Package: inlaops 0.1.2

inlaops: Deployment, Versioning and Scoring Tools for 'inlabru' Models

Serves Bayesian spatial and latent Gaussian models fit with 'inlabru' as versioned Representational State Transfer (REST) application programming interfaces (APIs). Includes a 'vetiver' bridge (versioning, REST API serving with posterior uncertainty, 'Docker' packaging) for 'bru' objects, a 'parsnip' model constructor scoped to fixed-effects models, and predictive scoring helpers (continuous ranked probability, log, squared-error and Dawid-Sebastiani scores) computed from inlabru::generate() posterior samples via 'scoringRules'.

Authors:Novica Nakov [aut, cre, cph]

inlaops_0.1.2.tar.gz
inlaops_0.1.2.zip(r-4.7-any)inlaops_0.1.2.zip(r-4.6-any)inlaops_0.1.2.zip(r-4.5-any)
inlaops_0.1.2.tgz(r-4.6-any)inlaops_0.1.2.tgz(r-4.5-any)
inlaops_0.1.2.tar.gz(r-4.7-any)inlaops_0.1.2.tar.gz(r-4.6-any)
inlaops_0.1.2.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
inlaops/json (API)

# Install 'inlaops' in R:
install.packages('inlaops', repos = c('https://novica.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/novica/inlaops/issues

Pkgdown/docs site:https://novica.github.io

On CRAN:

Conda:

inlainlabrumachine-learningmlopsmodelingquarto

5.13 score 20 downloads 6 exports 33 dependencies

Last updated from:778d4d18d4. Checks:8 OK, 1 ERROR. Indexed: yes.

TargetResultTimeFilesSyslog
linux-develOK178
source / vignettesOK283
linux-releaseOK175
macos-releaseOK153
macos-oldrelERROR205
windows-develOK113
windows-releaseOK150
windows-oldrelOK163
wasm-releaseOK154

Exports:bru_cv_predictbru_inlaops_fitbru_pred_formulabru_predict_conf_intbru_regbru_score

Dependencies:classclassIntcliDBIdplyre1071fmeshergenericsglueinlabruKernSmoothlatticelifecyclemagrittrMatrixMatrixModelspillarpkgconfigproxyR6Rcpprlangs2sfspsplancstibbletidyselectunitsutf8vctrswithrwk

Serving an inlabru model on Posit Connect
1. Install inlaops from GitHub | 2. Fit and pin the model | 3. Write plumber.R | 4. Publish | 5. Call the API | INLA on the server
Rendered fromconnect.qmdusingquarto::html

Last update: 2026-10-20
Started: 2026-10-15

Packaging an inlabru model API in a container
Why not vetiver_prepare_docker()? | 1. Write plumber.R | 2. Write the Dockerfile | 3. Build and test locally | 4. Spatial models | 5. Image size
Rendered fromdocker.qmdusingquarto::html

Last update: 2026-10-20
Started: 2026-10-14

How inlaops maps to the vetiver workflow
Step by step | In short
Rendered fromworkflow.qmdusingquarto::html

Last update: 2026-10-15
Started: 2026-10-15

Versioning and serving inlabru models with vetiver
Prerequisites | 1. Fit a model | 2. Wrap with vetiver | Aside — parsnip-fitted models | 3. Pin to a versioned board | 4. Serve the API and call it | 5. Track model quality across versions | Pins board options | Multi-likelihood models
Rendered fromvetiver.qmdusingquarto::html

Last update: 2026-10-15
Started: 2026-10-14

Spatial models: scope and limitations
What works | vetiver bridge — supported with spatial metadata | LGCP intensity surfaces | Out-of-scope for the bridge (rejected at pin time) | Areal lookup — planned | broom tidiers — supported for the fixed part | What does NOT work | bru_reg() parsnip constructor | bru_cv_predict() cross-validation helper | Recommended workflow for spatial models | Request body contract | Docker image for spatial fits | Multi-likelihood models
Rendered fromspatial.qmdusingquarto::html

Last update: 2026-10-15
Started: 2026-10-14

Model versioning and MLOps workflows
Setup | 1. Baseline model — version 1 | 2. New data, new version | 3. Compare versions before promoting | 4. Promote the better version | 5. Roll back | 6. Longitudinal quality tracking | 7. Serving the production model | 8. Real production boards | Posit Connect | S3 (or any object store) | Auth in serving code | A note on board hygiene | Versioning checklist
Rendered fromversioning.qmdusingquarto::html

Last update: 2026-10-15
Started: 2026-10-14

inlabru as a Bayesian GLM with parsnip
Scope | Setup | Example: Poisson regression for count data | Fit with bru_reg() | Inspect with broom | Predict | Comparing against a frequentist baseline | Cross-validation with bru_cv_predict() | Connecting to vetiver
Rendered fromglm.qmdusingquarto::html

Last update: 2026-10-14
Started: 2026-10-14

Proper scoring rules for Bayesian model comparison
Setup | The four scores | Example: comparing two models | Visualising score distributions | Relationship to WAIC and DIC | Scoring spatial and non-Gaussian models | Notes on n_samples
Rendered fromscoring.qmdusingquarto::html

Last update: 2026-10-14
Started: 2026-10-14

Serving LGCP intensity surfaces
Setup | 1. Fit the LGCP | 2. Serve via the vetiver bridge | 3. Fallback — custom plumber API | What you keep, what you give up
Rendered fromlgcp.qmdusingquarto::html

Last update: 2026-10-14
Started: 2026-10-14