1 gamlss2: Infrastructure for Flexible Distributional Regression

1.1 Overview

The gamlss2 package provides infrastructure for generalized additive models for location, scale, and shape (GAMLSS) and distributional regression more generally. It reimplements the classic gamlss framework with an emphasis on estimation speed and flexible model components. Its modular design supports alternative estimation algorithms, Bayesian sampling, and model terms such as regression trees and forests.

Models can use smooth terms from mgcv, custom special terms, and response families with more than four parameters. Separate predictors for the distribution parameters are specified in a single formula.

• The main model function is gamlss2().
• The default optimizer is RS(); alternative optimizers can be supplied.
• The main methods are summary(), plot(), and predict().
• Custom family objects are described in Family Objects.
• Custom predictor components are described in Special Model Terms.

Start with First Steps for a worked example, or browse the manual pages.

help(package = "gamlss2")

1.2 Installation

The development version of gamlss2 can be installed via

install.packages("gamlss2",
  repos = c("https://gamlss-dev.R-universe.dev",
            "https://cloud.R-project.org"))

1.3 License

The package is available under the General Public License version 3 or version 2