mdsic: Information criterion for Bayesian multidimensional scaling...

In dmbc: Model Based Clustering of Binary Dissimilarity Measurements

Information criterion for Bayesian multidimensional scaling (BMDS).

Description

mdsic computes the information criterion for a set of Bayesian multidimensional scaling (BMDS) solutions using the approach in Oh & Raftery (2001).

Usage

mdsic(x_star, rmin_ssr, n, min_p = 1, max_p = 6)

Arguments

x_star	An array containing the latent configurations estimated using bmds.
rmin_ssr	A numeric vector providing the ratios of SSR for the latent dimensions requested.
n	A length-one numeric vector providing the number of objects.
min_p	A length-one numeric vector providing the minimum value of the latent space dimension to use.
max_p	A length-one numeric vector providing the maximum value of the latent space dimension to use.

Value

A list with the following elements:

mdsic

A numeric vector with the values of MDSIC index.

bic

A numeric vector with the values of the BIC index.

Author(s)

Sergio Venturini [e-mail]

References

Oh, M.-S., Raftery, A. E. (2001), "Bayesian Multidimensional Scaling and Choice of Dimension", Journal of the American Statistical Association, 96, 1031-1044.

See Also

bmds for Bayesian (metric) multidimensional scaling and comp_ssr for the computation of SSR.

Examples

## Not run: 
# Road distances (in km) between 21 cities in Europe
data(eurodist, package = "datasets")

min_p <- 1
max_p <- 10
burnin <- 200
nsim <- 1000
totiter <- burnin + nsim

eurodist.mds <- cmdscale(eurodist, max_p)
eurodist.bmds <- bmds(eurodist, min_p, max_p, burnin, nsim)

plot((min_p:max_p), eurodist.bmds$mdsIC$mdsic, type = "b",
  main = "MDS Information Criterion", xlab = "p", ylab = "MDSIC")
MDSICmin <- which.min(eurodist.bmds$mdsIC$mdsic)
points((min_p:max_p)[MDSICmin], eurodist.bmds$mdsIC$mdsic[MDSICmin],
  col = "red", pch = 10, cex = 1.75, lwd = 1.5)

## End(Not run)

dmbc documentation built on April 26, 2022, 5:05 p.m.