Articles
Get started
Start here: a five-minute quickstart, then the ladder that builds up from lm() to location-scale on one shared dataset.
- Get started with symbolizer
A five-minute first contact: fit a model, then read its equation, assumptions, and plain-English interpretation.
- Building up: from lm to location-scale
One synthetic dataset, four models of increasing complexity, the same
symbolize()call. Watch the symbolic equation grow one row at a time as you add a factor, then a random effect, then a heterogeneity submodel.
Deep dives
One renderer or concept at a time: the drmTMB extractor, factor contrasts and interactions, non-Gaussian families, variance components, latent variables (gllvm), and structural model comparison.
- Understanding a drmTMB fit with symbolizer
Read a distributional (location-scale) regression: the mean submodel, the variance submodel, and what each coefficient means.
- Reading factors, dummies, and interactions
How factors, dummy coding, contrasts, and interactions appear in the equation – and the pitfalls that trip up interpretation.
- A tour of non-Gaussian families
Poisson, Beta, Gamma, lognormal and more – each family’s distribution, link function, and coefficient reading.
- Where the variation lives: ICC and repeatability
Partition variance into components, then read ICC and repeatability – and which scale each number lives on.
- Latent variables in ecology: a gllvmTMB worked example
Generalized linear latent-variable (GLLVM) models for multivariate ecology: factors, loadings, and the implied trait covariance.
- Comparing two symbolized models
See what changes between two model specifications – terms, assumptions, and structure – side by side.
Cross-package bridges
Models whose structure lives in another package: phylogenetic, animal-model, and spatial dependence, meta-analysis across metafor / brms / MCMCglmm, and multi-node structural equation models (piecewiseSEM / drmSEM).
- Structural dependence: phylogenetic, animal-model, and spatial random effects
Phylogenetic, animal-model, and spatial random effects as one grammar – the same dependence model across metafor, MCMCglmm, brms, and phylolm.
- Three flavors of meta-analysis: a cross-package tour
The same meta-analytic model across metafor, glmmTMB, and drmTMB – known sampling variances and the heterogeneity they leave behind.
- Multi-node structural equation models: piecewiseSEM and drmSEM
Where we’re going
The capability roadmap: what is Stable, what is a First slice, and what is planned or reserved.
- Roadmap and capability matrix
Where symbolizer is today, where it’s going, and what every Stable / First slice / Planned status word actually means.