checkMultRoot: Inspect estimated parameters

Description Usage Arguments Details Value Author(s) Examples

View source: R/qleTest.R

Description

Check out and compare estimated roots of the quasi-score vector

Usage

1
2
checkMultRoot(est, par = NULL, opts = list(ftol_stop = 1e-07, score_tol =
  0.001), verbose = FALSE)

Arguments

est

object of class qle, the estimation results from function qle

par

list or matrix of estimated parameters as roots of the quasi-score vector

opts

list of upper bounds for a root of the quasi-score vector, see details

verbose

logical, TRUE for intermediate output

Details

Only for the quasi-likelihood approach the function inspects the (numerical) consistency of the found parameters in 'par' by comparing each observed quasi-information matrix with the expected one. The degree of dissimilarity of both matrices is measured by certain scalar equivalent criteria (see vignette) and the parameter for which these are smallest is chosen. The numerical upper bounds to determine a root of the quasi-score are as follows: 'ftol_stop' for the quasi-deviance criterion value and 'score_tol' for the maximum of any of the components of the quasi-score vector.

Value

A data frame with columns named corresponding to each component of the investigated parameter, 'quasi-deviance', 'minor', 'det', 'max', 'trace' and 'score_max' (see vignette). The second column shows the leading minor of the observed QI matrix which is not positive definite. If so, then the corresponding parameter estimate cannot be consistent at least in theory. The rows show the corresponding values for each parameter passed by 'par'.

Author(s)

M. Baaske

Examples

1
2
3
4
5
6
 data(normal)
 # a dummy estimation result
 OPT <- qle(qsd,qsd$simfn,global.opts=list("maxeval"=0))
 
 # and just a single root 
 checkMultRoot(OPT,verbose=TRUE)

mbaaske/qle documentation built on Feb. 3, 2018, 11:02 a.m.