migpdCoefs function

Change values of parameters in a migpd object

Change the values of parameters in a migpd object. You might want to do this after modelling marginal distributions as functions of covariates.

migpdCoefs(object, which, coefs)

Arguments

• object: An object of class migpd.

• which: Which models in the migpd object you want to change.

• coefs: The coefficients that you want to change to. If which

  has length 1, coefs can be a vector of parameters. Otherwise, it should be a list of vectors, and the list should have the same length as which

Returns

A migpd object. See the help for migpd.

Examples

library(MASS)
liver <- liver
liver$ndose <- as.numeric(liver$dose)
d <- data.frame(alt = resid(rlm(log(ALT.M) ~ log(ALT.B) + ndose, data=liver)),
                ast = resid(rlm(log(AST.M) ~ log(AST.B) + ndose, data=liver)),
                alp = resid(rlm(log(ALP.M) ~ log(ALP.B) + ndose, data=liver)),
                tbl = resid(rlm(log(TBL.M) ~ log(TBL.B) + ndose, data=liver)))

Dgpds <- migpd(d[liver$dose == "D", 1:4], mqu=.7)

d$ndose <- liver$ndose
galt <- evm("alt", data=d, qu=.7, xi = ~ ndose)
gast <- evm("ast", data=d, qu=.7, xi = ~ ndose)
galp <- evm("alp", data=d, qu=.7, xi = ~ ndose)

altco <- predict(galt,type="lp",newdata=data.frame(ndose=4))$obj$link[1:2]
astco <- predict(gast,type="lp",newdata=data.frame(ndose=4))$obj$link[1:2]
alpco <- predict(galp,type="lp",newdata=data.frame(ndose=4))$obj$link[1:2]

Dgpd <- migpdCoefs(Dgpds, which=c("alt", "ast", "alp"),
                   coefs=list(altco, astco, alpco))

summary(Dgpd)
summary(Dgpds)

See Also

migpd

Author(s)

Harry Southworth