CLUSPAC
     Computer Programs for Mixture-Model Cluster Analysis
 
                 Copyright (C) 1991  Stanley L. Sclove

 
Developed and programmed by
 
          Prof. Stanley L. Sclove, Ph.D.                    312/996-2681
          Information & Decision Sciences Dept.   m/c 294   312/996-2676
          College of Business Administration
          University of Illinois at Chicago
          601 S. Morgan St.
          Chicago, IL  60607-7124
 
 
     CLUSPAC  is  a  package  of  FORTRAN  programs  for clustering and
classification of (univariate and) multivariate data.
 
     CLUSPAC is part of the ISOPAC library, which also  contains  TSPAC
for  time-series  segmentation  and IMPAC for segmentation of numerical
images.
 
     There are three sets of programs in CLUSPAC, denoted by  CLASS***,
ISDT****, and MIX*****.
 
     CLASS*** series:  classification (i.e., assignment or allocation)
                       of given observations (data)
                       into k multivariate normal distributions with
                       specified parameters and prior probabilities
 
     ISDT**** series:  clustering by ISODATA (Ball and Hall 1967) as
                       modified by Sclove (1977)
 
     MIX***** series:  clustering based on a mixture of normal
                       distributions (Wolfe 1970)
 
 
     The   programs   have  restrictions,  which  may  be  modified  by
relatively minor changes in some of the FORTRAN statements:
 
     N, sample size, at most 999;
     K, number of clusters, at most 29;
     ITER, maximum number of iterations, 20.
 
 
     The  programs  require  various  control  statements,  such as the
following:
 
     (1) dataset title
     (2) N, in format (2X,I4)
     (3) FMT, in format (18A4), e.g.,  (1X,F4.1).
         allow at least one blank in FMT:  it will also be used for
         output, where cc1 is for carriage control.  Allow a cc for the
         decimal point on output, whether or not there is one on input.
     (4) data, in format specified by FMT
 
CLUSPAC documentation:
 
CRIM Working Paper 91-8
CRIM Working Paper 92-1
CLUSPAC User Manual.  Available on-line on the ISOPAC public disk.
 
 
                              References
 
Ball, G.H., and Hall, D.J. (1967).  A clustering technique for
   summarizing multivariate data. Behavioral Science 12, 153-155. The
   basic reference for ISODATA.
 
Johari, Shyam, and Sclove, Stanley L.  Partitioning a distribution.
   Communications in Statistics (A) 5 (1976), 133-147. Gives optimal
   class probabilities for the normal (and other) distributions.
 
Sclove, Stanley L.  Population  mixture  models  and  clustering
   algorithms.  Communications in Statistics(A) 6 (1977), 417-434.
   Gives a probability interpretation for and certain modifications
   to ISODATA.
 
Wolfe, J. H. (1970).  Pattern clustering by multivariate mixture
   analysis. Multivariate Behavioral Research 5, 329-350.  A basic
   reference on normal mixture-model clustering.
 
 
18-Jan-92