Scope first | Start with the mathematical contract | Define the registry entry | Do not assume identity links | Connect the family to the builder | Implement the likelihood | Add simulation and method support | Write tests before calling the family implemented | Update user and developer documentation | Close with an after-task audit
Package: drmTMB 0.7.0
Shinichi Nakagawa
drmTMB: Distributional Regression Models Using Template Model Builder
Fast distributional regression models for univariate and bivariate responses using Template Model Builder. The current implementation focuses on Gaussian, Student-t, and skew-normal location-scale models, known sampling covariance, phylogenetic location effects, random-effect scale models, bivariate residual correlation, positive-continuous, 'Tweedie' semi-continuous, strict-proportion, zero-one bounded, and denominator-aware proportion families, fixed-effect Bernoulli/binomial event-probability models, and fixed-effect Poisson, negative-binomial, zero-inflated, zero-truncated, hurdle count, and ordinal cumulative-logit models. Additional response-family models are staged for later phases. Every fitted family also exposes a distributional-output and adequacy layer: randomized quantile-residual worm and QQ plots that detect fixed-effect shape and atom misspecification, and conditional-quantile, exceedance, and centile outputs with plug-in (uncalibrated) intervals.
Authors:
drmTMB_0.7.0.tar.gz
drmTMB_0.7.0.zip(r-4.7-x86_64)drmTMB_0.7.0.zip(r-4.7-arm64)drmTMB_0.7.0.zip(r-4.6-x86_64)drmTMB_0.7.0.zip(r-4.6-arm64)
drmTMB_0.7.0.tar.gz(r-4.7-arm64)drmTMB_0.7.0.tar.gz(r-4.7-x86_64)drmTMB_0.7.0.tar.gz(r-4.6-arm64)drmTMB_0.7.0.tar.gz(r-4.6-x86_64)
drmTMB_0.7.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
drmTMB/json (API)
| # Install 'drmTMB' in R: |
| install.packages('drmTMB', repos = c('https://cran.r-universe.dev', 'https://cloud.r-project.org')) |
Bug tracker:https://github.com/itchyshin/drmtmb/issues
Pkgdown/docs site:https://itchyshin.github.io
Last updated from:4996a6f768. Checks:10 OK. Indexed: yes.
| Target | Result | Time | Files | Syslog |
|---|---|---|---|---|
| linux-devel-arm64 | OK | 402 | ||
| linux-devel-x86_64 | OK | 441 | ||
| source / vignettes | OK | 584 | ||
| linux-release-arm64 | OK | 469 | ||
| linux-release-x86_64 | OK | 430 | ||
| windows-devel-arm64 | OK | 467 | ||
| windows-devel-x86_64 | OK | 567 | ||
| windows-release-arm64 | OK | 464 | ||
| windows-release-x86_64 | OK | 488 | ||
| wasm-release | OK | 206 |
Exports:animalassociate_pairsassociationbetabeta_binomialbfbiv_associatebiv_gaussianbiv_lognormalbiv_studentcategoricalcentile_chartcheck_drmcorpaircorpairscumulative_logitdrm_controldrm_formuladrm_phylo_penaltydrm_phylo_penalty_sweepdrmTMBexceedancefitted_distributionfixefgrimpute_modelimputedis_convergedlatent_normallognormalmake_meshmarginal_parametersmeta_known_Vmeta_Vmeta_vcov_bivariatemimiss_controlnbinom2phylophylo_interactionplot_corpairsplot_parameter_surfacepredict_parametersprediction_gridprofile_targetsqq_plotranefrelmatrho_latentrho12skew_normalspatialspatial_coordsstructured_effectsstudenttruncated_nbinom2tweedieworm_plotzero_one_beta
Adding distribution families
Last update: 2026-09-15
Started: 2026-09-15
Animal models and additive relatedness
What is fitted today | Start with the smallest useful model | What to inspect | Rendered checks
Last update: 2026-09-15
Started: 2026-09-15
Association between mixed outcome pairs
The question in an ecological example | One call, two stages | What a binary outcome contributes | Reviewed pair classes | Standard error and confidence interval | When the estimate or interval is withheld | Can the association itself be modelled? | How this differs from rho12
Last update: 2026-09-15
Started: 2026-09-15
Bivariate non-Gaussian models: choose a joint or staged association
Start by choosing the association scale | Direct joint model: two positive traits | Real example: penguin size after measured composition differences | Direct joint model: two heavy-tailed traits | Staged association: different kinds of outcome | What is currently outside these routes
Last update: 2026-09-15
Started: 2026-09-15
Can I fit and report this model?
Before reporting | Evidence and exact tested scopes | Capability tiers, defined once | Historical capability skim (technical evidence) | Detailed route evidence | Tier 1: inference-ready | Gaussian structured random effects: selected anchor cells | Ordinary (unstructured) random effects: point-trustworthy, coverage pending | Unstructured non-Gaussian fixed effects | Ordinary non-Gaussian random slopes: profile intervals with caveats | Gamma sigma random intercept: narrow coverage-backed interval | Tier 2: recovery-only | Tier 3: diagnostic-only | Tier 4: not available or rejected by design | Known limitations for 0.7.0 | Frozen-margin association intervals | Missing data | Full per-family capability map | Choosing an interval method | Where the exhaustive detail lives
Last update: 2026-09-15
Started: 2026-09-15
Changing residual coupling with rho12
Equations for the implemented model | Worked example: behaviour coupling and disturbance | What should I report? | Reading group-level covariance | First slope-slope covariance slice | Boundaries for this covariance layer
Last update: 2026-09-15
Started: 2026-09-15
Choosing response families
At a glance | Why sigma is not always phi | Reporting variation | Implemented univariate families | Bounded and ordered families | Bivariate Gaussian families
Last update: 2026-09-15
Started: 2026-09-15
Comparing drmTMB with other packages
What this article compares | Checked against a separate estimation engine | Comparison 1: contagious bovine pleuropneumonia across periods and herds | Comparison 2: the pooled BCG vaccine effect | Comparison 3: does heterogeneity itself differ by allocation design? | Checked against ordinal | Comparison 4: does serving temperature shift wine bitterness ratings? | Comparison 5: do judges differ, and does that change the temperature effect? | Checked against a package built on the same machinery | Comparison 6: does reaction-time variability grow with sleep deprivation? | Comparison 7: are male penguins more variable in body mass than females? | Comparison 8: are satiated owl broods more variable, not just quieter? | What these eight comparisons show | Reproducing these numbers
Last update: 2026-09-15
Started: 2026-09-15
Coordinate-spatial structured effects
What is fitted today | Start with the smallest useful model | Geographic coordinates and the fixed-kappa mesh route | What to inspect | Rendered checks | Boundaries
Last update: 2026-09-15
Started: 2026-09-15
Count abundance and extra zeros
Model Equation And Syntax | A Soil-Invertebrate Example | When Zeros Are A Separate Process | Current Boundary
Last update: 2026-09-15
Started: 2026-09-15
Distributional regression with drmTMB
Install the pre-CRAN release | Fit your first model | Learning path | Check before interpreting
Last update: 2026-09-15
Started: 2026-09-15
Errors, warnings, and convergence
Start with check_drm() | Use drm_control() for optimizer limits | If the optimizer did not converge | If pdHess is false | When bivariate phylogenetic correlations are near a boundary | Penalized / MAP estimation for weakly-identified phylogenetic components | Choosing between maximum likelihood and REML | Separate optimization from uncertainty | Model-specific advice | What is planned but not automatic yet | Further reading
Last update: 2026-09-15
Started: 2026-09-15
First-week intervals: fit, profile, and boundary
Default recipe (read once) | Fit a small random-intercept model | Inventory targets, then profile the RE-SD | Read profile.boundary before you trust the interval | What to read next
Last update: 2026-09-15
Started: 2026-09-15
Formula grammar
Current status map | Offsets
Last update: 2026-09-15
Started: 2026-09-15
Handling missing data
Capability matrix | What is fitted now | Choosing the predictor family | Relation to gllvmTMB and glmmTMB | Missing responses | Partial bivariate response pairs | Missing predictors | Binary missing predictors | Non-Gaussian responses with binary missing predictors | Ordered missing predictors | Unordered missing predictors | Strict proportion missing predictors | Boundary proportion missing predictors | Denominator-aware proportion missing predictors | Count missing predictors | Positive continuous missing predictors | Semi-continuous missing predictors with exact zeros | Grouped and structured predictor models | Current boundaries
Last update: 2026-09-15
Started: 2026-09-15
Implementation map
How to read the project maps | Family and component map | Current capability and evidence map | Random-effect and dependence map | Slope coverage at a glance | Current future-work lanes | Common planned requests
Last update: 2026-09-15
Started: 2026-09-15
Implemented source map
Routing overview | Implemented paths | What the source map protects | C++ modularization boundary | Validation-debt register | Current boundaries
Last update: 2026-09-15
Started: 2026-09-15
Julia engine: future support
What to use now | What future Julia support would need to establish | Scope boundary
Last update: 2026-09-15
Started: 2026-09-15
Known-matrix relatedness with relmat
When is relmat() the right tool? | Choosing K or Q | Two concrete examples | What is fitted today | Start with the smallest useful model | What to inspect | Rendered checks
Last update: 2026-09-15
Started: 2026-09-15
Model selection with AIC and BIC
Gaussian REML After Mean-Model Selection | Tail Assumption | Structural Zeros | Scale Formula | A 200-Replicate Article-Support Simulation | Practical Checklist
Last update: 2026-09-15
Started: 2026-09-15
Phylogenetic mixed models
The phylogenetic covariance | A small tree with ape::rcoal() | Simulating a Gaussian trait with phylogenetic signal | Fitting with phylo(1 | species) | Reading the phylogenetic signal | Uncertainty for the two SDs | Non-Gaussian responses | A larger fit, for reference | What phylo() fits today | References
Last update: 2026-09-15
Started: 2026-09-15
Proportions and success rates
Choose The Proportion Route | Plain Binomial Equation And Syntax | Beta-Binomial Equation And Syntax | Seed Germination Example | Strict Continuous Proportions | Continuous Proportions With Exact Boundaries | Current Boundary
Last update: 2026-09-15
Started: 2026-09-15
Random-effects meta-analysis with known sampling variances
The model | A simulated dataset | Fitting the model | The pooled effect | Between-study heterogeneity | Cross-check against metafor | REML for the heterogeneity | Meta-regression: moderators on the mean | Multiple effect sizes per study | Layered heterogeneity: an experimental contract | Known sampling variance is not a weight | Correlated effect sizes: a small dense covariance example | Notes on the function names | Reference
Last update: 2026-09-15
Started: 2026-09-15
Robust continuous responses
Model equation and R syntax | A seedling growth example | Check the fitted robust model | Interpret coefficients by parameter | Compare with the Gaussian model | Current boundary
Last update: 2026-09-15
Started: 2026-09-15
Structural dependence overview
Pick the route | What to read next | Fitted versus planned
Last update: 2026-09-15
Started: 2026-09-15
Testing likelihoods
The contract | Tier 1: comparator checks | Tier 2: simulation recovery | Independent likelihood checks | Bivariate and coscale checks | Boundary and rejection tests | Fast and long tests | Before closing a task
Last update: 2026-09-15
Started: 2026-09-15
Two-tree phylogenetic interactions
What is fitted today | A small count example | Independent pairs are a different question | When to reach for relmat() | What this does not claim
Last update: 2026-09-15
Started: 2026-09-15
What can I fit today?
Stable-core matrix | One response | Two responses | Structural dependence | Planned, not implemented | Correlation layers | A practical trait protocol | Where to go next
Last update: 2026-09-15
Started: 2026-09-15
When variance carries signal, Part 1: location-scale models
Model equations and matching R syntax | Grouped effects do not change the meaning of sigma | Interpret sigma on the SD or variance scale | Worked example: growth mean and predictability | Curved responses and interactions | Continue to Part 2 | Current implementation boundaries
Last update: 2026-09-15
Started: 2026-09-15
When variance carries signal, Part 2: location-scale-scale models
Personality, predictability, and repeatability | Simulate and fit the three submodels | Read the fitted scales and calculate repeatability | What this example supports | Reference
Last update: 2026-09-15
Started: 2026-09-15
Which scale are you modelling?
Residual standard deviation | Mean-model random intercept | Residual-scale random intercept | Random-effect scale formula | One location model, several scale quantities | A copy-run scale audit | Residual scale: sigma ~ temperature | Known sampling variance: meta_V(V = V) | Random-effect scale: sd(population) ~ habitat | Side-by-side guide | Correlations: group-level versus residual | Family A versus Family B | Current boundary
Last update: 2026-09-15
Started: 2026-09-15
Working with large data
What is implemented now | Large tip counts and se_group_sd | What this does not solve yet | Choose the interval route deliberately | Check the fit with the storage choice in mind | Avoid accidental full-row output | Benchmark before the full run | What the current development benchmarks say | Practical checklist
Last update: 2026-09-15
Started: 2026-09-15
