Skip to contents

Diagnose and fit

The two main functions. Both take an ordinary fitted inla object.

ng.check()
Check the latent Gaussian assumption of an INLA fit.
ngvb()
Fit a latent non-Gaussian model from a fitted INLA (LGM) object.

Model comparison

Integrate the mixing variables out and compare against the Gaussian model.

ngvb_sample()
Sample the mixing variables and refit INLA at each draw.
bayes.factor()
Bayes factor of a latent non-Gaussian model against its Gaussian counterpart.

Operators and custom models

The dependency-matrix building blocks, and how to add your own.

ngvb_operator()
Construct an ngvb operator descriptor.
ngvb_custom()
Define a custom latent non-Gaussian model from a precision "square root".
ngvb_rgeneric()
Build an INLA rgeneric model for an ngvb operator at a fixed V.
ngvb_precision()
Conditional precision matrix Q(theta, V) for an operator (main-process helper).

Plotting

Diagnostic plots and maps for areal and geostatistical data.

plot(<ngvb>)
Diagnostic plots for an ngvb fit.
plot(<ngvb.check>)
Diagnostic plots for a latent-Gaussianity check.
areal.plot()
Plot areal (regional) data on a map.
geo.plot()
Plot geostatistical (point-referenced) data, optionally over a mesh.

Methods

S3 methods for the fitted objects.

summary(<ngvb>)
Summary of an ngvb fit: the underlying INLA summaries plus, per component, the non-Gaussianity parameter and the most strongly down-weighted (outlying) indices.
density(<ngvb>)
Posterior density of a component's non-Gaussianity parameter eta.
fitted(<ngvb>)
Fitted values of the underlying INLA fit.
summary(<ngvb.samples>)
Importance-weighted posterior summary of the sampled fits.

Data

jumpts
Simulated time series with two jumps
Orthodont
Orthodontic growth data