
%***************************************************************************
%                              AutoSplit
%   This set of source codes is (c) Copyright 2002 by Jia-Yu Pan
%   (Computer Science Department, Carnegie Mellon University).
%   It may not be redistributed without the express consent of the 
%   author.
%***************************************************************************

Please run 'autosplit' in MATLAB for a demo.
The demo reads in a synthetic data set with points distributed in a 2-dimensional "X-shaped" data set.
Its takes approximately <1 minute (on an Intel Xeon CPU 2.80GHz) to complete on the X-shaped data.

To run on your data, replace the file 'synthX.txt' with your file of 2 columns giving the (X,Y)-coordinates of your data points.

When completed, see MATLAB window "figure 3" for the estimated basis vectors, plotted on top of the input data points.

Several EPS files will be generated in the subdirectory 'result/'.
Including:
   synthX_A_final.eps:
      A plot showing the data set with the final autosplit bases.
   synthX_A.mat:
      The final computed data file, in MATLAB format.
   synthX_likelihood.eps:
      The log-likelihood versus iteration plot.
   synthX_timeVSiteration.eps:
      The per-iteration time versus iteration.

Please cite the following paper for autosplit, thank you:

Jia-Yu Pan, Hiroyuki Kitagawa, Christos Faloutsos, and Masafumi Hamamoto. AutoSplit: Fast and Scalable Discovery of Hidden Variables in Stream and Multimedia Databases. In Proceedings of The Eighth Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) 2004.
