gamlss.lasso if installed.betareg.gamlss::ri) and optional Elastic-Net (gamlss.lasso::gnet).fastboost_interval() prototype for interval responses.