challenge {mcsm} | R Documentation |
This function illustrates a slice sampling implementation of the simulation from the posterior distribution associated with a logistic regression model
P(y=1|x)=exp(a+bx)/(1+exp(a+bx))
when applied to the challenger
dataset.
challenge(Nsim = 10^4)
Nsim |
Number of slice sampling iterations |
The output is a list
made of
a |
Sequence of values of the intercept a produced by the slice sampler |
b |
Sequence of values of the regression coefficient b produced by the slice sampler |
The function challenge
uses a function rtrun
that is replicated from
a function used in the package bayesm
. In the current case, the simulation of
the truncated normal distribution is done by a simple cdf inversion and may thus be
fragile in the tails.
Christian P. Robert and George Casella
Chapter 6 of EnteR Monte Carlo Statistical Methods
data(challenger) chares=challenge(10^4) plot(chares$a,chares$b,type="l",xlab="a",ylab="b",pch=19,cex=.4)