famish (version 0.2.0)

famish-package: famish: Flexibly Tune Families of Probability Distributions

Description

Fits probability distributions to data and plugs into the 'probaverse' suite of R packages so distribution objects are ready for further manipulation and evaluation. Supports methods such as maximum likelihood and L-moments, and provides diagnostics including empirical ranking and quantile score.

Overview

The famish package tunes probability distribution families so that they align with empirical data or expert-specified targets while remaining compatible with the probaverse ecosystem. With famish, you can:

1. Fit distribution families with a single interface that wraps established estimation routines.

2. Obtain results as distionary objects for downstream manipulation and comparison.

3. Access empirical ranking and quantile score as diagnostics.

The package acts as the estimation layer of probaverse, providing a bridge from distribution specifications to calibrated models.

Fitting Distribution Families

Use fit_dst() to fit any supported family by specifying the distribution family name and estimation method. Convenience wrappers such as fit_dst_gev(), fit_dst_gp(), and other fit_dst_*() functions expose documented combinations of families and fitting approaches. Supported methods include maximum likelihood, maximum goodness-of-fit method, method of moments, and L-moments (including log-scale variants for selected families). For references to these different methods, see the fitdistrplus package, or the lmom package for estimation by L-moments.

Getting Started

New users can begin with:

• the README for an overview of package goals and examples,

• vignette("fitting", package = "famish") for a tutorial on fitting workflows.

Author

Maintainer: Vincenzo Coia [copyright holder]

See Also

Useful links:

• https://famish.probaverse.com/

• https://github.com/probaverse/famish/

Examples

library(distionary)

set.seed(2024)
sample_data <- rgamma(100, shape = 2, rate = 0.4)

fitted_gamma <- fit_dst("gamma", x = sample_data, method = "mle")
fitted_gamma

parameters(fitted_gamma)
enframe_return(fitted_gamma, at = c(2, 5, 10))