brms: Bayesian Regression Models Using Stan

brms is an R package for fitting Bayesian generalized (non-)linear multivariate multilevel models using Stan for full Bayesian inference. It supports a wide range of distributions and link functions for models including linear, robust linear, count data, survival, ordinal, and zero-inflated models. The package was authored by Paul‐Christian Bürkner, with key references published in 2017, 2018, and 2021.

Use Cases

• Fit multilevel count data models based on the supported distributions and link functions.
• Perform distributional regression by predicting all parameters of the response distribution.
• Evaluate and compare models using methods assessing posterior or prior predictions.
• Fit survival or response time models in a Bayesian framework.
• Implement models with auto-correlation structures or censoring and truncation.

Strengths

• Supports a wide range of model types including linear, count data, survival, ordinal, and zero-inflated.
• Enables distributional regression by allowing prediction of all response distribution parameters.
• Provides flexible prior specifications to reflect user knowledge.
• Offers multiple methods for model evaluation and comparison.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• License is closed, restricting redistribution and modification.

Provenance

Source

Paul‐Christian Bürkner

Collection Method

Software package for statistical modeling.

Time Range

References span from 2017 to 2021.

Freshness

Last update date is unknown; freshness unverified.

Geography

Spatial coverage is not specified.