Manual page for apps.Clust program

Clust program finds groups of similar objects (cluters) by means of hierarchical clustering algorithm

-clust.in.distances=<file name> input file with a distance matrix. The matrix defines a distance between any two clustered objects.
-clust.in.n_data=<number> the number of clustered objects
-clust.in.tree=<file name> input file with a clustering tree. Such a tree can be computed computed in the first round of clustering and stored in a file. Then it may be used to re-select clusters by different criteriawithout repeating the clustering step.
-clust.out.clusters=<file name> output file for a resulting clusters
-clust.out.distances=<file name> output file for a resulting matrix of distances. The printed result is just the input distance matrix where columns and rows are reordered according to the order of clustering.
-clust.out.tree=<file name> save a resulting clustering tree into a file
-cluster.average enforces average-linkage clustering algorithm.
-cluster.complete enforces complete-linkage clustering algorithm.
-cluster.max_distance=<value> clustering stops when the merging distance at a given clustering step reaches a value provided by this option. Note, that in some clustering methods (e.g. average-linkage) merging distance is not always monotonic.
-cluster.max_ratio=<value> clustering stops when the given fraction of clutering steps were done. Fraction 1.0 means that there is a single cluster containing all the objects. Fraction 0.5 corresponds to N/2 clusters.
-cluster.min_size=<value> the program will print only these clusters that gather at least this number of elements
-cluster.missing_value=<value> value to be used as a distance between two points when an input file doesn’t define it
-cluster.single enforces single-linkage clustering algorithm.
-h print a brief summary of available options
-help=<name-part> print a help message on the screen - ANSI terminal version with visual enhancements. If <name-part> argumen is given, the program will print only these options that contains that substring
-help.dox=<name> print a help message in doxygen (*.dox) format on the screen for the Clust program
-help.md=<name> print a help message in markdown (*.md) format on the screen
-help.option=<option-name> print a help message for a single option on the screen.
-help.plain=<name-part> print a help message on the screen - plain text version. If <name-part> argumen is given, the program will print only these options that contains that substring
-mute suppress all messages from a given package or class, e.g. “-mute=jbcl.data.formats”, or “-mute=jbcl.calc.structural.Crmsd”. It is also possible to switch of a whole branch from the jbcl library, e.g. “-mute=jbcl.data” will mute all comming from jbcl.data.formats,jbcl.data.types, jbcl.data.dict and jbcl.data.basic. To switch all the messages, say: “-mute=jbcl” or simply “-mute” because the default behaviour is to mute everything. This option is executed AFTER -verbose, so user may increase verbosity level to a desired valueand then selectively switch off logging from some packages
-verbose=<integer> Sets up a verbosity level to a given value. The argument should be an integer from the rangefrom 0 (no messages at all, which is equivalent to -mute=jbcl) to 6 when everything is logged. See -mute for additional information.

EXAMPLES

      (1) Reads an input file "data" with distances between 51 objects and
calculates a hierarchical clustering according to a complete-link
strategy
    java  apps.Clust -clust.in.distances=data  -n=51 -max_ratio=0.75 -complete