Package

Start here

DRModels is the package module. For fitting choices, begin with What can I fit today?; for runnable syntax, use Getting started. This page documents the module overview rather than making a package-wide performance claim.

DRModels Module

DRModels

DRModels.jl fits distributional regression models for continuous, count, proportion, and ordered responses. It provides univariate routes and selected bivariate routes. Random effects and dependence from phylogenies, spatial coordinates, pedigrees, and supplied matrices are available on the combinations listed in the capability guide. The R package drmTMB can call selected routes through its optional Julia engine; DRModels.jl can also be used directly without R.

The q=4 phylogenetic bivariate location-scale model has random effects on two mean and two log-scale axes. Its engine uses an augmented-state Laplace approximation with an exact sparse gradient. Sparse factorisation cost depends on the model structure; this is not a package-wide linear-time guarantee.

Maximum likelihood is the default. Opt-in method = :REML is available on documented model-specific routes. The Gaussian phylogenetic-mean solver also supports algorithm = :em. Wald, profile, and parametric-bootstrap inference have route-specific restrictions; a computed interval does not establish its coverage.

See the model guides and capability matrix for runnable syntax, accepted combinations, and evidence limits. Measured comparisons with drmTMB retain their run conditions in report/comparison-grid.md; those results apply to the models and data measured, rather than every package workflow.

source