LinkageMap: Displays a Linkage Map

In VisuClust: Visualisation of Clusters in Multivariate Data

Description

A Linkage Map is a 2D Scatter Plot that displays the result of Sammon's Nonlinear Mapping. A number of ranges (>= 1) for the distance values d[i,j] can be defined. Each point pair whose distance is inside of the defined range is connected with a line.

The ranges are defined by thresholds t[k](1<=k<=3) in the following way range[1]: d[i,j] <= t[1] range[k]: t[k-1] < d[i,j] <= t[k] The ranges can be changed dynamically at runtime by changing the thresholds t[k] with sliders.

A estimation of the density function for the distances d[i,j] (for example with kernel density estimation, see function density) is helpful for finding the ideal thresholds t[k] (=local minima in the density function).

Usage

LinkageMap(xSammon, dist, lineTypes=c("solid","dotted", "dashed"), 
	lineColors=c("red","green","blue"), lineWidths=c(1,1,1), 
	labels = NULL, cluster = NULL, maxValue=0.5, legendDigits = 2, 
	xlab = "", ylab = "", main = "")

Arguments

xSammon	A matrix with size (nPoints,2) that contains the projected points from Sammon's Nonlinear Mapping.
dist	A distance matrix for the given data (to compute with dist).
lineTypes	An array of line types. The size must match with line colors and line widths. The various line types are described in par.
lineColors	An array of colors. The size must match with line types and line widths. see colors
lineWidths	An array of line widths. The size must match with line colors and line types. see par
labels	A string-array with labels for the points (size: nPoints).
cluster	A array containing cluster memberships of the points. The point membership will be indicated with different colors. The array size must match with nPoints.
maxValue	maxValue*max(d[i,j]) is the maximum value that can be adjusted with the sliders.
legendDigits	The number of fractional digits to be displayed in the legend.
xlab	A title for the x axis (as described in plot).
ylab	A title for the y axis (as described in plot).
main	An overall title for the plot (as described in plot).

Author(s)

Michael Sieger <[e-mail]>
Georg Ohmayer <[e-mail]>

Examples

library(MASS)

data("MilchSmall")
M <- scale(MilchSmall[3:6])

D <- dist(M)
S <- sammon(D)
K <- kmeans(D,center=5)
LinkageMap(S$points, D, cluster=K$cluster, labels=MilchSmall[,2])

VisuClust documentation built on May 2, 2019, 6:53 a.m.