plot.tclust {tclust}

Plot Method for tclust and tkmeans Objects

Description

One and two dimensional structures are treated separately (e.g. tolerance intervals/ellipses are displayed). Higher dimensional structures are displayed by plotting the two first Fisher's canonical coordinates (evaluated by tclust::discr_coords) and derived from the final cluster assignments (trimmed observations are not taken into account). plot.tclust.Nd can be called with one or two-dimensional tclust- or tkmeans-objects too. The function fails, if store.x = FALSE is specified in the tclust() or tkmeans() call, because the original data matrix is required here.

Usage

## S3 method for class 'tclust'
plot(x, ...)

## S3 method for class 'tkmeans'
plot(x, ...)

Arguments

x

The tclust or tkmeans object to be displayed

...

Further (optional) arguments which specify the details of the resulting plot (see section "Further Arguments").

Details

The plot method for classes tclust and tkmeans.

Further Arguments

• xlab, ylab, xlim, ylim, pch, col Arguments passed to plot().

• main The title of the plot. Use "/p" for displaying the chosen parameters alpha and k or "/r" for plotting the chosen restriction.

• main.pre An optional string which is added to the plot's caption.

• sub A string specifying the subtitle of the plot. Use "/p" (default) for displaying the chosen parameters alpha and k, "/r" for plotting the chosen restriction and "/pr" for both.

• sub1 A secondary (optional) subtitle.

• labels A string specifying the type of labels to be drawn. Either labels="none" (default), labels="cluster" or labels="observation" can be specified. If specified, parameter pch is ignored.

• text A vector of length n (the number of observations) containing strings which are used as labels for each observation. If specified, the parameters labels and pch are ignored.

• by.cluster Logical value indicating whether parameters pch and col refer to observations (FALSE) or clusters (TRUE).

• jitter.y Logical value, specifying whether the drawn values shall be jittered in y-direction for better visibility of structures in 1 dimensional data.

• tol The tolerance interval. 95% tolerance ellipsoids (assuming normality) are plotted by default.

• tol.col, tol.lty, tol.lwd Vectors of length k or 1 containing the col, lty and lwd arguments for the tolerance ellipses/lines.

Examples

 #--- EXAMPLE 1------------------------------
 sig <- diag (2)
 cen <- rep (1, 2)
 x <- rbind(MASS::mvrnorm(360, cen * 0,   sig),
 	       MASS::mvrnorm(540, cen * 5,   sig * 6 - 2),
 	       MASS::mvrnorm(100, cen * 2.5, sig * 50))
 # Two groups and 10\% trimming level
 a <- tclust(x, k = 2, alpha = 0.1, restr.fact = 12)
 plot (a)
 plot (a, labels = "observation")
 plot (a, labels = "cluster")
 plot (a, by.cluster = TRUE)
 #--- EXAMPLE 2------------------------------
 sig <- diag (2)
 cen <- rep (1, 2)
 x <- rbind(MASS::mvrnorm(360, cen * 0,   sig),
 	       MASS::mvrnorm(540, cen * 5,   sig),
 	       MASS::mvrnorm(100, cen * 2.5, sig))
 # Two groups and 10\% trimming level
 a <- tkmeans(x, k = 2, alpha = 0.1)
 plot (a)
 plot (a, labels = "observation")
 plot (a, labels = "cluster")
 plot (a, by.cluster = TRUE)