smacof (version 1.8-13)

inverseMDS: Inverse MDS

Description

Performs inverse MDS. That is, given a configuration matrix, it computes a dissimilarity matrix.

Usage

inverseMDS(x)

Arguments

x

Configuration matrix

Value

• Returns a list with dissimilarity matrices.

Details

Note that this is a very basic implementation. It can be that some dissimilarities are negative. Also, the weights are fixed to a value of 1.

References

De Leeuw J., Groenen P.J.F. (1997). Inverse multidimensional scaling. Journal of Classification, 12, 3-21.

De Leeuw, J. (2012). On inverse multidimensional scaling. Statistics preprint series, UCLA. gifi.stat.ucla.edu/janspubs/2012/ notes/deleeuw_U_12b.pdf.

See Also

mds

Examples

## MDS and inverse MDS on kinship data subset
D <- as.matrix(kinshipdelta)[1:6, 1:6]
fit <- mds(D)                       ## MDS D --> conf
ifit <- inverseMDS(fit$conf)        ## inverse MDS conf --> D

## MDS fit on dissimilarity matrix obtained by inverse MDS
op <- par(mfrow = c(3,3))           
plot(fit, main = "Original MDS")
for (i in 1:length(ifit)) {
  fit.i <- mds(ifit[[i]])             
  plot(fit.i, main = paste0("Inverse MDS (",i, ")"))
}
par(op)