OpenRepGrid (version 0.1.9)

bertinCluster: Bertin display with corresponding cluster anaylsis.

Description

Element columns and constructs rows are ordered according to cluster criterion. Various distance measures as well as cluster methods are supported.

Usage

bertinCluster(x, dmethod = c("euclidean", "euclidean"), cmethod = c("ward", "ward"), p = c(2, 2), align = TRUE, trim = NA, type = c("triangle"), xsegs = c(0, 0.2, 0.7, 0.9, 1), ysegs = c(0, 0.1, 0.7, 1), x.off = 0.01, y.off = 0.01, cex.axis = 0.6, col.axis = grey(0.4), draw.axis = TRUE, ...)

Arguments

x

repgrid object.

dmethod

The distance measure to be used. This must be one of "euclidean", "maximum", "manhattan", "canberra", "binary", or "minkowski". Default is "euclidean". Any unambiguous substring can be given (e.g. "euc" for "euclidean"). A vector of length two can be passed if a different distance measure for constructs and elements is wanted (e.g.c("euclidean", "manhattan")). This will apply euclidean distance to the constructs and manhattan distance to the elements. For additional information on the different types see ?dist.

cmethod

The agglomeration method to be used. This should be (an unambiguous abbreviation of) one of "ward", "single", "complete", "average", "mcquitty", "median" or "centroid". Default is "ward". A vector of length two can be passed if a different cluster method for constructs and elements is wanted (e.g.c("ward", "euclidean")). This will apply ward clustering to the constructs and single linkage clustering to the elements. If only one of either constructs or elements is to be clustered the value NA can be supplied. E.g. to cluster elements only use c(NA, "ward").

p

The power of the Minkowski distance, in case "minkowski" is used as argument for dmethod. p can be a vector of length two if different powers are wanted for constructs and elements respectively (e.g. c(2,1)).

align

Whether the constructs should be aligned before clustering (default is TRUE). If not, the grid matrix is clustered as is. See Details section in function cluster for more information.

trim

The number of characters a construct is trimmed to (default is 10). If NA no trimming is done. Trimming simply saves space when displaying the output.

type

Type of dendrogram. Either or "triangle" (default) or "rectangle" form.

xsegs

Numeric vector of normal device coordinates (ndc i.e. 0 to 1) to mark the widths of the regions for the left labels, for the bertin display, for the right labels and for the vertical dendrogram (i.e. for the constructs).

ysegs

Numeric vector of normal device coordinates (ndc i.e. 0 to 1) to mark the heights of the regions for the horizontal dendrogram (i.e. for the elements), for the bertin display and for the element names.

x.off

Horizontal offset between construct labels and construct dendrogram and (default is 0.01 in normal device coordinates).

y.off

Vertical offset between bertin display and element dendrogram and (default is 0.01 in normal device coordinates).

cex.axis

cex for axis labels, default is .6.

col.axis

Color for axis and axis labels, default is grey(.4).

draw.axis

Whether to draw axis showing the distance metric for the dendrograms (default is TRUE).

...

additional parameters to be passed to function bertin.

Value

A list of two hclust object, for elements and constructs respectively.

See Also

cluster

Examples

## Not run: 
# 
#    # default is euclidean distance and ward clustering
#    bertinCluster(bell2010)
# 
#    ### applying different distance measures and cluster methods
# 
#    # euclidean distance and single linkage clustering
#    bertinCluster(bell2010, cmethod="single")
#    # manhattan distance and single linkage clustering
#    bertinCluster(bell2010, dmethod="manhattan", cm="single")
#    # minkowksi distance with power of 2 = euclidean distance
#    bertinCluster(bell2010, dm="mink", p=2)
# 
#    ### using different methods for constructs and elements
# 
#    # ward clustering for constructs, single linkage for elements
#    bertinCluster(bell2010, cmethod=c("ward", "single"))
#    # euclidean distance measure for constructs, manhatten
#    # distance for elements
#    bertinCluster(bell2010, dmethod=c("euclidean", "man"))
#    # minkowski metric with different powers for constructs and elements
#    bertinCluster(bell2010, dmethod="mink", p=c(2,1)))
# 
#    ### clustering either constructs or elements only
#    # euclidean distance and ward clustering for constructs no
#    # clustering for elements
#    bertinCluster(bell2010, cmethod=c("ward", NA))
#    # euclidean distance and single linkage clustering for elements
#    # no clustering for constructs
#    bertinCluster(bell2010, cm=c(NA, "single"))
# 
#    ### changing the appearance
#    # different dendrogram type
#    bertinCluster(bell2010, type="rectangle")
#    # no axis drawn for dendrogram
#    bertinCluster(bell2010, draw.axis=F)
# 
#    ### passing on arguments to bertin function via ...
#     # grey cell borders in bertin display
#    bertinCluster(bell2010, border="grey")
#    # omit printing of grid scores, i.e. colors only
#    bertinCluster(bell2010, showvalues=FALSE)
# 
#    ### changing the layout
#    # making the vertical dendrogram bigger
#    bertinCluster(bell2010, xsegs=c(0, .2, .5, .7, 1))
#    # making the horizontal dendrogram bigger
#    bertinCluster(bell2010, ysegs=c(0, .3, .8, 1))
# ## End(Not run)