Scotch {bayesm} R Documentation

## Survey Data on Brands of Scotch Consumed

### Description

Data from Simmons Survey. Brands used in last year for those respondents who report consuming scotch.

### Usage

data(Scotch)

### Format

A data frame with 2218 observations on 21 brand variables.
All variables are numeric vectors that are coded 1 if consumed in last year, 0 if not.

### Source

Edwards, Yancy and Greg Allenby (2003), "Multivariate Analysis of Multiple Response Data," Journal of Marketing Research 40, 321–334.

### References

Chapter 4, Bayesian Statistics and Marketing by Rossi, Allenby, and McCulloch.

### Examples

data(Scotch)
cat(" Frequencies of Brands", fill=TRUE)
mat = apply(as.matrix(Scotch), 2, mean)
print(mat)

## use Scotch data to run Multivariate Probit Model
if(0) {
y = as.matrix(Scotch)
p = ncol(y)
n = nrow(y)
dimnames(y) = NULL
y = as.vector(t(y))
y = as.integer(y)
I_p = diag(p)
X = rep(I_p,n)
X = matrix(X, nrow=p)
X = t(X)

R = 2000
Data = list(p=p, X=X, y=y)
Mcmc = list(R=R)

set.seed(66)
out = rmvpGibbs(Data=Data, Mcmc=Mcmc)

ind = (0:(p-1))*p + (1:p)
mat = apply(out$betadraw/sqrt(out$sigmadraw[,ind]), 2 , quantile,
attributes(mat)$class = "bayesm.mat" summary(mat) rdraw = matrix(double((R)*p*p), ncol=p*p) rdraw = t(apply(out$sigmadraw, 1, nmat))