Beta Regression: R Package for Modeling Rates and Proportions

An R package for modeling beta-distributed dependent variables on the unit interval (0, 1), such as rates and proportions. The package implements the classical beta regression model from Cribari-Neto and Zeileis (2010) and extended-support models for boundary observations from Kosmidis and Zeileis (2025). It is authored by Achim Zeileis and includes alternative specifications like bias-corrected estimation and finite mixture models.

Use Cases

• Modeling rates and proportions based on the described beta-distributed dependent variables.
• Handling boundary observations at 0 and/or 1 using the extended-support beta regression models.
• Applying bias-corrected or bias-reduced estimation for improved inference in beta regression.
• Fitting finite mixture models for heterogeneous beta-distributed data.
• Using recursive partitioning to discover subgroups in beta regression analyses.

Strengths

• Implements the established classical beta regression model from a 2010 Journal of Statistical Software paper.
• Includes extended-support models for boundary observations, based on a 2025 Royal Statistical Society paper.
• Offers multiple alternative specifications, including bias-corrected estimation and finite mixture models.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

paperswithcode