$ show easystats/parameters
This is an R package (`parameters`) from the easystats ecosystem that provides a unified interface for extracting, formatting, and comparing parameters from a vast array of statistical models (lm, glmmTMB, brms, lme4, lavaan, fixest, etc.). The package functions as a model-parameter reporting engine: it takes a fitted model object and returns a tidy data frame with coefficients, standard errors, confidence intervals, p-values, and effect sizes, supporting various inference methods (Wald, profile, bootstrap, Kenward-Roger, Satterthwaite). Beyond basic extraction, it offers standardization, bootstrapping, equivalence testing, p-value functions, dominance analysis, and data reduction (PCA, EFA, clustering). The architecture relies heavily on S3 generic dispatch (e.g., `model_parameters.lm`, `model_parameters.glmmTMB`) and integrates with the easystats stack (`insight` for model introspection, `bayestestR` for Bayesian indices, `datawizard` for data manipulation). Output can be formatted as plain text, markdown, or HTML, making it suitable for both interactive use and automated reporting. The package is actively maintained with extensive test coverage and continuous integration.
The LLM identified 3 files worth highlighting for reuse:
Implements the core confidence interval computation logic used by the `ci()` generic. The internal function `.ci_generic()` supports Wald, profile-likelihood, bootstrap, and various degrees-of-freedom corrections (Kenward-Roger, Satterthwaite). It takes a parameter estimates data frame and returns a tidy CI table. This component is well-parameterized, separated from model-specific extraction, and could be reused in other R packages needing robust CI calculation.
Provides equivalence testing for frequentist models using the ROPE (Region of Practical Equivalence) approach, including classic TOST and the Conditional Equivalence Test (CET). The internal `.equivalence_test_numeric()` function implements the decision logic in a self-contained manner, relying only on base R statistics. This can be extracted and applied to any numeric estimate with accompanying confidence intervals.
Implements p-value functions (also called consonance or confidence curves), which show the relationship between p-values and parameter values. This is a statistically advanced technique not commonly available in R packages. The code is self-contained (uses only base R distribution functions) and can be reused to generate p-value functions for any model with Wald-type confidence intervals.