****************************************************************
* Copyright 1993 : Johns Hopkins University			*
*                  Department of Computer Science		*
****************************************************************
* Contacts : murthy@cs.jhu.edu (Sreerama K. Murthy)             *
*            salzberg@cs.jhu.edu (Steven Salzberg)              *
*            kasif@cs.jhu.edu (Simon Kasif)                     *
****************************************************************

Welcome to OC1 !

OC1 is a program to construct oblique decision trees from examples.
Oblique decision trees are trees in which each node may contain a
(linear) multivariate test on the attributes of the data.  OC1 also
constructs standard axis-parallel trees, which contain tests of just
one attribute at each node.  Oblique decision trees are a natural
generalization of axis-parallel trees.

As no one decision tree building method (or, for that matter, machine
learning method) is the best for all datasets, we feel that a machine
learning researcher should experiment with as many methods as possible
when attempting to solve a problem.  To aid in this goal, we are
making OC1 available in the public domain.  Please use it, experiment
with it, and let us know your questions, comments or suggestions.  OC1
is intended for non-commercial use only, and you should feel free to
use, copy, and modify it for such purposes.  Any commercial use of OC1
is strictly prohibited without the express written consent of the
authors.

The OC1 directory has four main components: 
  "gendata" generates artificial datasets, given (optionally) a 
            decision tree; 
  "mktree"  builds decision trees out of data, estimates classification 
            accuracies; 
  "display" displays 2D datasets and/or decision trees as PostScript(R) files. 
  "aaai93-paper.ps" contains the PostScript(R) version of our paper:

     S.K. Murthy, S. Kasif, S. Salzberg, and R.~Beigel.
     "OC1: Randomized Induction of Oblique Decision Trees."
     Proceedings of the Eleventh National Conference on Artificial
     Intelligence (pp. 322--327), Washington, D.C., 1993.

If you use the OC1 software in the context of any of your
publications, please reference the above paper.

To install OC1, after "FTP"ing, uncompressing and unarchiving ("tar"ing) 
the software, run the following commands:
      $ make mktree
      $ make gendata
      $ make display 

These commands will create the executable files for the three main
commands available in the OC1 system.  You can get help on the usage
of these commands (mktree, gendata and display), by
typing the command, with no arguments, at the $ prompt.  First,
though, we recommend that you look at the text file "sample_session",
which contains a session with OC1 illustrating many of the options
available.  More detailed description of these options can be found in
the text files mktree.readme, gendata.readme, and display.readme.

The complete source code of OC1 is available with this package, with 
what we hope is an adequate documentation. 

This directory also contains sample files (linear.data, sample.dt) giving
the formats of a data file and a decision tree file, respectively.

Finally, a note: using multivariate tests at each node of a
decision tree has both advantages and disadvantages. The resulting
trees may be smaller and/or more accurate, but they may be more
time-consuming to induce than univariate trees.

Enjoy !

-Sreerama K. Murthy
 Steven Salzberg
 Simon Kasif

 Department of Computer Science
 Johns Hopkins University
 Baltimore, MD 21218
 U.S.A.