sim: Simulation from order restricted linear models

In goric: Generalized Order-Restricted Information Criterion

Description

Simulation function for orlm and orgls objects

Usage

sim(object, n.sims)

## S3 method for class 'orlm'
sim(object, n.sims)

## S3 method for class 'orgls'
sim(object, n.sims)

Arguments

object	an object of class "orlm" or "orgls".
n.sims	number of simulation replications.

Details

Given the estimated coefficients of a orlm or orgls model, a set new parameters are generated. n.sims new sets of observations are generated based on the unrestricted model; these new datasets are used to estimate a new set of model coefficients incorporating the given order restrictions.

Value

a list with sets of simulated parameters.

See Also

orlm, orgls

Examples

########################
## Artificial example ##
########################
n <- 10
m <- c(1,1,2)
dat <- data.frame(grp=as.factor(rep(1:length(m), each=n)),
                  y=rnorm(n*length(m), rep(m, each=n), 1))
cm <- rbind(c(-1,1,0),
            c(0,-1,1))
fm <- orlm(y ~ grp-1, data=dat, constr=cm, rhs=rep(0,nrow(cm)), nec=0)
b <- sim(fm, n.sims=1000)$coef
pairs(t(b), cex=0.3)

goric documentation built on April 19, 2021, 5:06 p.m.