The purpose of "display" is to display, as PostScript(R) files, planar
decision trees and data.  This function is only available for 2-dimensional
data sets (including artificial data sets), but it is very useful as an
aid in visualizing and understanding the tree-building process.  It is
also a very useful teaching tool.

The following is a description of the various command line options 
for "display".

  -D  : File containing the decision tree to be displayed.
        For sample format, see sample.dt.
        This option overrides the -d and -c options, as the decision tree also
        lists the number of dimensions and the number of categories. "display"
        may not even allow you to specify this option along with -d and -c
        options on the command line.
        Default: None.
  -t or -T : File containing the data points
        For sample format, see linear.data.
        If a file name is specified with this option, the program 
        automatically computes the number of examples, the number of 
        attributes, and the number of categories.  If the number of
        attributes (dimensions) is specified with
        the -d option, and/or given in the decision tree file, it will
        override the value computed from the datafile. 
        Default: None.
  -o :  file to write the PostScript(R) output.
        Default=stdout (the screen)
        You can either redirect the output to a file or specify a filename
        using this option.
  -d :  number of dimensions (or attributes) 
        If this number is not equal to 2, an error results, since display
        currently works only in 2-D.  It is not necessary to specify this
        parameter.
  -p :  Pruning Threshold 
        Default = 0.000.
        Can be used to display only part of a decision tree.
        A node in the decision tree is pruned if
          1. number of examples at the node/total number of examples
             <= pruning threshold OR
          2. (initial error at the node - 
                error after placing the hyperplane)/ initial error 
              <= pruning threshold.
        The number of examples etc are computed from the examples file
        specified in the -t option.
  -v :  Verbose output. Default = FALSE
  -h :  Character string header (title) for the display.
        Default="<datafile name>" or
                "<decision tree file name>" or
                "<datafile name>-<decision tree file name>", depending on
                which of the -t and -D options are specified.
  -x :  Minimum x coordinate for the display. Default = 72. (These are 
        PostScript(R) screen coordinates). 
  -X :  Maximum x coordinate for the display. Default = 540.  
  -y :  Minimum y coordinate for the display. Default = 72.  
  -Y :  Maximum y coordinate for the display. Default = 640.  

If a datafile is specified with the -t or -T option, "display" uses
the numbers in the first column as the x coordinates, those in the
second column as the y coordinates, and the those in the third column
as the class numbers/labels. If a line has more than 3 columns, an
error message may be produced.

"display" first finds the minimum and maximum x and y coordinates
based on the dataset. (If no dataset is specified, these correspond to
the extreme coordinates of the display, which in turn can be set using
the -x, -X, -y and -Y options). These extreme coordinates, along with
the display boundary, define scaling factors along the x and y
dimensions.

The file formats used by "display" (especially the decision trees) are
currently not as flexible as we would like them to be. Efforts are
underway to be able to read more "general" decision tree formats.