NEWS

---

fracreg 1.1.0 (2026-08-20)

Output Standardization & Aesthetics

• Hierarchical Model Displays: Refined print and summary outputs for partial effects (fracreg.pe, fracregpd.pe, fracreghet.pe, fracregridge.pe, fracregmlogit.pe, fracregmlogit.wtp). Enforced a logical hierarchical model ordering (BIN0 -> BIN1 -> FRAC -> OVERALL) and adopted clear mathematical notation for component probabilities (e.g., Pr(y > 0)).
• Matrix Transposition: Restructured fracregmlogit.pe outputs to universally present discrete choices as columns and covariates as rows, standardizing the multidimensional matrix display.
• Specification Tests Standardization: Outputs for econometric specification tests (fracreg.reset, fracreghet.reset, fracreg.ggoff, fracreg.ptest) now display dynamic headers indicating precisely which parent model was tested (e.g. "RESET Test for Fractional logit regression") rather than static generic text.
• Willingness to Pay (WTP): Updated wtp() and its plotting functions (plot.fracregmlogit, plot.fracregmlogit.pe) to natively support the newly transposed fracregmlogit.pe outputs. Added stylistic formatting headers to wtp console prints.
• Native Print vs Summary: Restored native print() S3 methods across the package to output raw point-estimate matrices while reserving detailed statistical tables (with standard errors, p-values, and confidence intervals) for summary() S3 methods.

Tidyverse & gtsummary Integration

• Comprehensive Broom Support: Exported broom::tidy() S3 methods for all package models, partial effects objects (.pe), willingness-to-pay objects (.wtp), and econometric tests (.reset, .ggoff, .ptest). Tidied tibbles correctly map test outputs to term, statistic, and p.value columns.
• gtsummary Dispatch Resolution: Re-engineered the gtsummary::tbl_regression() integration. tbl_regression.fracreg() now correctly triggers S3 dispatch via NextMethod() without inadvertently stripping class attributes, preventing broom::tidy() fallback failures and resolving R CMD check warnings related to unexported gtsummary::: namespaces.
• Intercept Formatting: The gtsummary integration now correctly lists (Intercept) terms by default for non-mixture (1P) and mixture (2P) fractional models.
• Prefix Cleaning: Fixed an issue where broom::tidy retained frac_ internal prefixes for fracregridge components.

Bug Fixes & Code Quality Improvements

• Krinsky-Robb Standard Errors: Fixed a critical bug in fracregmlogit.pe where standard errors simulated via marg.type = "aveacr" calculated point estimates across a single sample mean instead of averaging across all individual observations. Average Partial Effect standard errors are now accurate across the full sample. Changed se = TRUE by default for fracregmlogit.pe.
• R CMD Check Compliance: Replaced all instances of T and F with TRUE and FALSE across R source files and roxygen2 documentation blocks. Added Depends: R (>= 3.5.0) to DESCRIPTION to comply with serialized dataset formats.
• Documentation Tags: Systematically added missing \value tags to exported R methods (like fitted, predict, and residuals) and removed redundant introductory phrases like "This package..." from description fields to meet strict CRAN policies.
• Example Scripts & Variances: Fixed simulation bugs in fracregpd.pe example scripts where random error terms possessed static variances, restoring R CMD check validation.
• GitHub Actions: Resolved 403 Forbidden errors during pkgdown automated documentation deployments by explicitly granting the contents: write permission to the GitHub Actions workflow.

fracreg 1.0.1 (2026-08-05)

New Features & Enhancements

• Fractional Ridge Regression: Added fracregridge and fracregridge.pe to implement fractional ridge regression and its corresponding partial effects.
• Fractional Multinomial Logit & Krinsky-Robb: Refactored fracregmlogit.pe to use MASS::mvrnorm for Krinsky-Robb standard error simulations, resolving hardcoded constraints.
• Panel Data Partial Effects: Implemented fracregpd.pe for computing Average Partial Effects (APE) in panel data settings, including full support for Correlated Random Effects (CRE) models and Delta method standard errors.
• Odds Ratios: Added support for computing Odds Ratios and adjustable confidence intervals (level) in output summaries.
• Offset Parameter: Added support for an offset parameter in model specifications.
• S3 Methods & Output Display: Standardised console output by adding dedicated summary() and print() S3 methods across the package to align with R's standard practices. Changed output tables to say "regression" rather than "model" for clarity.
• Econometric Terminology & Standardization: Renamed Estimate column headers to dy/dx universally across all partial effects tables (fracreg.pe, fracreghet.pe, fracregmlogit.pe, fracregpd.pe, fracregridge.pe), while enforcing Coefficient for non-partial effects outputs.
• Standard Error Annotations: Output tables now dynamically label column headers according to standard error type (e.g. Robust Std.Err., Cluster Std.Err.) and cleanly append specific methodological notes to the tables (e.g., Delta Method, Krinsky-Robb, HC0, CRVE).

Documentation & Website

• Roxygen2 Migration: Fully automated package documentation by migrating all manual .Rd files into roxygen2 comments embedded directly within the R source files, ensuring airtight synchronization between code and documentation.
• Citations & References: Added citations for the foundational fmlogit (Ji and Woodill) and fracridge (Rokem and Kay) packages across manual pages, README, and vignettes, formatting URLs properly for rendering.
• Logo & Badges: Improved the package hex logo and added status badges to the README.
• Website (pkgdown): Added a sitemap.xml and reordered the README and vignettes sections to properly showcase Fractional Multinomial Logit and Fractional Ridge Regression alongside Panel Data models. Added Willingness to Pay (WTP) plot visualizations in tutorials.
• Help Files: Updated .Rd files to properly reflect table = FALSE default arguments and fixed \usage widths. Updated the package title for clarity and consistency.

Bug Fixes & Under-the-Hood Improvements

• Namespace Imports: Added comprehensive @importFrom tags for base R stats, graphics, and grid functions, alongside ggplot2, eliminating all undefined global function NOTEs during checking.
• S3 Method Registration: Explicitly registered S3 methods (like summary and plot) using @exportS3Method to comply with strict R CMD check standards.
• CRAN Compliance: Wrapped computationally heavy examples (like plot.fracregmlogit.pe) in \donttest{} blocks to strictly respect CRAN's 5-second execution limit.
• Resolved length > 1 logical condition errors in plot.fracregmlogit standard error rendering by properly adopting match.arg().
• Resolved WTP matrix subsetting issues by enforcing drop = FALSE.
• Cleaned up legacy fmlogit code dependencies and unified the namespace fully under fracregmlogit.
• Fixed NA/NaN evaluation errors in nlminb during fracreghet estimations.
• Resolved devtools::check() formatting issues, including S3 class inheritance checks (inherits()), stats::approx scoping, and fixing missing alt-text for accessibility.

fracreg 1.0.0

• Initial CRAN release.