smacof: SMACOF

In mkampert/rCOSA: Clustering Objects on Subsets of Attributes (COSA)

Description

A metric multidimensional scaling algorithm that minimizes a least squares loss function based on the dissimilarities Heiser \& De Leeuw, 80 and 77; Gutman 68). The lettters in SMACOF stand for Scaling by MAjorizing a COmplicated Function.

Usage

smacof(
  D,
  niter = 100,
  interc = 1,
  inicon = NULL,
  groupnr = NULL,
  colv = palette()[c(8, 2, 4, 3, 5, 6, 7, 1)],
  main = "Multidimensional Scaling",
  k = 2,
  pch = 16,
  PLOT = TRUE,
  VERBOSE = TRUE,
  ...
)

Arguments

D	dissimilarities/distances of class dist.
niter	max number of iterations to use, default is set to 20: 'to iterate is heaven, to converge is divine'. Convergence does not necessarily give you the best non-linear mapping of the dissimilarity structure.
interc	logical for including an additive constant in the smacof algorithm.
inicon	intial configuration. By default this is specified as NULL, here the SMACOF algorithm specifies...
groupnr	groupnr when known, each object can be given a number to which group it belongs
colv	vector containing color names for each group number
main	title of the plot, by default there is no title.
k	number of dimensions
pch	plotting 'character', i.e. symbol to use. This can either be a single character or an integer code for one of a set of graphics symbols. The full set of S symbols is available with pch = 0:18. For more information see the help file of the function points.
PLOT	whether to plot or not to plot, by default TRUE.
VERBOSE	whether to give a stdout on the criterion for each iteration, by default set to equal TRUE.
...	arguments which can be parsed to either the hclust or cmdscale procedures.

Value

Similar output as cmdscale

mkampert/rCOSA documentation built on Dec. 23, 2019, 8:21 p.m.