resp.check: Diagnostic plot for a variable

In GJRM: Generalised Joint Regression Modelling

Diagnostic plot for a variable

Description

It produces a normal Q-Q plot for the (randomised) normalised quantile response. It also provides the log-likelihood for AIC calculation, for instance. It is also used for internal purposes.

Usage

resp.check(y, margin = "N", print.par = FALSE, plots = TRUE, 
           loglik = FALSE, os = FALSE, i.f = FALSE, 
           min.dn = 1e-40, min.pr = 1e-16, max.pr = 0.999999,
           left.trunc = 0)

Arguments

y	Response.
margin	The distributions allowed are: normal ("N"), log-normal ("LN"), generelised Pareto ("GP"), discrete generelised Pareto ("DGP"), Gumbel ("GU"), reverse Gumbel ("rGU"), logistic ("LO"), Weibull ("WEI"), inverse Gaussian ("iG"), gamma ("GA"), Dagum ("DAGUM"), Singh-Maddala ("SM"), beta ("BE"), Fisk ("FISK"), Poisson ("P"), zero truncated Poisson ("ZTP"), negative binomial - type I ("NBI"), negative binomial - type II ("NBII"), Poisson inverse Gaussian ("PIG").
print.par	If TRUE then the estimated parameters used to construct the plot are returned.
plots	If FALSE then no plot is produced and only parameter estimates returned.
loglik	If TRUE then it returns the logLik.
os	If TRUE then the estimated parameters are returned on the original scale.
i.f	Internal fitting option. This is not for user purposes.
min.dn, min.pr, max.pr	Allowed minimum and maximum for estimated probabities and densities for parameter estimation.
left.trunc	Value of truncation at left. Currently done for count distributions only.

Details

Prior to fitting a model with discrete and/or continuous margins, the distributions for the outcome variables may be chosen by checking the normalised quantile responses. These will provide a rough guide to the adequacy of the chosen distribution. The latter are defined as the quantile standard normal function of the cumulative distribution function of the response with scale and location estimated by MLE. These should behave approximately as normally distributed variables (even though the original observations are not). Therefore, a normal Q-Q plot is appropriate here.

If loglik = TRUE then this function also provides the log-likelihood for AIC calculation, for instance.

Author(s)

Maintainer: Giampiero Marra [e-mail]

See Also

gjrm

GJRM documentation built on Oct. 25, 2024, 5:07 p.m.