Using the Clustering Tool

The software is packaged as a Java jar file, which means that is can be run from the commandline with the following command

java -jar clustertool.jar

This will bring up the application in normal interactive mode. In this case the data should be loaded with the file->open menu and then the desired parameters set up within the GUI before running a fit with the tools->clustering menu.

Command Line Mode

The application can be run to completion without using the interactive GUI, but rather specifying all of the model parameters on the commandline as shown below.

java -jar clustertool.jar --nogui [options] data_file_name  
                                --nogui do not run the gui, but process a file directly as specified by other arguments 
                                 -t <d>  set the fitting tolerance 
                                 -o      include outliers 
                                 -m      use minimum message length methods to determine number of clusters
                                 -c <i>  number of continuous variables
                                 -e <i>  number of continuous variables with errors  
                                 -b <i>  number of binary variables
                                 -d <i>  number of modal variables
                                 -i <i>  number of integer variables
                                 -n <i>  number of iterations to use
                                 -x <i>  number of clusters
                                 -v common|diagonal|free type of covariance to use

The results will be written out to a VOTable format file "results.vot". The results consist of the original data with the columns appended which contain the probability that each data point is a member of the particular cluster.