Changes in version 0.4.5 (2025-11-11)                  

  - 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.

                        Changes in version 0.4.4                        

  - Add NEWS.md documenting development history.

                        Changes in version 0.4.3                        

  - Added sb_beta() to run the full SelectBoost correlated-resampling
    loop with beta-regression selectors, plus a vignette illustrating
    the workflow.
  - Added vignette section demonstrating the extended simulator and
    interval selection.

                        Changes in version 0.4.2                        

  - New simulation_DATA() to generate interval-valued Beta-regression
    data:
      - interval = "jitter" (symmetric) or "quantile" (Beta quantile
        intervals).
      - Works with fastboost_interval(); added a small vignette and unit
        test.
      - Supports mixed mechanisms (row-wise jitter vs quantile),
        asymmetric jitter widths (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).

                        Changes in version 0.4.1                        

  - Comparison helpers and visualizations:
      - 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.
      - Vignette expanded to include simulated data and two real-ish
        datasets after scaling to (0,1).
  - Kept fastboost_interval() (interval response stability selection),
    C++ IRLS speedups, and prestandardize option for betareg_glmnet().
      - ENet requires gamlss.lasso if installed.

                        Changes in version 0.4.0                        

  - Initial beta-regression integration for SelectBoost:
      - Stepwise AIC, BIC, and AICc wrappers around betareg.
      - GAMLSS LASSO (gamlss::ri) and optional Elastic-Net
        (gamlss.lasso::gnet).
      - Pure glmnet (IRLS + weighted Gaussian ENet) with optional
        prestandardize.
      - fastboost_interval() prototype for interval responses.
      - First vignette + roxygenized docs.