LABSORT

Sort label array of a multiclustering

    [LABOUT,IND] = LABSORT(LABIN,TYPE,MAXSIZE)
    [LABOUT,IND] = LABIN*LABSORT(TYPE,MAXSIZE)

Input
LABIN	Numeric label array, size MxL, M objects, L levels.
TYPE	'up' sort columns according to increasing number of clusters.  'down' sort columns according to decreasing number of clusters.  'upred' sort columns according to increasing number of  clusters and remove columns with equal numbers.  'downred' sort columns according to increasing number of  clusters and remove columns with equal numbers.  Default: 'up
MAXSIZE	Clustering with more than MAXSIZE clusters will be removed.

Output
LABOUT	MxN array with the results of the multilevel clusterings of the
M	objects. The sorted columns refer to N clusterings in an order  as defined by TYPE.
IND	Sorting index used, LABOUT = LABIN(:,IND)

Description

This is a support routine used by several procedures that analyse the  output of a multiclustering. The rows of LABIN are ranked according to  the numbers of clusters. In addition rows pointing to clusterings with  the same number of clusters are reduced.

See also

datasets, mappings, dclusth, cluste, clustf, clustk, clusth, clustkh, clustm, reclustk, reclustn, reclusth, clusteval, clustcerr, clustc, clustnum, plotdg,