Infusion (version 1.0.0)

infer_surface: Infer a (summary) likelihood or tail probability surface from inferred likelihoods

Description

The logLs method uses a standard smoothing method (prediction under linear mixed models, a.k.a. Kriging) to infer a likelihood surface, using as input likelihood values themselves inferred with some error for different parameter values. The tailp method use a similar approach for smoothing binomial response data, using the algorithms implemented in the spaMM package for fitting GLMMs with autocorrelated random effects.

Usage

"infer_surface"(object, method="REML",verbose=interactive(),allFix=NULL,...)
"infer_surface"(object, method="PQL",verbose=interactive(),allFix,...)

Arguments

object

A data frame with attributes, containing independent prediction of logL or of LR tail probabilities for different parameter points, as produced by infer_logLs or infer_tailp.

method

methods used to estimate the smoothing parameters. If method="GCV", a generalized cross-validation procedure is used (for logLs method only). Other methods are as described in the HLfit documentation.

verbose

Whether to display some information about progress or not.

allFix

Fixed values in the estimation of smoothing parameters. For development purposes, not for routine use. For infer_surface.logLs, this should typically include values of all parameters fitted by spaMM::corrHLfit ($\rho,\nu,\phi,\lambda$, and $etaFix=$\beta$).

...

further arguments passed to or from other methods (currently not used).

Value

An object of class SLik or SLikp, which is a list including an HLfit object as returned by corrHLfit, and additional members not documented here.

Examples

## see main documentation page for the package